The Protein ATG16L1 Suppresses Inflammatory Cytokines

Immunity Article

The Protein ATG16L1 Suppresses Inﬂammatory Cytokines Induced by the Intracellular Sensors Nod1 and Nod2 in an Autophagy-Independent Manner

Matthew T. Sorbara,1 Lisa K. Ellison,1 Mahendrasingh Ramjeet,2,6 Leonardo H. Travassos,1,5 Nicola L. Jones,3 Stephen E. Girardin,2,4 and Dana J. Philpott1,4,* 1Department of Immunology, University of Toronto, Toronto M5S1A8, Canada 2Department of Laboratory Medicine and Pathobiology, University of Toronto, Toronto M5S1A8, Canada 3Cell Biology Program, Research Institute, The Hospital for Sick Children and the Department of Physiology, University of Toronto, Toronto M5S1A8, Canada 4These authors contributed equally to this work 5Present address: Laborato´ rio de Fisiologia da Respirac¸a˜ o, Instituto de Biofı´sica Carlos Chagas Filho, Rio de Janeiro CEP 21941-902, Brazil 6Present address: bioMe´ rieux/R&D Microbio-Innovation, 3 route de Port Michaud, 38390 La Balme-les-Grottes, France *Correspondence: [email protected] http://dx.doi.org/10.1016/j.immuni.2013.10.013

SUMMARY ually as well as in concert to promote balanced inflammatory responses at the level of the gut mucosa remains unresolved. The peptidoglycan sensor Nod2 and the autophagy Nod1 and Nod2 are members of the Nod-like receptor (NLR) protein ATG16L1 have been linked to Crohn’s dis- family of intracellular pattern-recognition molecules (PRMs) ease (CD). Although Nod2 and the related sensor, that recognize fragments of bacterial peptidoglycan. Nod1 rec- Nod1, direct ATG16L1 to initiate anti-bacterial auto- ognizes D-glutamyl-meso-diaminopimelic acid, a peptido- phagy, whether ATG16L1 affects Nod-driven inflam- glycan-derived dipeptide found primarily in Gram-negative mation has not been examined. Here, we uncover bacteria as well as select species of Gram-positive bacteria, including Listeria monocytogenes (Chamaillard et al., 2003; an unanticipated autophagy-independent role for Girardin et al., 2003a). Nod2 recognizes muramyl dipeptide ATG16L1 in negatively regulating Nod-driven inflam- (MDP), which is present in both Gram-positive and Gram- matory responses. Knockdown of ATG16L1 expres- negative bacteria (Girardin et al., 2003b; Inohara et al., 2003). sion, but not that of ATG5 or ATG9a, specifically Recruitment of Nod1 and Nod2 to the plasma membrane has enhanced Nod-driven cytokine production. In addi- been reported, and these PRMs are thought to primarily detect tion, autophagy-incompetent truncated forms of intracellular bacteria during the invasion process (Philpott and ATG16L1 regulated Nod-driven cytokine responses. Girardin, 2010). Indeed, Nod1 and Nod2 are critical in initiating Mechanistically, we demonstrated that ATG16L1 innate immune responses to Shigella flexneri (Shigella), Listeria interfered with poly-ubiquitination of the Rip2 monocytogenes (Listeria), and Salmonella enterica serovar adaptor and recruitment of Rip2 into large signaling Typhimurium (Salmonella) (Sorbara and Philpott, 2011). complexes. The CD-associated allele of ATG16L1 Recognition of peptidoglycan motifs by Nod1 or Nod2 in the cytosol leads to the recruitment of the adaptor protein Rip2 to was impaired in its ability to regulate Nod-driven a complex of oligomerized Nod proteins, and this recruitment inflammatory responses. Overall, these results then drives NF-kB- and MAPK-dependent inflammatory re- suggest that ATG16L1 is critical for Nod-dependent sponses (Inohara et al., 2000; Ogura et al., 2001b). Activation regulation of cytokine responses and that disruption of Rip2 is critical for the recruitment of the IKK complex and other of this Nod1- or Nod2-ATG16L1 signaling axis could downstream components of the Nod signaling pathway, and this contribute to the chronic inflammation associated recruitment is controlled by the addition of K63-linked polyubi- with CD. quitin on Rip2. Several E3 ligases, including XIAP, cIAP1, cIAP2, ITCH, TRAF6, and TRAF2 and TRAF5, have been pro- posed to drive the ubiquitination of Rip2 (Sorbara and Philpott, INTRODUCTION 2011). ATG16L1 is a key component of the autophagy machinery. Crohn’s disease (CD) is characterized by dyseregulated mucosal Autophagy is a bulk degradation system in which cytoplasmic immune responses in the intestine; these responses lead to cargo is sequestered in double-membrane autophagosomes aberrant inflammation that is most likely directed toward innoc- for subsequent degradation in lysosomes. ATG16L1 functions uous components of the microbiota. Several genes involved in in autophagy through its interaction with an ATG5-ATG12 conju- the innate immune response to pathogens, including NOD2 gate to direct the site of autophagosome formation (Fujita et al., and ATG16L1, have been identified as susceptibility loci for the 2008; Fujita et al., 2009; Mizushima et al., 2003). Recruitment of development of CD (Hampe et al., 2007; Hugot et al., 2001; ATG16L1:ATG5-ATG12 promotes the localized conversion of Ogura et al., 2001a). However, how these genes function individ- LC3 to a phosphotidyl-ethanolamine-conjugated form, LC3-II,

