1. Cell Biology
  2. Immunology and Inflammation
Download icon

Autophagy regulates inflammatory programmed cell death via turnover of RHIM-domain proteins

Research Article
  • Cited 0
  • Views 626
  • Annotations
Cite this article as: eLife 2019;8:e44452 doi: 10.7554/eLife.44452

Abstract

RIPK1, RIPK3, ZBP1 and TRIF, the four mammalian proteins harboring RIP homotypic interaction motif (RHIM) domains, are key components of inflammatory signaling and programmed cell death. RHIM-domain protein activation is mediated by their oligomerization; however, mechanisms that promote a return to homeostasis remain unknown. Here we show that autophagy is critical for the turnover of all RHIM-domain proteins. Macrophages lacking the autophagy gene Atg16l1accumulated highly insoluble forms of RIPK1, RIPK3, TRIF and ZBP1. Defective autophagy enhanced necroptosis by Tumor necrosis factor (TNF) and Toll-like receptor (TLR) ligands. TNF-mediated necroptosis was mediated by RIPK1 kinase activity, whereas TLR3- or TLR4-mediated death was dependent on TRIF and RIPK3. Unexpectedly, combined deletion of Atg16l1 and Zbp1 accelerated LPS-mediated necroptosis and sepsis in mice. Thus, ZBP1 drives necroptosis in the absence of the RIPK1-RHIM, but suppresses this process when multiple RHIM-domain containing proteins accumulate. These findings identify autophagy as a central regulator of innate inflammation governed by RHIM-domain proteins.

https://doi.org/10.7554/eLife.44452.001

Introduction

Programmed cell death plays a central role in dictating tolerogenic or immuno-stimulatory responses. To leverage these pathways therapeutically, it is critical to understand how immune-suppressive versus inflammatory modes of cell death (e.g. necroptosis and pyroptosis) are regulated. RHIM-domain containing proteins have emerged as central nodes in inflammatory signaling mediated by cytokines or microbial antigens (also known as microbe-associated molecular patterns/MAMPs) (de Almagro and Vucic, 2015; Kajava et al., 2014; Pasparakis and Vandenabeele, 2015). Receptor interacting protein kinase 1 (RIPK1) kinase activity drives caspase 8-dependent apoptosis as well as pro-inflammatory necroptosis dependent on RIPK3 and its substrate mixed lineage kinase domain like (MLKL) (Cho et al., 2009; He et al., 2009; Sun et al., 2012; Zhang et al., 2009; Zhao et al., 2012). Additionally, the cytosolic adaptor Toll/IL-1 receptor (TIR) domain-containing adaptor protein inducing interferon−β (TRIF) and innate sensor Z-DNA binding protein 1 (ZBP1) can directly interact with RIPK1 and/or RIPK3 via their RHIM domains, thereby stabilizing downstream signaling (He et al., 2011; Kaiser et al., 2013; Lin et al., 2016; Newton et al., 2016; Thapa et al., 2016). While the role of these proteins in inflammation and cell death have been elucidated via genetic deletion, mechanisms which drive a return to homeostasis have remained poorly described (Cuchet-Lourenço et al., 2018; Ito et al., 2016; Mocarski et al., 2014; Ofengeim et al., 2017; Weinlich et al., 2017). RHIM-dependent oligomerization of TRIF, RIPK1, RIPK3 and ZBP1 is required for their function, but this also results in the generation of amyloid-like structures which require regulated turnover to prevent signal amplification (Kaiser et al., 2013; Li et al., 2012; Rebsamen et al., 2009).

Selective autophagy targets the autophagic machinery to specific cargo via autophagy receptors. Forms of selective autophagy have been identified in the turnover of organelles, cytosolic pathogens and protein aggregates, ultimately driving their turnover via lysosomal degradation (Anding and Baehrecke, 2017; Khaminets et al., 2016; Levine et al., 2011; Mizushima, 2007). Putative links between autophagy and TNF-mediated necroptosis of epithelial cells have emerged, but contradictory observations in primary versus transformed cells question how autophagy impacts necroptosis in these models (Goodall et al., 2016; Matsuzawa-Ishimoto et al., 2017). In the current study, we investigated the role of autophagy in innate immunity driven by macrophage activation. We found that autophagy is critical for the turnover of highly insoluble complexes containing TRIF, RIPK1, RIPK3 or ZBP1. Defective autophagy enhanced cytokine production and necroptosis driven by activators of RHIM-domain proteins. Unexpectedly, we observed that ZBP1 dampens necroptosis in a context-specific manner, since deletion of Zbp1 in an autophagy deficient background exacerbated necroptosis driven by TRIF. Thus, we identify autophagy as an upstream regulator of RHIM-domain proteins and reveal a non-canonical, immunosuppressive function of ZBP1 upon defective autophagy.

Results

Defective autophagy enhances RIPK1-dependent and independent forms of macrophage death

We first asked whether TNF or TLR ligands promote cell death in autophagy-deficient macrophages by deleting Atg16l1, a core gene in the autophagy pathway. Stimulation with TNF or TLR ligands does not significantly induce death of wild-type macrophages, with the exception of TLR3 which can drive caspase-8 mediated apoptosis via TRIF (Kaiser et al., 2013; Gentle et al., 2017; Kawasaki and Kawai, 2014). However, combining inflammatory stimuli with caspase-inhibitors are established methods to study inflammatory cell death via necroptosis in vitro and in vivo (de Almagro and Vucic, 2015; Pasparakis and Vandenabeele, 2015; Weinlich et al., 2017). Atg16l1 deficient bone marrow-derived macrophages (Atg16l1-cKO BMDMs) did not exhibit increased sensitivity to TNF alone, but they were more sensitive than wild-type (Atg16l1-WT) BMDMs to necroptotic stimulus of TNF plus pancaspase inhibitor zVAD-fmk (Figure 1A). Blocking the kinase activity of RIPK1 with Necrostatin-1 (Nec-1) reduced death in both Atg16l1-WT and Atg16l1-cKO BMDMs (Figure 1A), consistent with active RIPK1 engaging RIPK3 downstream of TNFR1 (Newton et al., 2014). Death induced by TLR2, TLR3, TLR4, TLR7/8 or TLR9 ligands plus zVAD-fmk was also enhanced by Atg16l1 deficiency (Figure 1B; Figure 1—figure supplement 1A). While Nec-1 suppressed necroptosis by TLR2, TLR7/8 and TLR9 ligands in both genotypes, it was less effective at preventing TLR3- or TLR4-mediated death in Atg16l1-cKO BMDMs compared with Atg16l1-WT BMDMs (Figure 1B; Figure 1—figure supplement 1A). IL-1β release by Atg16l1-cKO BMDMs was also elevated upon LPS-mediated necroptosis independent of RIPK1 inhibition (Figure 1—figure supplement 1B). The modest effect of Nec-1 in Atg16l1-cKO BMDMs may stem from it blocking necroptosis due to autocrine TNF production, which was elevated upon Atg16l1 deletion (Figure 1—figure supplement 1C), but not death due to other mechanisms activating RIPK3.

Figure 1 with 2 supplements see all
Defective autophagy enhances RIPK1-dependent and independent necroptosis.

(A, B) Cell death assayed by Propidium Iodide (PI) staining and live-cell imaging for 12–16 hr (n = 5). BMDMs from mice of the indicated genotypes were treated with combinations of TNF/zVAD/Nec-1 (A) or PolyI:C/zVAD/Nec-1 and LPS/zVAD/Nec-1 (B). (C) Immunoblots confirming deletion of autophagy genes in BMDMs of indicated genotypes using RNP electroporation. NTC = non targeting control gRNA. (D, E) Cell death assayed under combinations of PolyI:C/zVAD/Nec-1 (D) or LPS/zVAD/Nec-1 (E) treatment (n = 4). Data in (A, B) are representative of four independent experiments; (C–E) are representative of two independent experiments. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Bar graphs depict mean.

https://doi.org/10.7554/eLife.44452.002

Core autophagy genes can contribute to autophagy-independent functions in innate immunity (Codogno et al., 2011; Fletcher et al., 2018; Heckmann et al., 2017). Conclusive proof of autophagy in suppressing necroptosis therefore requires a genetic approach assessing multiple autophagy-related genes. We established a non-viral gene editing protocol in primary murine macrophages by comparing CRISPR/Cas9-mediated deletion of enhanced green fluorescent protein (eGFP; schematic in Figure 1—figure supplement 2A). Efficient gene knockdown was achieved in monocyte- and bone marrow-derived macrophages (Figure 1—figure supplement 2B, eGFP deletion; Figure 1—figure supplement 2C,D, Ptprc/CD45 deletion). We used this method in wild-type (WT) BMDMs to delete canonical autophagy genes Atg14l, Rb1cc1 (encoding FIP200) and Atg16l1, as well as Rubcn (encoding Rubicon), the principal gene associated with LC3-associated phagocytosis (LAP) (Heckmann et al., 2017) (Figure 1C). Deletion of Atg14l and Rb1cc1 significantly enhanced PolyI:C/zVAD- or LPS/zVAD-induced necroptosis, whereas Rubcn deletion resulted in cell death levels comparable to Atg16l1-WT controls (Figure 1D,E). As with loss of Atg16l1, Atg14l and Rb1cc1-deficient BMDMs maintained elevated levels of cell death upon Nec-1 treatment (Figure 1D,E). These results demonstrate that canonical autophagy suppresses TLR3/4-induced necroptosis in BMDMs and uncover a RIPK1 kinase-independent mode of cell death by TLR3/4 activation when autophagy is perturbed.

TRIF and RIPK1 drive necroptosis in Atg16l1-deficient macrophages

To characterize mode of cell death in Atg16l1-cKO BMDMs, we performed CRISPR-mediated deletion of necroptosis mediators Ripk3 and Mlkl as well as Gsdmd (Gasdermin d), the final executor of pyroptosis (Kayagaki et al., 2015; Sarhan et al., 2018; Shi et al., 2015). Deletion of Ripk3 and Mlkl but not Gsdmd prevented LPS/zVAD- and TNF/zVAD-induced death in both Atg16l1-WT and Atg16l1-cKO BMDMs, confirming that macrophage death was due to necroptosis (Figure 2A,B; Figure 2—figure supplement 1A). Recently, Gsdmd- and caspase-1-independent secondary pyroptosis mediated by Nlrp3 and Pycard (encoding ASC) was described in murine bone-marrow derived dendritic cells (BMDCs) (Schneider et al., 2017). However, neither Nlrp3 nor Pycard knockdown prevented RIPK1 kinase-independent cell death in Atg16l1-cKO BMDMs (Figure 2—figure supplement 1B,C). Therefore, secondary pyroptosis does not appear to contribute to the death of Atg16l1-cKO BMDMs treated with LPS/zVAD or PolyI:C/zVAD. As expected, knockdown of Ticam1 (encoding TRIF) in Atg16l1-WT or Atg16l1-cKO BMDMs also decreased necroptosis induced by PolyI:C/zVAD or LPS/zVAD, although additional inhibition of RIPK1 provided a more complete rescue of cell death (Figure 2C–E). TLR3 only signals via TRIF (reviewed in Kawasaki and Kawai, 2014), so the added protection offered by Nec-1 to wild-type cells is consistent with incomplete knockdown of Ticam1 (Figure 2C).

Figure 2 with 2 supplements see all
RIPK3, MLKL and TRIF are required for RIPK1-independent necroptosis in Atg16l1-deficient BMDMs.

(A-E) Immunoblot (A, C) and cell death assays (B, D, E) of BMDMs from mice of indicated genotypes treated with combinations of LPS/zVAD/Nec-1 following CRISPR-mediated deletion of RIPK3, MLKL or GSDMD (A, B) (n = 4) or TRIF (C–E) (n = 6). Cell death assayed by PI staining and live-cell imaging for 12–16 hr. Data in (A, B) are representative of three independent experiments; (C, D, E) are representative of four independent experiments. **p<0.01, ***p<0.001, ****p<0.0001. Bar graphs depict mean. NTC = non targeting gRNA.

https://doi.org/10.7554/eLife.44452.008

TNF and type I interferons are proposed to license necroptosis in murine macrophages (Sarhan et al., 2019; Siegmund et al., 2016), so we tested whether pharmacological blockade of these cytokines would rescue necroptosis in Atg16l1-deficient BMDMs. Cells were pre-treated for 36 hr with a control antibody (anti-Ragweed), TNFR2-Fc to block TNF, or anti-IFNAR1 to block Interferon-α Receptor 1, and then a necroptosis stimulus was applied. TNFR2-Fc attenuated necroptosis of Atg16l1-WT BMDMs comparably to Nec-1, especially with TLR2 or TLR9 ligands or TNF itself (Figure 2—figure supplement 2A,E,F), but also with TLR3, TLR4 or TLR7 ligands (Figure 2—figure supplement 2B,C,D). In contrast, TNFR2-Fc failed to inhibit necroptosis in Atg16l1-cKO BMDMs treated with TLR2, TLR7 or TLR9 ligands (Figure 2—figure supplement 2A,D,E), despite inhibiting necroptosis induced by TLR3 or TLR4 ligands or TNF comparably to Nec-1 (TNFR2-Fc, Figure 2—figure supplement 2B, C, F). Thus, while TNF contributes to enhanced necroptosis of Atg16l1-deficient BMDMs, activation of RIPK1 contributes more to this phenotype, perhaps indicating the involvement of multiple death receptors.

Consistent with the results of McComb et al. (2014), pharmacological blockade of IFNAR1 prevented TLR2, TLR3, TLR4, or TLR9-induced necroptosis in Atg16l1-WT BMDMs (Figure 2—figure supplement 2A,B,C,E). IFNAR1 blockade also decreased TLR3- or TLR4-induced necroptosis of Atg16l1-deficient BMDMs more than Nec-1 (Figure 2—figure supplement 2B,C). However, it only provided modest protection to Atg16l1-cKO BMDMs treated with TLR2, TLR7, or TLR9 ligands or TNF (Figure 2—figure supplement 2A,D,E,F). Accordingly, phosphorylation of STAT1, which is associated with Type I interferon signaling, was increased in Atg16l1-cKO BMDMs following LPS/zVAD treatment (Figure 2—figure supplement 2G). These data indicate that signaling by Type I interferons contributes to enhanced necroptosis of Atg16l1-deficient BMDMs, particularly with ligands that activate TRIF.

Loss of autophagy results in accumulation of active forms of TRIF, RIPK1 and RIPK3 during necroptosis

We asked whether loss of Atg16l1 caused TRIF, RIPK1, or RIPK3 to accumulate in necroptotic BMDMs, as this would provide direct biochemical evidence that autophagy promotes their turnover. TRIF appeared to be monomeric in untreated BMDMs of both genotypes, running as a single band in the detergent (NP-40) soluble fraction below the 100-kDa mark in an SDS-PAGE gel (Figure 3A,B). LPS was sufficient to elicit a smear of NP40-insoluble, slower migrating TRIF species that was far more prominent in Atg16l1-cKO BMDMs than in Atg16l1-WT BMDMs (Figure 3B). CRISPR-knockdown of Ticam1 (encoding TRIF) in Atg16l1-cKO cells confirmed that the high-MW species were TRIF (Figure 3B, Atg16l1-cKO, Ticam1-gRNA sample). High-MW species of autophosphorylated RIPK1 (p-RIPK1, Ser166/Thr169) as well as total RIPK1 largely accumulated in the NP-40 insoluble fraction after treatment with LPS plus zVAD; these were more abundant in Atg16l1-cKO BMDMs than in Atg16l1-WT BMDMs (Figure 3C,D). Slower migrating species of both autophosphorylated RIPK3 (Thr231/Ser232) and total RIPK3 in the insoluble fraction after treatment with LPS plus zVAD were also elevated in Atg16l1-cKO BMDMs when compared with Atg16l1-WT BMDMs (Figure 3E,F). Knockdown of Ticam1 decreased the amount of autophosphorylated RIPK3 and autophosphorylated RIPK1 in Atg16l1-cKO cells (Figure 3D,F). Loss of Atg16l1 did not appear to affect the abundance of monomeric TRIF, RIPK1, or RIPK3 in the detergent soluble fraction (Figure 3A,C,E), suggesting that autophagic turnover specifically regulates levels of activated/modified forms of TRIF, RIPK1 and RIPK3. Consistent with reduced autophagic turnover of RHIM proteins enhancing TLR4-induced necroptosis, Atg16l1-cKO BMDMs treated with LPS plus zVAD contained more MLKL phosphorylated at Ser345, a marker of necroptosis (Sun et al., 2012; Cai et al., 2014; Chen et al., 2014; Dondelinger et al., 2014; Wang et al., 2014), when compared to their wild-type counterparts (Figure 4—figure supplement 1A).

Loss of Atg16l1 drives accumulation of detergent insoluble, high molecular weight TRIF, RIPK1, RIPK3 and enhances RIPK1/RIPK3 phosphorylation.

(A, B) Immunoblots of TRIF in Atg16l1-WT and Atg16l1-cKO BMDM lysates following 4 hr of treatment with indicated combinations of LPS/zVAD/Nec-1 and enrichment of NP-40 soluble (A) or insoluble (B) fractions. (C, D) immunoblots for autophosphorylated RIPK1 (Ser166/Thr169, p-RIPK1) and total RIPK1 in Atg16l1-WT and Atg16l1-cKO BMDM lysates following 4 hr of treatment with indicated combinations LPS/zVAD/Nec-1 and enrichment of NP-40 soluble (C) or insoluble (D) fractions. (E, F) immunoblot assay for autophosphorylated RIPK3 (Thr231/Ser232, p-RIPK3) and total RIPK3 in Atg16l1-WT and Atg16l1-cKO BMDM lysates following 4 hr of treatment with indicated combinations of LPS/zVAD/Nec-1 and enrichment of NP-40 soluble (E) or insoluble (F) fractions. Representative data shown from three independent experiments. In all immunoblots, CRISPR-mediated TRIF deletion was performed in Atg16l1-cKO BMDMs followed by LPS/zVAD treatment as a negative control. *=non specific bands (n.s.).

https://doi.org/10.7554/eLife.44452.014

Measuring the kinetics of cell death revealed that LPS/zVAD-induced necroptosis killed more than 80% of Atg16l1-cKO BMDMs within 3 hr of treatment, compared to approximately 50% of Atg16l1-WT BMDMs (Figure 4A, Figure 4—figure supplement 1B). Interestingly, Atg16l1-deficient BMDMs were more sensitive than Atg16l1-WT BMDMs to TLR3 engagement alone without zVAD (Figure 4—figure supplement 1C), consistent with TLR3 and TRIF having the capacity to assemble a death-inducing signaling complex that activates caspase-8 (Zinngrebe et al., 2016). We analyzed TRIF, RIPK1 and RIPK3 in the detergent-insoluble fraction in the first 6 hr after treatment with LPS/zVAD, and found that high-MW forms of TRIF accumulated transiently and with similar kinetics in both genotypes, peaking at 1 hr, and to a greater extent in the absence of Atg16l1 (Figure 4B). Accumulation of autophosphorylated, high-MW RIPK1 and RIPK3 peaked at 2 hr after LPS/zVAD treatment in both genotypes, with greater accumulation in Atg16l1-cKO BMDMs (Figure 4C,D). Autophosphorylated RIPK1 and RIPK3 were also ubiquitinated with Met1- or Lys63-linked chains, with greater abundance in Atg16l1-cKO BMDMs (Figure 4E). Indeed, accumulation of ubiquitinated protein aggregates is a hallmark of defective autophagy (Kwon and Ciechanover, 2017; Dikic, 2017).

Figure 4 with 2 supplements see all
Overabundance of TRIF, phosphorylated and ubiquitinated RIPK1 and RIPK3 coincides with accelerated necroptosis of Atg16l1 deficient BMDMs.

(A) Kinetic measurement of cell death over 18 hr of LPS/zVAD treatment (n = 5). (B) Immunoblot of TRIF in NP-40 insoluble fractions of BMDM lysates over 6 hr of LPS/zVAD treatment. (C, D) Immunoblots of autophosphorylated and total RIPK1 (C), RIPK3 (D) in NP-40 insoluble fractions of BMDM lysates treated as in (B). (E) Immunoblots of autophosphorylated RIPK1, RIPK3 and ubiquitin in BMDM lysates following immunoprecipitation of M1 or K63-ubiquitinated proteins after 4 hr of LPS/zVAD treatment. Data in (A) are representative of four independent experiments; (B–D) are representative of three independent experiments; (E) are representative of three independent experiments. *=P < 0.05.

https://doi.org/10.7554/eLife.44452.015

To confirm that accumulation of modified forms of TRIF, RIPK1 and RIPK3 occurred due to defective lysosomal turnover via autophagy, pharmacological inhibition of lysosomal function was performed in WT BMDMs during necroptosis. Consistent with our genetic models, treatment of WT BMDMs with Bafilomycin A1, an inhibitor lysosomal vacuolar H-ATPases, resulted in accumulation of high-MW forms of TRIF and RIPK1 in detergent insoluble fractions during LPS/zVAD-mediated necroptosis over 6 hr (Figure 4—figure supplement 2A–C). In contrast, basal turnover of low-MW TRIF, RIPK1, and RIPK3 was not perturbed in a reproducible manner by Bafilomycin A1. For comparison, inhibition of proteasomal degradation with MG-132 caused a very subtle increase in low-MW TRIF and RIPK1, whereas RIPK3 appeared unaffected Figure 4—figure supplement 2D–F). Collectively, these data indicate that lysosomal turnover via autophagy is critical for preventing the accumulation of active TRIF, RIPK1 and RIPK3, and its loss exacerbates necroptotic signaling.

The autophagy receptor TAX1BP1 prevents TRIF-mediated necroptosis

Having demonstrated a role for core autophagy genes in macrophage necroptosis (Figure 1), we analyzed autophagic flux in WT BMDMs during LPS-mediated necroptosis. LC3, a critical component of the mature autophagosome membrane that receives autophagic cargo, is lipidated during the process of autophagy. Additionally, selective autophagy receptors which can potentially associate with cytosolic substrates are trafficked to autophagosomes as a consequence of autophagic flux (Anding and Baehrecke, 2017; Dikic, 2017). Treatment with LPS/zVAD in the presence of Bafilomycin A1 to halt autophagic flux revealed accumulation of lipidated LC3B (LC3-II) in WT BMDMs (Figure 4—figure supplement 2G). Levels of the autophagy receptors SQSTM1/p62, TAX1BP1 and CALCOCO1 were elevated upon Bafilomycin A1 treatment, especially in the detergent-insoluble fraction (Figure 4—figure supplement 2G,H). These data suggest that autophagic flux was ongoing during necroptosis and that autophagy receptors accumulated in the same subcellular compartment as active TRIF, RIPK1 and RIPK3. We and others have observed that the selective autophagy receptor TAX1BP1 suppresses TRIF abundance and TRIF-dependent IFNβ production in BMDMs (Samie et al., 2018; Yang et al., 2017). These findings are consistent with our unbiased proteomics-based identification of SQSTM1/p62, TAX1BP1 and CALCOCO1 as candidate selective autophagy receptors in BMDMs during TLR4 activation (Samie et al., 2018). Immunoblotting revealed accumulation of SQSTM1/p62, TAX1BP1 and CALCOCO1 with varying kinetics after treatment with LPS/zVAD, and loss of Atg16l1 further increased the levels of these receptors in the detergent-insoluble fraction (Figure 5A,B). We utilized CRISPR-mediated knockdown of these autophagy receptors in WT BMDMs to assess their role in necroptosis induced by ligands that engage TRIF (Figure 5C). Loss of Tax1bp1 enhanced BMDM death by either PolyI:C/zVAD or LPS/zVAD treatment. Necroptosis in both settings was only partially blocked by Nec-1. In contrast, knockdown of Sqstm1 or Calcoco1 did not increase TLR3- or TLR4-induced necroptosis in BMDMs (Figure 5D). Thus, the autophagy receptor TAX1BP1 suppresses BMDM necroptosis downstream of TRIF.

The autophagy receptor TAX1BP1 protects against necroptosis by TLR3 or TLR4 ligands.

(A, B) Immunoblots of indicated autophagy receptors in total (A) or NP-40 insoluble fractions (B) of BMDM lysates over 6 hr of LPS/zVAD treatment. (C) Immunoblots confirming CRISPR-mediated deletion of indicated autophagy receptor genes in wild-type BMDMs. (D) Cell death assayed by PI staining and live-cell imaging for 12–16 hr following treatment with indicated ligands. Data in (A, B) are representative of three independent experiment; (C, D) are representative of four independent experiments. **p<0.01, ****p<0.0001. Bar graphs depict mean. NTC = non targeting control gRNA.

https://doi.org/10.7554/eLife.44452.021

Accumulation of ZBP1 protects against necroptosis in Atg16l1-deficient macrophages

We recently identified the RHIM-domain containing protein ZBP1 as one of the most highly accumulated proteins in Atg16l1-deficient macrophages (Samie et al., 2018). Loss of Atg16l1 resulted in basal accumulation of ZBP1 as shown by a cycloheximide-chase assay (Figure 6A). In contrast to TRIF, RIPK1 and RIPK3, basal turnover of ZBP1 was attenuated by lysosomal inhibition, because treatment with Bafilomycin A1, but not MG132, resulted in ZBP1 accumulation in WT BMDMs (Figure 6—figure supplement 1A). ZBP1 also accumulated during LPS/zVAD-induced necroptosis, and this was enhanced by Atg16l1 deletion (Figure 6—figure supplement 1B). No high-MW forms of ZBP1 were detected using currently available reagents. The role of the Zα1/Zα2- or RHIM-domains of ZBP1 in its accumulation will need to be addressed in future studies.