858 Immunity 39, 858–873, November 14, 2013 ª2013 Elsevier Inc. Immunity ATG16L1 Negatively Regulates Nod1 and Nod2

and thus drives autophagosome formation. A role for autophagy compared to Atg5/ MEFs, ATG16L1 DCCD MEFs infected in restricting the growth of invasive bacteria has been described. with Shigella unexpectedly produced reduced amounts of Infection with invasive pathogens such as Shigella, Listeria, or CXCL1 (Figure 1B). In the case of Salmonella and Listeria, Salmonella triggers an amino acid starvation stress response although these pathogens did not trigger increased secretion (Tattoli et al., 2012b; Tattoli et al., 2013) that induces autophagy of CXCL1 in Atg5/ cells as compared to WT cells, the cytokine as a host-mediated bactericidal mechanism (Birmingham et al., response observed in ATG16 DCCD MEFs was significantly 2007; Birmingham et al., 2006; Ogawa et al., 2005). blunted in comparison to the response in either WT or Atg5/ A critical role for Nod1 and Nod2 in the initiation and targeting cells. Together, these results reveal an unanticipated difference of anti-bacterial autophagy has been identified (Cooney et al., between Atg5/ and ATG16L1 DCCD cells with regard to the 2010; Homer et al., 2010; Travassos et al., 2010). Stimulation response to invasive bacterial pathogens and imply that a of Nod1 or Nod2 induces autophagosome formation, and both macroautophagy-independent function of ATG16L1 contributes Nod1 and Nod2 interact with ATG16L1 to target the autophagy to the cytokine response to these pathogens. machinery to the plasma membrane at the site of bacterial Although autophagy deficient, the ATG16L1 DCCD MEFs do invasion (Travassos et al., 2010). In addition to functioning as a express a truncated form of ATG16L1 lacking amino acids bactericidal mechanism, Nod-induced autophagy also pro- 70–212 (Figures 1C and 1D), and we thus hypothesized that motes antigen presentation on MHC class II molecules (Cooney the expression of this truncated ATG16L1 protein might be the et al., 2010). However, the impact of ATG16L1 interaction with cause of the differential cytokine response observed in Atg5/ Nod1 and Nod2 on canonical Nod signaling and Nod-driven and ATG16L1 DCCD MEFs. In order to examine this possibility, inflammation during infection has not been elucidated. we knocked down expression of the DCCD fragment in the auto- Here we have characterized the functional interaction between phagy-deficient MEFs by using shRNA directed against a region ATG16L1 and Nod proteins and identified a previously unappre- in the remaining fragment of ATG16L1, and we confirmed the ciated role of ATG16L1 in specifically regulating Nod1 and Nod2 efficiency of the knockdown at the protein and RNA levels (Fig- signaling. We demonstrate that ATG16L1 negatively regulates ures 1E and 1F). Knockdown of the DCCD fragment in Nod1- and Nod2-driven inflammation independently of its ATG16L1 DCCD MEFs resulted in an increased secretion of canonical function in autophagosome formation. Moreover, CXCL1 in response to Listeria and Salmonella and a trend toward this Nod regulatory function of ATG16L1 was specific to this an increased response to Shigella (Figure 1G). Thus, a domain autophagy protein given that autophagy interference by of ATG16L1, distinct from the coiled-coil region implicated silencing ATG5 or ATG9a or by chemical inhibition did not regu- in autophagy, was responsible for controlling the cytokine late Nod responses. The mechanism of Nod regulation by response to bacterial pathogens. ATG16L1 