Figure 6 with 1 supplement see all
Elevated ZBP1 in Atg16l1-deficient BMDMs suppresses TRIF-mediated necroptosis.

(A) ZBP1 turnover in Atg16l1-WT and Atg16l1-cKO BMDMs following cycloheximide (CHX) treatment for indicated time points. Representative immunoblot (top), ZBP1 quantification by densitometry (bottom) normalized to ZBP1 band intensity in WT samples at 0 hr. (B, C) immunoblot (B) and cell death (C) assays of BMDMs from mice of indicated genotypes treated with combinations of LPS/zVAD/Nec-1 following CRISPR-mediated deletion of Zbp1, Ticam1 or both (n = 4). (D, E) cell death assayed in Atg16l1-WT or Atg16l1-cKO BMDMs following CRISPR-mediated Zbp1 deletion and a dose titration of LPS in the presence of 20 μM zVAD and/or 30 μM Nec-1 (n = 4). Dot-plots depict mean ±S.D. (F–H) immunoblots depicting accumulation of TRIF (F), autophosphorylated and total RIPK1 (G), autophosphorylated and total RIPK3 (H) in NP-40 insoluble lysates of BMDMs lacking both Atg16l1 and Zbp1 following induction of necroptosis via LPS/zVAD for 3 hr. Top panels represent short exposures; middle panels represent long exposures. *=non specific band. Data (A) are representative of four independent experiments, densitometry is pooled from four independent experiments. Data in (B, C) are representative of three independent experiments; (D–H) are representative of two independent experiments. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Bar graphs depict mean. NTC = non targeting control gRNA.

https://doi.org/10.7554/eLife.44452.023

ZBP1 has been shown to promote cell death upon accumulation of endogenous or viral nucleic acids (Thapa et al., 2016; Kesavardhana et al., 2017; Kuriakose et al., 2016; Maelfait et al., 2017), or genetic deletion of RIPK1-RHIM (Lin et al., 2016; Newton et al., 2016). Thus, we hypothesized that elevated ZBP1 might engage RIPK3 and contribute to enhanced necroptosis in Atg16l1-deficient BMDMs. However, CRISPR-mediated deletion of Zbp1 in Atg16l1-cKO BMDMs (Figure 6B) further enhanced LPS/zVAD-induced necroptosis, and this death was prevented by deletion of Ticam1 but not by Nec-1 (Figure 6C; Figure 6—figure supplement 1C). Therefore, contrary to expectations, ZBP1 appears to suppress TRIF-mediated necroptosis (Figure 6C). Notably, loss of Zbp1 did not impact the death of autophagy-sufficient BMDMs, suggesting that levels of ZBP1 must cross a certain threshold before suppressing TRIF-dependent necroptosis. To more thoroughly characterize the sensitization conferred by Zbp1 loss, we measured cell death after performing a dose-titration of LPS in the presence of 20 μM zVAD. CRISPR-mediated deletion of Zbp1 alone did not impact cell death at any dose of LPS (Figure 6D), but combined with defective autophagy, Zbp1 deletion sensitized cells to necroptosis at low doses of LPS, even in the presence of Nec-1 (Figure 6E). Thus, overabundant ZBP1 can antagonize TRIF-mediated necroptosis in Atg16l1-deficient macrophages. We confirmed our CRISPR-based observations by generating conditional-knockout mice lacking Atg16l1 (Atg16l1-cKO; Zbp1-WT) or Atg16l1 and Zbp1 (Atg16l1-cKO; Zbp1-cKO) in myeloid cells. Consistent with our earlier results, Zbp1 deletion accelerated LPS/zVAD-mediated necroptosis of Atg16l1-deficient macrophages (Figure 6—figure supplement 1D). ZBP1 deficiency did not affect the accumulation of TRIF in the detergent insoluble fraction of Atg16l1-cKO BMDMs treated with LPS/zVAD (Figure 6F), arguing that ZBP1 interferes with signaling events further downstream. Indeed, compared to Atg16l1-cKO BMBMs, BMDMs lacking both Atg16l1 and Zbp1 contained more high MW species of autophosphorylated RIPK1 (Figure 6G) and autophosphorylated RIPK3 (Figure 6H) after treatment with LPS/zVAD.

Combined deletion of Atg16l1 and Zbp1 in myeloid cells accelerates LPS-mediated sepsis

Beyond cellular necroptosis, myeloid-specific loss of Atg16l1 sensitizes mice to LPS-mediated sepsis in vivo (Samie et al., 2018). Since Zbp1 deletion enhanced TRIF-mediated necroptosis in Atg16l1-cKO BMDMs ex vivo, we asked whether loss of Atg16l1 and Zbp1 in myeloid cells would further exacerbate LPS-mediated sepsis. Intraperitoneal administration of LPS (10 mg per kg body weight) reproduced the previously observed sensitization of Atg16l1-cKO mice. Combined loss of Atg16l1 and Zbp1 significantly accelerated morbidity, with double-knockout mice succumbing to LPS-driven mortality by 14 hr. Loss of Zbp1 alone did not impact morbidity (Figure 7A; Figure 7—figure supplement 1A). Atg16l1 deficiency in myeloid cells exacerbated LPS-induced IL-1β and TNF in the serum, and the amount of IL-1β was yet higher in Atg16l1-cKO; Zbp1-cKO double-knockout mice (Figure 7B). Together, these results demonstrate that: 1) autophagy regulates ZBP1 abundance in macrophages, 2) elevated ZBP1 suppresses necroptosis when autophagy is perturbed, and 3) loss of both Atg16l1 and Zbp1 in myeloid cells accelerates LPS-mediated inflammation, enhancing morbidity in vivo.

Figure 7 with 1 supplement see all
Combined loss of myeloid-specific Atg16l1 and Zbp1 accelerates LPS-mediated sepsis in mice.

(A) Kaplan-Meier survival plots for mice following challenge with 10 mg/kg LPS administered intraperitoneally. Statistical analysis Figure 7—figure supplement 1A was performed using log-rank test (Figure 7—figure supplement 1; Figure 7—figure supplement 1A). (B) Serum cytokine measurements of IL-1β and TNFα performed by ELISA following 4 hr of intraperitoneal LPS administration at 10 mg/kg. Data in A are representative of two independent experiments. Data in B are pooled from two independent experiments. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

https://doi.org/10.7554/eLife.44452.027

Discussion

Macrophages represent primary cellular sensors of the innate immune system. During an inflammatory response, these phagocytes are armed to either propagate or resolve inflammation via antigen uptake and presentation, cytokine production, or induction of programmed cell death. Here, we show that defective autophagy in macrophages leads to an accumulation of modified RIPK1, RIPK3, and TRIF in response to pro-inflammatory signals, precipitating inflammation and necroptosis in a stimulus-dependent manner (Figure 7—figure supplement 1B,C). While RIPK1, RIPK3 and TRIF are well-established signaling factors required for necroptosis, our understanding of ZBP1 in this context is less developed. Recent studies have expanded on cytosolic nucleic-acid sensing by ZBP1 (Thapa et al., 2016; Kesavardhana et al., 2017; Kuriakose et al., 2016; Maelfait et al., 2017; Guo et al., 2018). ZBP1 was also shown to drive RIPK3-mediated necroptosis when the RIPK1-RHIM domain was mutated in vivo (Lin et al., 2016; Newton et al., 2016). However, its role in other forms of innate signaling is not known. Using an optimized protocol for CRISPR-mediated gene editing in primary macrophages, we revealed context-specific roles of TRIF and ZBP1 in regulating necroptosis and inflammatory cytokine production (Figure 7—figure supplement 1C). Specifically, we noted that TRIF can promote RIPK1 kinase-independent, RIPK3-dependent necroptosis when its accumulation is not checked by autophagy. TRIF can engage RIPK3 directly without the need for RIPK1 (Kaiser et al., 2013; Newton et al., 2016), so a scaffolding role for RIPK1 is unlikely. This notion cannot be confirmed genetically because RIPK1-deficiency alone triggers macrophage death (Newton et al., 2016). ZBP1 contains multiple RHIM-domains that support formation of amyloid-like structures (Li et al., 2012; Rebsamen et al., 2009). Accumulation of ZBP1 induced by defective autophagy may perturb optimal TRIF- or RIPK3- signaling via RHIM-mediated interference. Conclusive evidence for this possibility requires the mutation of ZBP1-RHIM domains in autophagy-deficient cells.

In mammalian cells, RHIM-dependent stacking of RIPK1 and RIPK3 has been described as a key feature of the necrosome (Li et al., 2012). RHIM-dependent accumulation of TRIF is also acknowledged as a pre-requisite for TRIF-mediated inflammatory signaling (Gentle et al., 2017). Biophysical characteristics of RHIM-domain protein complexes include a highly insoluble nature and resistance to denaturation/degradation (Fowler et al., 2007; Kleino et al., 2017; Lamour et al., 2017). Ubiquitination of cytosolic proteins is an established mechanism of substrate-identification by selective autophagy, and TRIF, RIPK1 and RIPK3 are ubiquitinated during inflammatory signaling (Yang et al., 2017; Choi et al., 2018; de Almagro et al., 2017). Although ubiquitinated RIPK1, and RIPK3 were more abundant after LPS/zVAD-induced necroptosis when autophagy was compromised, further investigation is needed to define the components of the selective autophagy machinery that drive the turnover of RHIM-domain proteins, and to determine whether ubiquitination is a critical step in this process.

Necroptosis is acknowledged as a potent pro-inflammatory mode of cell death, but there is an incomplete understanding of its role in normal tissue homeostasis and anti-microbial immunity. Genome-wide association (GWA) and functional studies have revealed that defects in autophagy promote inflammatory diseases such as Crohn’s disease, rheumatoid arthritis, lupus and neurodegeneration (Levine et al., 2011; Mizushima, 2007; Matsuzawa-Ishimoto et al., 2017; Samie et al., 2018). Our findings provide a potential link between defective autophagy and necroptotic signaling, with autophagy promoting the turnover of RHIM-containing proteins.

Materials and methods

Key resources table
Reagent type
(species) or
resource
DesignationSource or
reference
IdentifiersAdditional
information
Genetic reagent (M. musculus)Atg16l1loxP/loxPPMID: 24553140Dr. Aditya Murthy (Genentech, Inc)
Genetic reagent (M. musculus)Zbp1loxP/loxPNewton et al., 2016Dr. Kim Newton (Genentech, Inc)
Commercial assay or kitMouse monocyte isolation kitMiltenyi BiotecCat#: 130-100-629
Peptide, recombinant proteinCas9 V3IDTCat#: 108105810 μg per reaction
Peptide, recombinant proteinmurine TNFαPeprotechCat#: 315-01A50 ng/ml
Peptide, recombinant proteinPam3CSK4InvivogenCat#: tlrl-pms1 μg/ml
Peptide, recombinant proteinPolyI:C (LMW)InvivogenCat#: tlrl-picw10 μg/ml
Peptide, recombinant proteinLPS-EB ultrapure (E. coli O111:B4)InvivogenCat#: tlrl-3pelps100 ng/ml
Peptide, recombinant proteinR848 (Resiquimod)InvivogenCat#: tlrl-r8481 μg/ml
Peptide, recombinant proteinCpG-ODN 1826InvivogenCat#: tlrl-18265 μM
Peptide, recombinant proteinzVAD-fmkPromegaCat#: G723220 μM
Chemical compound, drugNecrostatin-1Enzo Life SciencesCat#: BML-AP309-010030 μM
Chemical compound, drugBafilomycin A1SigmaCat#: B1793100 nM
Chemical compound, drugMG132SigmaCat#: M74492 μM
Peptide, recombinant proteinFcR-BlockBD biosciencesCat#: 5331441
Chemical compound, drugFixable viability dye efluor780InvitrogenCat#: 65–0865
Antibodyanti-CD62L PerCP Cy5.5
Rat monoclonal
BD biosciencesCat#: 560513
RRID: AB_10611578
Flow cytometry
Antibodyanti-CCR2 APCR and D SystemsCat#: FAB5538A
RRID: AB_10645617
Flow cytometry
Antibodyanti-F4/80 efluor450
Rat monoclonal
eBioscienceCat#: 48-4801-82
RRID: AB_1548747
Flow cytometry
Antibodyanti-CSF1R BV711
Rat monoclonal
BiolegendCat#: 135515
RRID: AB_2562679
Flow cytometry
Antibodyanti-Ly6G BUV395
Rat monoclonal
BD biosciencesCat#: 565964
RRID: AB_2739417
Flow cytometry
Antibodyanti-CD11b BUV737
Rat monoclonal
BD biosciencesCat#: 564443
RRID: AB_2738811
Flow cytometry
Antibodyanti-MHCII (IA/IE) PE
Rat monoclonal
eBioscienceCat#: 12-5322-81
RRID: AB_465930
Flow cytometry
Antibodyanti-Ly6C-PECy7
Rat monoclonal
eBioscienceCat#: 25-5932-82
RRID: AB_2573503
Flow cytometry
Antibodyanti-CD45 FITC
Rat monoclonal
eBioscienceCat#: 11-0451-82
RRID: AB_465050
Flow cytometry
Antibodyanti-F4/80 BV421
Rat monoclonal
BiolegendCat#: 123131
RRID: AB_10901171
Flow cytometry
Antibodyanti-CD11b BUV395
Rat monoclonal
BD biosciencesCat#: 563553
RRID: AB_2738276
Flow cytometry
Antibodyanti-ATG16L1
Mouse monoclonal
MBL internationalCat#: M150-3
RRID: AB_1278758
Immunoblot
Antibodyanti-ATG14L
Rabbit polyclonal
MBL internationalCat#: PD026
RRID: AB_1953054
Immunoblot
Antibodyanti-FIP200
Rabbit monoclonal
Cell Signaling TechnologyCat#: 12436
RRID: AB_2797913
Immunoblot
Antibodyanti-Rubicon
Mouse monoclonal
MBL internationalCat#: M170-3
RRID: AB_10598340
Immunoblot
Antibodyanti-TRIF
Host: Rat
Genentech, IncCat#: 1.3.5Immunoblot
Antibodyanti-MLKL
Host: Rabbit
Genentech, IncCat#: 1G12Immunoblot
Antibodyanti-p-MLKL
Rabbit monoclonal
AbcamCat#: ab196436
RRID: AB_2687465
Immunoblot
Antibodyanti-RIPK1
Mouse monoclonal
BD biosciencesCat#: 610459
RRID: AB_397832
Immunoblot
Antibodyanti-p- RIPK1
Host: Rabbit
Genentech, IncCat#: GNE175.DP.B1Immunoblot
Antibodyanti-RIPK3
Rabbit polyclonal
Novus BiologicalsCat#: NBP1-77299
RRID: AB_11040928
Immunoblot
Antibodyanti-p-RIPK3
Host: Rabbit
Genentech, IncCat#: GEN-135-35-9Immunoblot
Antibodyanti-GSDMD
Host: Rat
Genentech, IncCat#: GN20-13Immunoblot
Antibodyanti-LC3B
Rabbit polyclonal
Cell Signaling TechnologyCat#: 2775
RRID: AB_915950
Immunoblot
Antibodyanti-CALCOCO1 Rabbit polyclonalProteintechCat#: 19843–1-AP
RRID: AB_10637265
Immunoblot
Antibodyanti-TAX1BP1
Rabbit monoclonal
AbcamCat#: ab176572Immunoblot
Antibodyanti-p62
Guinea pig polyclonal
Progen biotechnicCat#: gp62-c
RRID: AB_2687531
Immunoblot
Antibodyanti-NLRP3
Rabbit monoclonal
Cell Signaling TechnologyCat#: 15101
RRID: AB_2722591
Immunoblot
Antibodyanti-ASC
Rabbit monoclonal
Cell Signaling TechnologyCat#: 67824
RRID: AB_2799736
Immunoblot
Antibodyanti-STAT1
Rabbit monoclonal
Cell Signaling TechnologyCat#: 14995
RRID: AB_2716280
Immunoblot
Antibodyanti-p- STAT1
Rabbit monoclonal
Cell Signaling TechnologyCat#: 7649
RRID: AB_10950970
Immunoblot
Antibodyanti-M1-polyubiquitin linkage specific antibodyGenentech, IncN/AImmunoprecipitation
Antibodyanti-K63-polyubiquitin linkage specific antibodyGenentech, IncN/AImmunoprecipitation
AntibodyAnti-Ubiquitin
Mouse monoclonal
Cell Signaling TechnologyCat#: 3936
RRID: AB_331292
Immunoblot
Antibodyanti-beta ActinCell Signaling TechnologyCat#: 3700
RRID: AB_2242334
Immunoblot
Antibodyanti-rabbit IgG HRP
Goat polyclonal
Cell Signaling TechnologyCat#: 7074
RRID: AB_2099233
Immunoblot
Antibodyanti-mouse IgG HRP
Horse polyclonal
Cell Signaling TechnologyCat#: 7076
RRID: AB_330924
Immunoblot
Antibodyanti-rat IgG HRP
Goat polyclonal
Cell Signaling TechnologyCat#: 7077
RRID: AB_10694715
Immunoblot
AntibodyAnti-RagweedGenentech, IncN/AInhibition
AntibodyAnti-mIFNAR1
Mouse monoclonal
Leinco TechnologiesCat#: I-401
RRID: AB_2737538
Inhibition
AntibodymTNFR2-Fc
Mouse Fc
Genentech, IncN/AInhibition
Tools (software)Image JImmunoblot densitometry
Tools (software)Graphpad Prism 7GraphpadData visualization and statistics

Mice

Myeloid-specific deletion of Atg16l1 was achieved by crossing Lyz2-Cre + mice with Atg16l1loxP/loxP mice (described in 45). Conditional targeting of the Zbp1 locus was generated in C57BL/6NTac ES cells (Taconic) by introduction of loxP-sites flanking the ATG-containing exon 1, spanning the Zbp1 5’UTR and exon one corresponding to NCBI37/mm9 chr2:173,043,537–173,045,687 (described in 12). A 3xFLAG-tag was inserted in-frame with the ATG. Addition of the 3xFLAG-tag did not compromise ZBP1 function, since it failed to rescue the previously described lethality of Ripk1-RHIM mutant mice (Guo et al., 2018) (Supplementary file 1). Myeloid-specific deletion of Zbp1 was achieved by crossing Lyz2-Cre + mice with Zbp1loxP/loxP mice. Combined deletion of Atg16l1 and Zbp1 was generated by crossing the above two strains of mice. eGFP-reporter mice were obtained from Jackson labs (strain 57BL/6-Tg(CAG-EGFP)1Osb/J, Stock No: 003291). None of the in vivo experiments were randomized. No statistical method was used to pre-determine group sample size, and investigators were not blinded to group allocations or study outcomes.

LPS-driven sepsis

Intraperitoneal administration of LPS (E. coli O111:B4, Sigma L2630) was performed at 10 mg/kg dissolved in a maximum of 200 μL sterile phosphate-buffered saline (PBS). Mice were monitored for morbidity and body temperature every 4 hr for the first 14 hr, followed by monitoring at 24 and 48 hr. Blood was obtained at 4 hr post LPS-administration for serum cytokine analysis. Experiments were performed using age- and sex-matched cohorts from a single colony. All protocols were approved by the Genentech Institutional Animal Care and use Committee; all studies were executed by following relevant ethical regulations detailed in animal use protocols (internal protocol 18–1823).

Cell culture

Murine monocytes were obtained from femoral bone marrow by negative selection using a monocyte isolation kit (Miltenyi Biotec). Monocyte-derived macrophages were cultured in macrophage medium [high glucose Dulbecco’s Minimum Essential Media (DMEM) + 10% FBS+GlutaMAX (Gibco) +Pen/Strep (Gibco) supplemented with 50 ng/ml recombinant murine macrophage-colony stimulating factor (rmM-CSF, Genentech)]. Bone marrow-derived macrophages were generated by culture of total femoral bone marrow in macrophage medium on 15-cm non-TC treated plates for 5 days (Petri dish, VWR). Fresh medium was added on day 3 without removal of original media. On day 5, macrophages were gently scraped from dishes, counted and re-plated on TC treated plates of the desired format for downstream assays. After overnight culture in macrophage medium, assays were performed on day 6 BMDMs. CRISPR-edited BMDMs were treated on day 10 to permit complete protein loss of targeted genes. BMDMs were stimulated with 50 ng/ml murine TNFα (Peprotech), 1 μg/ml Pam3CSK4, 10 μg/ml poly(I:C) LMW, 100 ng/ml ultra-pure LPS unless otherwise stated (LPS, E. coli 0111:B4), 1 μg/ml R848 or 5 μM CpG-ODN 1826 (all from Invivogen). zVAD-fmk was added at 20 μM (Promega). Necrostatin-1 (Nec-1) was added at 30 μM (Enzo Life Science). DMSO was added at 0.1% as vehicle control (Sigma). For cell death assays, BMDMs were plated at 2 × 104 cells/well in flat-bottom 96-well plates. The following day, cells were stimulated as indicated to induce necroptosis. BMDM viability was assessed by propidium iodide (PI, Invitrogen) staining using live-cell imaging, measuring PI-positive cells per mm2 (Incucyte systems, Essen Biosciences). Percent cell death was calculated by dividing PI-positive cells per mm2 with total plated cells per mm2. Total cell plated cells were enumerated by independently plating BMDMs and staining with a nuclear dye fluorescing in the same channel as PI (Nuclear-ID Red, Enzo Life Science) or addition of 0.1% Triton X-100 in the presence of PI. Time points between 12 and 16 hr were used to compare cell death unless otherwise stated. ZBP1 turnover measured by cycloheximide-chase assays was performed with 100 μg/ml CHX in DMSO for the indicated time points (Sigma). ZBP1 degradation assays in WT BMDMs were performed using 100 nM Bafilomycin A1 (Sigma) or 2 μM MG132 (Sigma) for the indicated time points.

Gene editing

CRISPR/Cas9-mediated deletion of genes was performed by electroporation of Cas9 RNP in monocytes and BMDMs. Briefly, 5 × 106 primary monocytes from the bone marrow or day five cultured BMDMs were electroporated with recombinant Cas9 (IDT) complexed with gene-specific guide RNAs (Supplementary file 2). Briefly, locus-specific crRNAs were annealed with tracrRNAs at a 1:1 stoichiometric ratio at 95°C for 5 min followed by complexing with recombinant Cas9 at a 3 µL:1 µL gRNA:Cas9 ratio per guide RNA to generate the RNP complex. two guide RNAs were combined per gene. Cells were resuspended in 20 µL nucleofector solution P3 and RNP complex added. This mixture was aliquoted into 16-well nucleofector strips (Lonza) and electroporated using program CM-137 (4D-Nucleofector, Lonza). Following electroporation, cells were grown in non-tissue culture treated dishes (VWR) for an additional 5 days in macrophage media. On day 5, cells were scraped from dishes and re-plated as required for functional assays in tissue-culture treated multi-well plates.

Flow cytometry

Monocyte-derived macrophages or BMDMs were harvested and washed in cold PBS. Cells were incubated in Fc-block reagent (BD biosciences) and fixable viability dye eFluor 780 (Invitrogen) for 15 min at 4°C in cold PBS. Monocyte-derived macrophages were washed once and stained with the following antibodies: anti-CD62L PerCp-Cy5.5 (BD biosciences), anti-CCR2 APC (R and D Systems), anti-F4/80 efluor450 (eBioscience), anti-CSF1R BV711 (Biolegend), anti-Ly6G BUV395 (BD biosciences), anti-CD11b BUV737 (BD biosciences), anti-MHCII(I-A/I-E) PE (eBioscience) and anti-Ly6C PE-Cy7 (eBioscience). BMDMs were washed once and stained with the following antibodies: anti-CD45 FITC (eBioscience), anti-F4/80 BV421 (Biolegend), anti-CD11b BUV395 (BD biosciences). Stained cells were analyzed on by flow cytometry using a BD FACS CANTO instrument. Loss of eGFP or CD45 was assessed by gating on live F4/80 + macrophages using FlowJo X.

ELISA

BMDM cell culture medium or murine serum was analyzed for measurement of cytokines IL-1β and TNFα (eBioscience) by ELISA following manufacturer’s protocols.

Immunoblotting

To assay CRISPR-mediated gene deletions, cell pellets were lysed in RIPA buffer +protease and phosphatase inhibitors (Roche). Supernatants were obtained after high speed centrifugation and protein concentration measured using the BCA assay (Thermo Fisher). To perform detergent soluble and insoluble fractionation, cell pellets were lysed in 1% NP-40 lysis buffer (150 mM NaCl, 20 mM Tris-HCl pH 7.5, 1% NP-40, 1 mM EDTA, protease and phosphatase inhibitors). Lysates were flash frozen on dry-ice, thawed on ice and vortexed for 10 s followed by centrifugation at 1000 g for 10 min to remove nuclear pellets. Supernatants were centrifuged at 15000 rpm (or highest speed) for 15 min in a refrigerated table-top centrifuge. Resultant supernatants were collected as NP-40 soluble fractions. NP-40 insoluble pellets were resuspended in 1% NP-40 lysis buffer supplemented with 1% SDS. The suspension was homogenized by passing through a 26-gauge needle, and protein quantification performed using BCA assay. To assess ubiquitination, cells were lysed under denaturing conditions with lysis buffer (9 M urea and 20 mM HEPES, pH 8.0) containing 1 mM sodium orthovanadate, 2.5 mM sodium pyrophosphate, 1 mM beta-glycerolphosphate and incubated for 20 min with vigorous shaking at 900 rpm at room temperature. Following incubation, cell lysates were centrifuged for 10 min at 14,000 rpm. Lysates were then diluted two times with buffer (20 mM HEPES, pH 8.0) containing Roche protease inhibitor cocktail, 100 µM PR-619 (SI9619, Life Sensors) and 100 µM 1,10-phenanthroline (SI9649, Life Sensors) and used for immunoprecipitation with ubiquitin chain-specific antibodies and protein-A/G beads overnight at 4°C as previously described in Goncharov et al. (2018). SDS-PAGE was performed using a 4–12% gradient Bis-Tris gel (Novex), followed by protein transfer onto PVDF membranes and antibody incubation. Immunoblots were detected by enhanced chemiluminescence (western lightning-plus ECL, Perkin Elmer). Antibodies used: anti-Atg16l1 (MBL international), anti-Atg14l (MBL international), anti-FIP200 (Cell Signaling Technology), anti-Rubicon (MBL), anti-TRIF (Genentech), anti-MLKL (Genentech), anti-RIPK1 (BD Biosciences), anti-pSer345 MLKL (Abcam), anti-pSer166/Thr169 RIPK1 (Genentech), anti-RIPK3 (Novus Biologicals), anti-pThr231/Ser232 RIPK3 (Genentech), anti-GSDMD (Genentech,), anti-LC3B (Cell Signaling Technology), anti-Calcoco1 (Proteintech), anti-Tax1bp1 (Abcam), Sqstm1/p62 (Progen biotechnic), anti-NLRP3 (Cell Signaling Technology), anti-ASC (Cell Signaling Technology), anti-STAT1 (Cell Signaling Technology), anti-pTyr701 STAT1 (Cell Signaling Technology), anti-M1 polyubiquitin linkage-specific antibody (Genentech), anti-K63 polyubiquitin linkage-specific antibody (Genentech), anti-β-actin (Cell Signaling Technology), anti-rabbit IgG-HRP (Cell Signaling Technology), anti-mouse IgG-HRP (Cell Signaling Technology), anti-rat-HRP Ig (Cell Signaling Technology). ImageJ was used to quantify immunoblot density.

Statistical analysis

Pairwise statistical analyses were performed using an unpaired Student’s two-sided t-test to determine if the values in two sets of data differ. Correction for multiple-comparisons was performed using the Holm-Sidak method with α = 0.05. Scatterplot bars and connected dot plots present means of data. Analysis of kinetic (time) or LPS dose-response datasets were performed using two-way ANOVA followed by multiple comparison testing. Line graphs and associated data points represent means of data; error bars represent standard deviation from mean. For LPS-mediated sepsis studies, a log-rank (Mantel-Cox) test was used to assess significance of the differences between indicated groups in their survival. GraphPad Prism seven was used for data analysis and representation.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58

Decision letter

  1. Tadatsugu Taniguchi
    Senior Editor; Institute of Industrial Science, The University of Tokyo, Japan
  2. Facundo D Batista
    Reviewing Editor; Ragon Institute of MGH, MIT and Harvard, United States
  3. Rupert Beale
    Reviewer
  4. Katherine Fitzgerald
    Reviewer; UMASS, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "Autophagy regulates inflammatory programmed cell death via turnover of RHIM-domain proteins" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Tadatsugu Taniguchi as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Rupert Beale (Reviewer #2); Katherine Fitzgerald (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission. As you will see from their reports, all three reviewers have found the manuscript interesting and worthy of inviting for a re-submission. However, they have raised several points that need to be addressed to increase the general impact of the work. In particular, there was a general concern among reviewers to better address the mechanism.

1) The authors should perform a time course of the process comparing WT vs. Atg16L1 KO to substantiate their claim that autophagy is important to attenuate/resolve necroptosis in their system. In this time course, different parameters must be analyzed –% cell death, accumulation of detergent-insoluble forms of TRIF, RIPK1 and RIPK3 among others.

2) Autophagy is presented as an essential process, however the mechanism is only studied according to the consequences of knocking down different autophagy proteins. The authors should characterize autophagy in more detail by monitor autophagy flux after cell death induction (LC3 lipidation, p62 levels, mTORC1 activation among others).

3) The authors should provide more evidence as to the type of death – necroptosis vs. pyroptosis.

4) Figure 3 shows accumulation of detergent-insoluble forms of TRIF, RIPK1, RIPK3, We consider that detecting ubiquitination of these proteins will support their hypothesis. An analysis of the role of TaxBP1 (and other autophagy receptors) would greatly strengthen the manuscript.

5) The comments on the statistics need to be addressed.

I am adding the reviewers' comments so that you get an idea of their different views as well as experimental work that you might carry out to add to accommodate their concerns.

Reviewer #1:

In this paper, Lim et al., proposes a novel mechanism in macrophages, in which autophagy, promoting the turnover of RHIM-domain proteins (RIPK1, RIPK3, TRIF and ZBP1), induces the suppression/attenuation of necroptosis, a type of necrotic cell death.

Necrotic cell death (necroptosis and Pyroptosis) protect host against microbial pathogens and dysregulation of this process has been involved in autoimmune and auto-inflammatory conditions. In our opinion authors try to address an important question in this context: what is the mechanism that drives returning to homeostasis after necrotic cell death induction?. They do so mainly doing extensive in vitro work, claiming that autophagy attenuates necroptosis through the selective degradation of RHIM-domain proteins. To reach this conclusion they use a very well-known model: conditional Atg16L KO mice lacking this key autophagy protein in myeloid cells.

However, although the work is interesting and with high potential to end up being an important contribution to the field, we find that authors overestimated their results making too simple conclusions. In general:

- They claim that the type of necrotic cell death that is being detected/quantified is Necroptosis. In our opinion a deeper analysis is required to make this claim. Considering the crosstalk between necroptosis and pyroptosis, we think that authors have oversimplified their conclusions.

- They make a strong statement describing canonical autophagy as a mechanism driving the attenuation of necroptosis. Although it is potentially interesting, especially considering the current interest in selective autophagy, authors must improve this part of the work.

Therefore, authors have to revise the work obtaining convincing data to support their main claims. We think that the following points have to be addressed to make this work suitable for publication:

- Authors quantify% of cell death, and they conclude that this is necroptosis. We do understand that using the combination of TNF/LPS+zVAD is a well-established system to induce this type of death cell in the context of WT cells. However, it is not known if this increase of% death cells in Atg16L KO cells is necroptosis or pyroptosis. Considering the crosstalk among these two processes, the existence of Gsdmd-independent pyroptosis (Schneider et al., 2017), and the failure rescue with Nec-1 in the case of LPS, we think that the authors should further characterize this kind of cell death.

- Authors claim that autophagy is important to attenuate/resolve necroptosis in their system. To do this claim, we think that it is necessary to do a time course of the process comparing WT vs. Atg16L1 KO. In this time course, different parameters must be analyzed –% cell death, accumulation of detergent-insoluble forms of TRIF, RIPK1 and RIPK3 among others.

- Autophagy is presented as an essential process. This mechanism is only studied according to the consequences of knocking down different autophagy proteins. However, we consider that authors should describe autophagy in more detail. They could monitor autophagy flux after cell death induction. Actually, an important issue in all this work is the fact of obtaining different results with TLR3 and TLR4. Maybe autophagy flux is different in these two contexts.

- Authors see an accumulation of RHIM-domain proteins in Atg16L KO cells. They should try to recapitulate this accumulation using lysosome inhibitors to prove that autophagy core proteins are being used in a lysosomal degradation process (they do so for ZBP1, Figure 6—figure supplement 1).

- As we highlighted before, the technique used to induce necroptosis is well established in the field. However, it is known that this method induces autophagy. Would it be possible to use another technique to induce necroptosis, without targeting autophagy?

- Figure 3 shows accumulation of detergent-insoluble forms of TRIF, RIPK1 and RIPK3. Authors comment in the Discussion (second paragraph) about the possibility of protein ubiquitination playing a role in this process. We consider that detecting ubiquitination of these proteins will support their hypothesis considering that authors make a strong statement about the role of selective autophagy in this process.

Reviewer #2:

This manuscript extends on observations reported by this laboratory in Samie et al., 2018, which reported Tax1BP1 mediated selective autophagy of TRIF. Here, the authors demonstrate that oligomerised RHIM domain proteins (including TRIF) accumulate in ATG16L1 deficient macrophages. They investigate autophagy as a possible homeostatic mechanism for clearing these oligomers, thereby opposing inflammation and cell death. Consistent with this, loss of essential autophagy genes resulted in enhanced cell death in BMDMs stimulated with LPS or Poly I:C and zVAD. A notable strength of this paper which distinguishes it from previous observations on RIP kinases and autophagy is that the authors provide evidence of involvement of the FIP200 and ATG14, which indicates this is due to canonical macroautophagy and not a 'non-canonical' or LAP-like process. The loss of Rubicon makes no difference, and whilst this does not rule out a contribution from 'non-canonical' autophagy (see discussion in Fletcher, Ulferts et al. EMBOJ 2018), taken together with the other genetic data it strongly argues against LAP being required here. The data to support this are shown in Figures 1F and Supplementary Figure 2B, the paper would be much easier to read if these data were grouped together.

The authors demonstrate that Necrostatin fails to completely suppress necroptosis in autophagy deficient cells, and show this depends on Mlkl and Ripk3. ATG16L1 deficient cells accumulate detergent insoluble forms of RHIM domain proteins, consistent with a role for autophagy in their degradation. The authors go on to show that ZBP1 accumulates in ATG16L1 deficient cells, but that curiously this protects against necroptosis. This is interesting, but the effect only seems to take place in the setting of ATG16L1 deficiency so it remains unclear if there is a physiological relevance.

Absent from this paper are any direct measurements of autophagy or autophagic flux. It would be very helpful to the authors' argument to demonstrate that oligomerised RHIM-domain proteins do indeed become selectively engulfed by autophagosomes. As a minimum I would expect to see data on accumulation of autophagosomes and associated lipidation of LC3 after RHIM-domain oligomerisation, and (crucially) the effect of ATG14/FIP200/ATG16L1 deletion on this process. This would provide additional strong evidence of regulation of signalling by canonical autophagy. I accept the authors argument that a full mechanistic explanation of exactly how the proteins are ubiquitinated and degraded (assuming this is the case) is beyond the scope of the paper, but given their previous findings of Tax1BP1 being a critical component required for degradation of TRIF (at least in mice, the situation in humans may well be different), it is surprising that the authors do not comment at all on this. It would greatly strengthen the paper to see data on the requirement (or otherwise) of autophagy receptors such as TaxBP1, p62 and (in humans) NDP52.

In summary this paper provides an important if somewhat limited advance in our understanding of the regulation of necroptosis. The conclusions are in general well supported by the data. The paper would be greatly strengthened by exploring the mechanisms of selective autophagy in this setting.

Additional data files and statistical comments:

Many of the pairwise analyses throughout require correction for multiple hypotheses.

Reviewer #3:

In this manuscript, Lim et al. show that autophagy is involved in clearing activation of the necroptotic machinery in order to limit cell death. Inhibition of autophagy leads to increased cell death following ligand /caspase inhibition. This was shown to be caused by increased oligomerization of necroptotic components and reduced clearance. Interestingly, whereas ZBP1 has previously been shown to be capable of triggering necroptosis, loss of ZBP1 in autophagy-deficient cells lead to further increase in death, suggesting that in this scenario, ZBP1 keeps necroptosis in check. The authors further show that mice lacking canonical autophagy and ZBP1 do worse in a model of septic shock.

The paper makes a valuable contribution to understanding the regulatory machinery of cell death pathways and reveals the importance of autophagy in limiting necroptosis. They also show that when autophagy is perturbed, necroptosis can progress independently of RIPK1. Overall the paper is well written and the data very compelling. However some additional insights on how ZBP1 can repress necroptosis would strengthen this study.

– The authors should provide some insights into how ZBP1 functions in repressing necroptosis.

– Ripk1 inhibition could not completely block necroptosis when TRIF is involved (TLR3 and TLR4 stimuli in autophagy deficient cells). Can RIPK3 be recruited to the TRIFosome and activated in the absence of RIPK1? Is it just the kinase activity of RIPK1 that becomes redundant (i.e. is scaffolding still required)? Use of RIPK1 sgRNA could address this.

– How is necroptosis triggered in non-TRIF mediated TLR triggered necroptosis? Through autocrine secretion of TNF or IFN?

– The recent paper (Sarhan et al., 2018) suggests that autocrine IFN is required for necroptosis. Is IFN involved in necroptosis when autophagy is not functional? Does perturbed clearance of TRIF and RIPK1 result in higher IFN signaling and thus higher cell death?

– Figure 3D and F suggests that when autophagy is deranged, RIPK1- and RIPK3- oligomerization is dependent on TRIF but not phosphorylation of RIPK1 and RIPK3?

– The term "screening" is a bit misleading in Figure 1. They optimized electroporation conditions for delivery of sgRNA:Cas9 RNPs into primary BMDMs. This could be added to the supplementary figure.

https://doi.org/10.7554/eLife.44452.035

Author response

As you will see from their reports, all three reviewers have found the manuscript interesting and worthy of inviting for a re-submission. However, they have raised several points that need to be addressed to increase the general impact of the work. In particular, there was a general concern among reviewers to better address the mechanism.

1) The authors should perform a time course of the process comparing WT vs. Atg16L1 KO to substantiate their claim that autophagy is important to attenuate/resolve necroptosis in their system. In this time course, different parameters must be analyzed –% cell death, accumulation of detergent-insoluble forms of TRIF, RIPK1 and RIPK3 among others.

We have performed time-course analysis of necroptosis in WT vs. Atg16l1-cKO BMDMs by LPS or PolyI:C. These demonstrate accelerated necroptosis in Atg16l1-cKO BMDMs over 18 hours of treatment (Figure 4, Figure 4—figure supplement 1B, C). Interestingly, PolyI:C alone exhibits elevated death of Atg16l1-cKO BMDMs (Figure 4—figure supplement 1C, PolyI:C time-course). This demonstrates that TLR3, which only signals through TRIF, can induce death of autophagy-deficient macrophages and is consistent with the observed over-accumulation of TRIF in these cells. We also show accumulation of detergent-insoluble forms of TRIF (Figure 4B), autophosphorylated and total RIPK1 (Figure 4C), and autophosphorylated and total RIPK3 (Figure 4D) in both wild-type and Atg16l1-cKO BMDMs over 6 hours of LPS/zVAD treatment. Loss of Atg16l1 enhances accumulation of active TRIF, RIPK1 and RIPK3 with varying kinetics. These new findings are discussed in the second paragraph of the subsection “Loss of autophagy results in accumulation of active forms of TRIF, RIPK1 and RIPK3 during necroptosis”.

2) Autophagy is presented as an essential process, however the mechanism is only studied according to the consequences of knocking down different autophagy proteins. The authors should characterize autophagy in more detail by monitor autophagy flux after cell death induction (LC3 lipidation, p62 levels, mTORC1 activation among others).

We analyzed wild-type BMDMs in a time-course over 6 hours using the well-established method of halting lysosomal degradation by the lysosomal inhibitor Bafilomycin A1 to block autophagic flux. It is not possible to study kinetics of autophagy in ATG16L1 deficient cells since they cannot perform autophagy. We observe a rapid accumulation of lipidated LC3B (LC3-II), a marker of autophagosome maturation, during LPS/zVAD mediated necroptosis in the presence of Bafilomycin A1 (Figure 4—figure supplement 2G). Importantly, selective autophagy receptors SQSTM1/p62, CALCOCO1 and TAX1BP1 preferentially accumulate in detergent-insoluble fractions when lysosomal function is blocked (compare Figure 4—figure supplement 2G vs. 2H). These new findings are discussed in the subsection “The autophagy receptor TAX1BP1 prevents TRIF-mediated necroptosis”.

Along with the above autophagy receptors, we now show that active TRIF, RIPK1 and RIPK3 also accumulate in detergent-insoluble fractions of wild-type BMDM lysates when lysosomal function is inhibited (Figure 4—figure supplement 2A-C). In contrast, basal turnover of TRIF, RIPK1 and RIPK3 does not seem to depend on lysosomal turnover via autophagy (Figure 4—figure supplement 2D-F). Together, these data clearly demonstrate that autophagic flux and lysosomal function are critical for the turnover of active forms of TRIF, RIPK1 and RIPK3 during necroptosis. These are discussed in the last paragraph of the subsection “Loss of autophagy results in accumulation of active forms of TRIF, RIPK1 and RIPK3 during necroptosis”.

None of our studies were performed under nutrient-deprived conditions used to induce mTOR-mediated autophagy. We thus focused our analysis on markers of autophagosome maturation and selective autophagy. The investigation of starvation-induced signals (i.e. mTOR, AKT) are outside the scope of the current manuscript. We believe that our efforts provide sufficient evidence to demonstrate that autophagic flux is induced in wild-type BMDMs during necroptosis.

3) The authors should provide more evidence as to the type of death - necroptosis vs. pyroptosis.

Reviewer #1 proposed that we investigate secondary pyroptosis as it was recently demonstrated to drive Gsdmd-independent pyroptosis. Schneider et al., 2017, identified NLRP3 and ASC were critical drivers of GSDMD-independent pyroptosis in BMDCs. Thus, in addition to our previous demonstration that RIPK1-independent BMDM death depended on RIPK3 and MLKL, we have now generated Nlrp3 and Pycard/ASC knockdown by CRISPR/Cas9 (Figure 2—figure supplement 1B). Loss of Nlrp3 or Pycard/ASC did not protect BMDMs from death by LPS/zVAD or PolyI:C/zVAD. Furthermore, it did not fully rescue RIPK1-independent death (Figure 2—figure supplement 1C). These data are consistent with our previous observations that necroptosis driven by RIPK3 and MLKL is the relevant type of cell death in our experimental system. We discuss these new findings in the first paragraph of the subsection “TRIF and RIPK1 drive necroptosis in Atg16l1-deficient macrophages”.

4) Figure 3 shows accumulation of detergent-insoluble forms of TRIF, RIPK1, RIPK3, We consider that detecting ubiquitination of these proteins will support their hypothesis. An analysis of the role of TaxBP1 (and other autophagy receptors) would greatly strengthen the manuscript.

Detecting ubiquitination of detergent-insoluble proteins is a technically challenging assay due to the harsh denaturing conditions (9M Urea) required. Nonetheless, we have successfully measured ubiquitination of RIPK1 and RIPK3 in primary BMDMs under these conditions. Using a mixture of M1 and K63-linkage-specific antibodies, we successfully immunoprecipitated autophosphorylated RIPK1 and RIPK3 from wild-type and Atg16l1-cKO BMDMs following LPS/zVAD treatment. As shown in Figure 4E, we demonstrate that loss of Atg16l1 results in accumulation of poly-ubiquitinated forms of autophosphorylated RIPK1 and RIPK3 upon LPS/zVAD treatment. This is consistent with our previous observations that high molecular weight forms of autophosphorylated RIPK1 and RIPK3 accumulate in Atg16l1-cKO BMDMs (Figure 3D, F; Figure 4C, D). These data are discussed in the second paragraph of the subsection “Loss of autophagy results in accumulation of active forms of TRIF, RIPK1 and RIPK3 during necroptosis”.

In line with findings in response to question 2, we demonstrate that loss of Atg1l1 phenocopies Bafilomycin A1-mediated lysosomal inhibition, since SQSTM1/p62, TAX1BP1 and CALCOCO1 accumulate in detergent-insoluble fractions of Atg16l1-cKO BMDMs to a greater degree than wild-type controls over 6 hours of necroptosis by LPS/zVAD (Figure 5B). To demonstrate a functional role for autophagy receptors in BMDM necroptosis, we have successfully generated CRISPR-mediated knockdown of Sqstm1/p62, Tax1bp1 or Calcoco1 in wild-type BMDMs (Figure 5C). These autophagy receptors were previously identified as candidate receptors accumulating in Atg16l1-cKO BMDMs (Samie et al., 2018). Induction of LPS/zVAD or PolyI:C/zVAD-mediated necroptosis revealed that knockdown of Tax1bp1 significantly increased BMDM necroptosis (Figure 5D). Consistent with the phenotypes of BMDMs lacking ATG16L1, ATG14L or FIP200, TAX1BP1 deficiency resulted in Necrostatin-1 resistant cell death following activation of TRIF signaling via TLR3 or TLR4 (Figure 5D, PolyI:C+zVAD+Nec-1 or LPS+zVAD+Nec-1 conditions). Together, these new data demonstrate that autophagic flux is rapidly induced as a mechanism of RHIM-protein clearance during necroptosis. The selective autophagy receptor TAX1BP1 appears important in this pathway, since its deletion in wild-type BMDMs phenocopies autophagy deficiency. These data are described in the subsection “The autophagy receptor TAX1BP1 prevents TRIF-mediated necroptosis”.

5) The comments on the statistics need to be addressed.

We have revised our pairwise analyses to correct for multiple comparisons (Holms-Sidack method, α=0.05). This has not impacted any conclusions in the original submission. We have amended our Materials and methods appropriately to reflect this revision (subsection “Statistical analysis”).

I am adding the reviewers' comments so that you get an idea of their different views as well as experimental work that you might carry out to add to accommodate their concerns.

Reviewer #1:

[…] Although the work is interesting and with high potential to end up being an important contribution to the field, we find that authors overestimated their results making too simple conclusions. In general:

- They claim that the type of necrotic cell death that is being detected/quantified is Necroptosis. In our opinion a deeper analysis is required to make this claim. Considering the crosstalk between necroptosis and pyroptosis, we think that authors have oversimplified their conclusions.

- They make a strong statement describing canonical autophagy as a mechanism driving the attenuation of necroptosis. Although it is potentially interesting, especially considering the current interest in selective autophagy, authors must improve this part of the work.

Therefore, authors have to do a better work obtaining convincing data to support their main claims. We think that the following points have to be addressed to make this work suitable for publication:

- Authors quantify% of cell death, and they conclude that this is necroptosis. We do understand that using the combination of TNF/LPS +z VAD is a well-established system to induce this type of death cell in the context of WT cells. However, it is not known if this increase of% death cells in Atg16L KO cells is necroptosis or pyroptosis. Considering the crosstalk among these two processes, the existence of Gsdmd-independent pyroptosis (Schneider et al., 2017), and the failure to rescue with Nec-1 in the case of LPS, we think that the authors should further characterize this kind of cell death.

We thank the Reviewer for this comment. RIPK3 and MLKL are well-established as bona-fide gatekeepers of necroptosis, and in our initial submission we demonstrated that these genes were required for death of Atg16l1-cKO BMDMs, even under RIPK1-independent (Nec-1 resistant) conditions. Nonetheless, we acknowledge that additional modes of death could account for the Nec-1 resistant phenotype. In our revised manuscript, we provide additional evidence to support our conclusion that the type of cell death observed in our study is necroptosis. CRISPR-mediated knockdown of Nlrp3 or Pycard/ASC, recently demonstrated to drive GSDMD-independent pyroptosis (Schneider et al., 2017), did not rescue Nec-1 resistant death of Atg16L1-cKO BMDMs (Figure 2—figure supplement 1B, C). In contrast, knockdown of Ripk3 or Mlkl fully rescued Nec-1 resistant death of Atg16l1-cKO BMDMs (Figure 2B, Figure 2—figure supplement 1A). Together with the lack of rescue observed by Gsdmd deletion (Figure 2B, Figure 2—figure supplement 1A), we believe we have stronger evidence to support necroptosis as the primary form of cell death in our study. These data are discussed in the first paragraph of the subsection “TRIF and RIPK1 drive necroptosis in Atg16l1-deficient macrophages”.

- Authors claim that autophagy is important to attenuate/resolve necroptosis in their system. To do this claim, we think that it is necessary to do a time course of the process comparing WT vs. Atg16L1 KO. In this time course, different parameters must be analyzed –% cell death, accumulation of detergent-insoluble forms of TRIF, RIPK1 and RIPK3 among others.

The studies prompted by this comment proved highly informative in better describing the process of necroptosis in Atg16l1-cKO BMDMs. Thus, we have generated a new main figure (Figure 4) and supplementary figure (Figure 4— figure supplement 1). In these new datasets, we describe accelerated death of Atg16l1-cKO BMDMs following necroptosis induced by LPS or PolyI:C. Loss of Atg16l1 resulted in >80% cell death by 3 hours in both settings, compared to approximately 50% of wild-type BMDMs (Figure 4A; Figure 4—figure supplement 1B, C). We thus focused our time course to 6 hours of treatment, and show that TRIF, RIPK1 and RIPK3 accumulate in detergent-insoluble fractions with distinct kinetics. TRIF accumulates rapidly and transiently, with a greater accumulation observed in Atg16l1-cKO BMDMs (Figure 4B). Autophosphorylated and total RIPK1 accumulate at 2 hours, with sustained and increased levels in Atg16l1-cKO BMDMs (Figure 4C). Autophosphorylated and total RIPK3 accumulates with different kinetics between wild-type and Atg16l1-cKO BMDMs. Whereas maximal levels are observed at 6 hours in wild-type BMDMs, loss of Atg16l1 accelerates this process. Maximum levels of p-RIPK3 and total RIPK3 are observed at 2 hours of LPS/zVAD treatment (Figure 4D). These new datasets provide insight into the accelerated cell death of BMDMs during defective autophagy.

- Autophagy is presented as an essential process. This mechanism is only studied according to the consequences of knocking down different autophagy proteins. However, we consider that authors should describe autophagy in more detail. They could monitor autophagy flux after cell death induction. Actually, an important issue in all this work is the fact of obtaining different results with TLR3 and TLR4. Maybe autophagy flux is different in these two contexts.

We provide new datasets describing autophagic flux during necroptosis in wild-type BMDMs (loss of Atg16L1 prevents analysis of autophagic flux in these cells). Lysosomal inhibition with Bafilomycin A1 is a well-established method to perturb autophagic flux and permit the accumulation of autophagic cargo destined for lysosomal degradation. By adding Bafilomycin A1 during LPS/zVAD mediated necroptosis, we observe rapid accumulation of lipidated LC3B (LC3-II), a hallmark of autophagosome biogenesis (Figure 4—figure supplement 2G). Additionally, autophagy receptors are known to traffic to the lysosome to regulate turnover of autophagic cargo. Consistent with this, we observe accumulation of specific autophagy receptors SQSTM1/p62, TAX1BP1 and CALCOCO1 in detergent-insoluble fractions of necroptotic wild-type BMDMs in the presence of Bafilomycin A1 (Figure 4—figure supplement 2G). Loss of Atg16l1 phenocopies Bafilomycin A1-mediated lysosomal inhibition, since SQSTM1/p62, TAX1BP1 and CALCOCO1 accumulate in detergent-insoluble fractions of Atg16l1-cKO BMDMs to a greater degree than wild-type controls over 6 hours of necroptosis` by LPS/zVAD (Figure 5B). Together, these findings more conclusively demonstrate that autophagy is rapidly induced in BMDMs during necroptosis. We discuss these findings in the subsection “The autophagy receptor TAX1BP1 prevents TRIF-mediated necroptosis”.

Comparing kinetics of cell death by TLR3 and TLR4 stimulation revealed that PolyI:C was sufficient to reveal enhanced cell death in Atg16L1-cKO BMDMs even in the absence of zVAD-fmk (Figure 4—figure supplement 1B, LPS only vs. Figure 4—figure supplement 1C, PolyI:C only). This is consistent with the knowledge that TLR3 only signals through TRIF, resulting in the generation of a death-inducing complex termed the RIPoptosome (Kaiser, 2013). While TLR4 is capable of this as well, it can engage MYD88 to promote pro-survival pathways that antagonize RIPoptosome activity. We thank the reviewer for this comment and discuss these points in the second paragraph of the subsection “Loss of autophagy results in accumulation of active forms of TRIF, RIPK1 and RIPK3 during necroptosis” and in the first paragraph of the subsection “TRIF and RIPK1 drive necroptosis in Atg16l1-deficient macrophages”.

- Authors see an accumulation of RHIM-domain proteins in Atg16L KO cells. They should try to recapitulate this accumulation using lysosome inhibitors to prove that autophagy core proteins are being used in a lysosomal degradation process (they do so for ZBP1, Figure 6—figure supplement 1).

In order to demonstrate that lysosomal degradation is responsible for the turnover of TRIF, RIPK1 and RIPK3, we induced LPZ/zVAD-mediated necroptosis in wild-type BMDMs in the presence of Bafilomycin A1. Over a 6-hour time-course, we observed accumulation of high molecular weight forms of TRIF, RIPK1 and RIPK3 in detergent-insoluble fractions of wild-type BMDMs varying kinetics (Figure 4—figure supplement 2A-C). TRIF maximally accumulated by approximately 1 hour (Figure 4—figure supplement 1A), whereas maximal levels of RIPK1 and RIPK3 were observed at approximately 6 hours of treatment (Figure 4—figure supplement 1B, C). This is consistent with the kinetics observed in Atg16l1-cKO BMDMs over the same time course (Figure 4B, TRIF; Figure 4C, RIPK1; Figure 4D, RIPK3). In contrast, basal turnover of monomeric TRIF, RIPK1 and RIPK3 was not significantly affected by lysosomal inhibition (Figure 4—figure supplement 2D-F). Proteasomal inhibition with MG132 provided modest and transient accumulation of TRIF or RIPK1 (Figure 4—figure supplement 2D, TRIF; Figure 4—figure supplement 2E, RIPK1), while basal RIPK3 levels were not affected (Figure 4—figure supplement 2F). Thus, we conclude that autophagy specifically regulates the lysosomal turnover of active forms of TRIF, RIPK1 and RIPK3. These new data are discussed in the subsection “Loss of autophagy results in accumulation of active forms of TRIF, RIPK1 and RIPK3 during necroptosis.

- As we highlighted before, the technique used to induce necroptosis is well established in the field. However, it is known that this method induces autophagy. Would it be possible to use another technique to induce necroptosis, without targeting autophagy?

This is challenging to address, since autophagy is an acknowledged stress response pathway induced under numerous inflammatory states. We have utilized multiple TLR and cytokine (TNF) ligands to induce necroptosis and arrived at a consistent conclusion that autophagy/ATG16L1 is cytoprotective under all states, thus we expect autophagy to be broadly induced during necroptotic innate immune response in macrophages. Our data support a framework where autophagic flux promotes the degradation of RHIM-domain proteins under multiple contexts, thus it is likely that most stimuli promoting the aggregation of TRIF, RIPK1 or RIPK3 depend on autophagy to curb necroptosis. We have revised our schematic to depict the role of autophagy in suppressing TLR and cytokine-mediated necroptosis (Figure 7—figure supplement 1B, C), and hope this provides a satisfactory description of the role of autophagy in necroptosis.

- Figure 3 shows accumulation of detergent-insoluble forms of TRIF, RIPK1 and RIPK3. Authors comment in the Discussion (second paragraph) about the possibility of protein ubiquitination playing a role in this process. We consider that detecting ubiquitination of these proteins will support their hypothesis considering that authors make a strong statement about the role of selective autophagy in this process.

Demonstrating ubiquitination of active forms of RIPK1, RIPK3 and TRIF in primary macrophages is a technically challenging request, given the requirement of highly denaturing lysis conditions. Nonetheless, we have attempted to measure ubiquitination of TRIF, RIPK1 and RIPK3 in primary BMDMs under these conditions. Using a mixture of M1 and K63-linkage-specific antibodies, we successfully immunoprecipitated autophosphorylated RIPK1 and RIPK3 from wild-type and Atg16l1-cKO BMDMs following LPS/zVAD treatment. As shown in Figure 4E, we demonstrate that loss of Atg16l1 results in accumulation of poly-ubiquitinated forms of autophosphorylated RIPK1 and RIPK3 upon LPS/zVAD treatment. This is entirely consistent with our previous observations that high molecular weight forms of autophosphorylated RIPK1 and RIPK3 accumulate in Atg16l1-cKO BMDMs (Figure 3D, F; Figure 4C, D).

Unfortunately, we have been unable to obtain publication-quality immunoblots demonstrating the accumulation of ubiquitinated TRIF in the same experimental setting. While we consistently observe high MW ubiquitinated TRIF in Atg16L1-cKO BMDM lysates, a clear demonstration is precluded by a considerable non-specific signal, perhaps from degradation of primary antibody (Author response image 1). Thus, we have focused our discussion on RIPK1 and RIPK3 ubiquitination in the revised manuscript. We hope this effort provides sufficient evidence to demonstrate that active RIPK1 and RIPK3 are indeed ubiquitinated during necroptosis, and that defective autophagy enhances the accumulation of ubiquitinated RIPK1 and RIPK3. These data are discussed in the second paragraph of the subsection “Loss of autophagy results in accumulation of active forms of TRIF, RIPK1 and RIPK3 during necroptosis”.

Author response image 1
TRIF, autophosphorylated RIPK1 and autophosphorylated RIPK3 are ubiquitinated during necroptosis.

Immunoblots of TRIF, autophosphorylated RIPK1, autophosphorylated RIPK3 and ubiquitin in BMDM lysates following immunoprecipitation of M1 or K63-ubiquitinated proteins after 4 hours of LPS/zVAD treatment. Red arrow depicts specific high MW TRIF signal.

https://doi.org/10.7554/eLife.44452.034

Reviewer #2:

This manuscript extends on observations reported by this laboratory in Samie et al., 2018, which reported Tax1BP1 mediated selective autophagy of TRIF. Here, the authors demonstrate that oligomerised RHIM domain proteins (including TRIF) accumulate in ATG16L1 deficient macrophages. They investigate autophagy as a possible homeostatic mechanism for clearing these oligomers, thereby opposing inflammation and cell death. Consistent with this, loss of essential autophagy genes resulted in enhanced cell death in BMDMs stimulated with LPS or Poly I:C and zVAD. A notable strength of this paper which distinguishes it from previous observations on RIP kinases and autophagy is that the authors provide evidence of involvement of the FIP200 and ATG14, which indicates this is due to canonical macroautophagy and not a 'non-canonical' or LAP-like process. The loss of Rubicon makes no difference, and whilst this does not rule out a contribution from 'non-canonical' autophagy (see discussion in Fletcher, Ulferts et al. EMBOJ 2018), taken together with the other genetic data it strongly argues against LAP being required here. The data to support this are shown in Figures 1F and Supplementary Figure 2B, the paper would be much easier to read if these data were grouped together.

These datasets are now grouped together as Figure 1D and E in the revised manuscript.

The authors demonstrate that Necrostatin fails to completely suppress necroptosis in autophagy deficient cells, and show this depends on Mlkl and Ripk3. ATG16L1 deficient cells accumulate detergent insoluble forms of RHIM domain proteins, consistent with a role for autophagy in their degradation. The authors go on to show that ZBP1 accumulates in ATG16L1 deficient cells, but that curiously this protects against necroptosis. This is interesting, but the effect only seems to take place in the setting of ATG16L1 deficiency so it remains unclear if there is a physiological relevance.

Absent from this paper are any direct measurements of autophagy or autophagic flux. It would be very helpful to the authors' argument to demonstrate that oligomerised RHIM-domain proteins do indeed become selectively engulfed by autophagosomes. As a minimum I would expect to see data on accumulation of autophagosomes and associated lipidation of LC3 after RHIM-domain oligomerisation, and (crucially) the effect of ATG14/FIP200/ATG16L1 deletion on this process. This would provide additional strong evidence of regulation of signalling by canonical autophagy.

We thank the reviewer for suggesting a more direct measurement of autophagic flux. While deletion of Atg16l1 revealed the accumulation of RHIM-domain proteins in macrophages, we acknowledge that it is not possible to study autophagic flux in this genotype. Thus, we treated wild-type BMDMs with the lysosomal inhibitor Bafilomycin A1 during a kinetic analysis of necroptosis. This is an established method to investigate the process of autophagic degradation of cytosolic cargo, as it blocks the lysosomal degradation of putative autophagy substrates, forcing their accumulation in the lysosomal compartment. Consistent with our observations in Atg16l1-cKO BMDMs, blocking lysosomal degradation in wild-type BMDMs resulted in the accumulation of TRIF, RIPK1 and RIPK3 in detergent-insoluble compartments over 6 hours of LPS/zVAD treatment (Figure 4—figure supplement 2A-C). Importantly, LC3B lipidation was rapidly induced (at approximately 30min) upon LPS/zVAD treatment, as observed by accumulation of LC3-II in Bafilomycin A1 treated BMDMs (Figure 4—figure supplement 2G). Finally, we measured the accumulation of multiple autophagy receptors under the same conditions and observed their accumulation in detergent-insoluble fractions (Figure 4—figure supplement 2H). Together, these data provide evidence that autophagic flux is induced during necroptosis, and that active forms of TRIF, RIPK1, RIPK3 are indeed trafficked to the lysosomal compartment for degradation. Comparing kinetics of RHIM-protein aggregation between wild-type and Atg16l1-cKO BMDMs reveals a consistent kinetic, with TRIF accumulation preceding that of autophosphorylated as well as total RIPK1 and RIPK3. In all these settings, loss of Atg16l1 enhanced the accumulation of RHIM-domain proteins (Figure 4B-D). Additionally, we were able to assay ubiquitinated forms of active (autophosphorylated) RIPK1 and RIPK3 and demonstrate increased accumulation in ATG16L1 deficient BMDMs during necroptosis (Figure 4E). These data are described in the second paragraph of the subsection “Loss of autophagy results in accumulation of active forms of TRIF, RIPK1 and RIPK3 during necroptosis”.

I accept the authors argument that a full mechanistic explanation of exactly how the proteins are ubiquitinated and degraded (assuming this is the case) is beyond the scope of the paper, but given their previous findings of Tax1BP1 being a critical component required for degradation of TRIF (at least in mice, the situation in humans may well be different), it is surprising that the authors do not comment at all on this. It would greatly strengthen the paper to see data on the requirement (or otherwise) of autophagy receptors such as TaxBP1, p62 and (in humans) NDP52.

We have successfully generated CRISPR-mediated knockdown of Sqstm1/p62, Tax1bp1 or Calcoco1 in wild-type BMDMs (Figure 5C). These autophagy receptors were previously identified as candidate receptors accumulating in Atg16L1-cKO BMDMs (Samie et al., 2018). Induction of LPS/zVAD or PolyI:C/zVAD-mediated necroptosis revealed that loss of Tax1bp1 significantly increased BMDM necroptosis (Figure 5D). Consistent with the phenotypes of BMDMs lacking ATG16L1, ATG14L or FIP200, TAX1BP1 deficiency resulted in Necrostatin-1 resistant cell death following activation of TRIF signaling via TLR3 or TLR4 (Figure 5D, PolyI:C+zVAD+Nec-1 or LPS+zVAD+Nec-1 conditions). These data are described in the subsection “The autophagy receptor TAX1BP1 prevents TRIF-mediated necroptosis”.

In summary this paper provides an important if somewhat limited advance in our understanding of the regulation of necroptosis. The conclusions are in general well supported by the data. The paper would be greatly strengthened by exploring the mechanisms of selective autophagy in this setting.

Additional data files and statistical comments:

Many of the pairwise analyses throughout require correction for multiple hypotheses.

We have updated all pairwise analyses to correct for multiple hypotheses (Holms-Sidak method, α=0.05). This has not affected our conclusions.

Reviewer #3:

[…] Overall the paper is well written and the data very compelling. However some additional insights on how ZBP1 can repress necroptosis would strengthen this study.

– The authors should provide some insights into how ZBP1 functions in repressing necroptosis.

This is a highly relevant comment but broad in its implications. Studies to address the molecular mechanism(s) by which ZBP1 functions in repressing necroptosis requires additional tools (e.g. domain-specific ZBP1 mutants) for primary macrophage complementation assays. Specifically, RHIM-mutant vs. Za-domain mutant versions of ZBP1 would need to be generated and knock-in cells obtained both in wild-type and autophagy-deficient backgrounds. These requirements place a thorough assessment of ZBP1 function outside the scope of the current manuscript. However, we have highlighted this possibility in the Discussion to encourage future investigation on the molecular mechanism by which ZBP1 represses TRIF-mediated necroptosis (Discussion, first paragraph).

– Ripk1 inhibition could not completely block necroptosis when TRIF is involved (TLR3 and TLR4 stimuli in autophagy deficient cells). Can RIPK3 be recruited to the TRIFosome and activated in the absence of RIPK1? Is it just the kinase activity of RIPK1 that becomes redundant (i.e. is scaffolding still required)? Use of RIPK1 sgRNA could address this.

Since RIPK1 deletion significantly compromises hematopoietic cell and macrophage fitness (Rickard, J.A., et al. Cell 2014; Roderick, J.E. et al. PNAS 2014; Newton, et al., 2016), assessment of TRIFfosome formation in RIPK1-deficient BMDMs requires the generation of Atg16l1-cKO; Ripk1-KO; Caspase-8-KO; Ticam1-KO quadruple knockout or cells. Additional deletion of Mlkl may be needed to maintain viability in the absence of Ripk1 or Caspase-8 if TLR3 or TLR4-mediated pathways are investigated. This complicated genetic background poses a significant technical challenge and is out of scope for the current study.

Previous work by the Mockarski lab (Kaiser et al., 2013) and Wang lab (He, S. et al. PNAS 2011) have shown that TRIF activation directly stimulates RIPK3 auto-phosphorylation and downstream necroptosis via MLKL. Kaiser et al. demonstrate that the TRIF/RIPK3 pathway is activated in either the absence of RIPK1 protein or its kinase activity, thus revealing a RIPK1-independent pathway of TRIF-mediated necroptosis. Our observations are consistent with these previously described mechanisms; defective autophagy likely enhances TRIF/RIPK3-mediated necroptosis in the absence of RIPK1. We raise this possibility in the Discussion (first paragraph) and have updated the references to cite the above works in the revised manuscript.

– How is necroptosis triggered in non-TRIF mediated TLR triggered necroptosis? Through autocrine secretion of TNF or IFN?

To address this query, we blocked TNF- or IFN-signaling by pre-treating BMDMs with TNFR2-Fc or anti-IFNAR1, respectively. Consistent with previous studies, we observed that non-TRIF mediated necroptosis of wild-type BMDMs was rescued by blockade of TNF or IFNAR1 (Figure 2—figure supplement 2A, E, F). However, this failed to rescue necroptosis of Atg16l1-cKO BMDMs to the same level as Necrostatin-1 (Figure 2—figure supplement 2A, D, E). These findings reveal that RIPK1 activity dominates the necroptotic phenotype when autophagy is defective and non-TRIF mediated TLR signaling engaged. This is likely due to additional death ligands contributing to RIPK1 activity. We discuss these new findings in the second paragraph of the subsection “TRIF and RIPK1 drive necroptosis in Atg16l1-deficient macrophages” and provide new data (Figure 2—figure supplement 2A-F) in the revised manuscript.

– The recent paper (Sarhan et al., 2018) suggests that autocrine IFN is required for necroptosis. Is IFN involved in necroptosis when autophagy is not functional? Does perturbed clearance of TRIF and RIPK1 result in higher IFN signaling and thus higher cell death?

We thank the reviewer for this query. To assay IFN signaling, we measured STAT1 phosphorylation in wild-type and Atg16l1-cKO BMDMs over a 6-hour time-course of necroptosis induced by LPS/zVAD. Loss of Atg16l1 enhanced STAT1 phosphorylation, both in magnitude and kinetics (Figure 2—figure supplement 2G). Pharmacological blockade of IFNAR1 rescued necroptosis of wild-type BMDMs, consistent with Sarhan et al., 2018. Interestingly, it also decreased necroptosis of Atg16l1-cKO BMDMs comparably to Necrostatin-1, indicating that elevated IFN signaling contributes to elevated cell death (Figure 2—figure supplement 2B, C). These data provide an additional insight into autocrine signaling that precipitates inflammation when autophagy is perturbed. We present and discuss these new findings in the third paragraph of the subsection “TRIF and RIPK1 drive necroptosis in Atg16l1-deficient macrophages”.

– Figure 3D and F suggests that when autophagy is deranged, RIPK1- and RIPK3- oligomerization is dependent on TRIF but not phosphorylation of RIPK1 and RIPK3?

When autophagy is defective (e.g. Atg16l1 deletion), TRIF-mediated TLR stimulation elevates autophosphorylation as well as oligomerization of RIPK1 and RIPK3. Knockdown of TRIF in ATG16L1-deficient BMDMs decreased autophosphorylation of both high MW and monomeric RIPK1 (Figure 3D). Levels of autophosphorylated, high MW RIPK3 were more significantly affected, with a considerable loss of this form of RIPK3 when TRIF is deleted. These findings are consistent with the previously discussed observations that TRIF can directly engage RIPK3 and drive its signaling (He, S. 2011; Kaiser, 2013). We discuss direct RIPK3 engagement by TRIF and the impact of defective autophagy in the first paragraph of the Discussion.

– The term "screening" is a bit misleading in Figure 1. They optimized electroporation conditions for delivery of sgRNA:Cas9 RNPs into primary BMDMs. This could be added to the supplementary figure.

We have replaced “screening” with “comparing” to address this query. We have also moved the schematic to Figure 1—figure supplement 2A, B in the revised manuscript.

https://doi.org/10.7554/eLife.44452.036

Article and author information

Author details

  1. Junghyun Lim

    Department of Cancer Immunology, Genentech, South San Francisco, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Validation, Investigation, Methodology, Writing—review and editing
    Competing interests
    Affiliated with Genentech. No other competing interests to declare
  2. Hyunjoo Park

    Department of Translational Immunology, Genentech, South San Francisco, United States
    Contribution
    Investigation
    Competing interests
    Affiliated with Genentech. No other competing interests to declare
  3. Jason Heisler

    Department of Translational Immunology, Genentech, South San Francisco, United States
    Contribution
    Investigation
    Competing interests
    Affiliated with Genentech. No other competing interests to declare
  4. Timurs Maculins

    Department of Cancer Immunology, Genentech, South San Francisco, United States
    Contribution
    Validation, Investigation, Writing—review and editing
    Competing interests
    Affiliated with Genentech. No other competing interests to declare
  5. Merone Roose-Girma

    Department of Molecular Biology, Genentech, South San Francisco, United States
    Contribution
    Resources
    Competing interests
    Affiliated with Genentech. No other competing interests to declare
  6. Min Xu

    Department of Translational Immunology, Genentech, South San Francisco, United States
    Contribution
    Resources, Supervision, Methodology, Project administration
    Competing interests
    Affiliated with Genentech. No other competing interests to declare
  7. Brent Mckenzie

    Department of Translational Immunology, Genentech, South San Francisco, United States
    Contribution
    Resources, Supervision, Project administration
    Competing interests
    Affiliated with Genentech. No other competing interests to declare
  8. Menno van Lookeren Campagne

    Department of Immunology, Genentech, South San Francisco, United States
    Contribution
    Supervision, Writing—original draft, Writing—review and editing
    Competing interests
    Affiliated with Genentech. No other competing interests to declare
  9. Kim Newton

    Department of Physiological Chemistry, Genentech, South San Francisco, United States
    Contribution
    Resources, Supervision, Methodology, Writing—original draft, Project administration, Writing—review and editing
    Competing interests
    Affiliated with Genentech. No other competing interests to declare
  10. Aditya Murthy

    Department of Cancer Immunology, Genentech, South San Francisco, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Supervision, Investigation, Visualization, Methodology, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    murthy.aditya@gene.com
    Competing interests
    Affiliated with Genentech. No other competing interests to declare
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6130-9568

Funding

Genentech

  • Aditya Murthy

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank K Cherry, T Scholl, B Torres and W Ortiz for animal husbandry, K Wickliffe, K Rajasekaran, K Heger and E Freund for technical assistance, and T Yi, M Albert, S Turley and I Mellman for critical review of this work.

Ethics

Animal experimentation: All protocols were approved by the Genentech Institutional Animal Care and use Committee; all studies were executed by following relevant ethical regulations detailed in animal use protocols.(internal protocol 18-1823).

Senior Editor

  1. Tadatsugu Taniguchi, Institute of Industrial Science, The University of Tokyo, Japan

Reviewing Editor

  1. Facundo D Batista, Ragon Institute of MGH, MIT and Harvard, United States

Reviewers

  1. Rupert Beale
  2. Katherine Fitzgerald, UMASS, United States

Publication history

  1. Received: December 17, 2018
  2. Accepted: June 14, 2019
  3. Version of Record published: July 9, 2019 (version 1)

Copyright

© 2019, Lim et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 626
    Page views
  • 135
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Cell Biology
    Naama Flint Brodsly et al.
    Research Article
    1. Biochemistry and Chemical Biology
    2. Cell Biology
    Amal Alex et al.
    Tools and Resources