was the blocking of Rip2 activation, as marked by reduced polyubiquitination of Rip2 and diminished incorporation ATG16L1, but not ATG5, Regulates the Cytokine into Nod signaling complexes. Finally, cells expressing the CD Response to Invasive Bacterial Pathogens in Epithelial risk allele of ATG16L1 had defective regulation of Nod-driven Cells inflammation. Altogether, these findings extend the role of the Shigella, Listeria, and Salmonella are enteric pathogens that, ATG16L1-Nod1 and -Nod2 axis beyond the targeting of bacteria during infection in vivo, are able to invade intestinal epthelial cells to the autophagy system and demonstrate that ATG16L1 has a and replicate intracellularly. On the basis of our findings in fibro- critical function in the control of Nod-driven inflammatory re- blasts, we sought to establish a model system to investigate the sponses. Disruption of this regulation by ATG16L1 could cause respective roles of ATG5 and ATG16L1 in intestinal epithelial a predisposition to aberrant inflammation, as observed in CD. cells during bacterial infection. We took the approach of transiently knocking down these transcripts in ModeK mouse intes- RESULTS tinal epithelial cells to minimize the accumulated effects from the prolonged loss of autophagy on essential cellular metabolic pro- Expression of an Autophagy-Deficient Truncation cesses. Knockdown of ATG5 or ATG16L1 was confirmed at the of ATG16L1 Regulates Cytokine Responses protein and mRNA levels (Figures S1A, S1B, and S1C in the Sup- We aimed to determine whether the proteins of the autophagic plemental Information available with this article online). As ex- systems contributed to the host response to bacterial pathogens pected, ATG5 or ATG16L1 deficiency led to impaired conversion independently of their ability to induce autophagy. In Atg5/ of endogenous LC3-I to LC3-II after rapamycin treatment (Fig- and ATG16L1 DCCD murine embryonic fibroblasts (MEFs), ure S1E), and defective targeting of intracellular Shigella to which are defective in autophagy induction (Fujita et al., 2009; GFP-LC3+ autophagosomes (Figures S1G and S1H). This Ogawa et al., 2005), rapamycin-induced LC3-II generation was demonstrated that the knockdown procedure not only reduced undetectable, as expected (Figure 1A). In addition, the amounts amounts of ATG5 or ATG16L1 but also effectively blocked auto- of p62 were increased in the autophagy-deficient cell lines (Fig- phagy induction in ModeK cells. In contrast to the chronic auto- ure 1A), in line with the previous evidence that this protein is nor- phagy-deficient conditions in the autophagy-deficient MEFs mally constitutively degraded by autophagy (Lee et al., 2012). (Figure 1A), we did not observe a change in p62 amounts by pro- Next we infected the control and autophagy-deficient MEFs tein blot after transient knockdown (data not shown), thus with the invasive bacteria Shigella, Listeria, and Salmonella.As excluding reported confounding effects of changes in p62 previously reported, Atg5/ MEFs produced increased amounts on inflammatory responses. Knockdown of ATG16L1 amounts of the proinflammatory cytokine CXCL1 after Shigella but not ATG5 triggered a significant upregulation of CXCL1 infection (Figure 1B) (Dupont et al., 2009). However, as secretion after infection with Shigella (Figure 2A), Listeria

Immunity 39, 858–873, November 14, 2013 ª2013 Elsevier Inc. 859 Immunity ATG16L1 Negatively Regulates Nod1 and Nod2

Figure 1. An Autophagy-Deﬁcient Fragment of ATG16L1 Regulates the Cytokine Response to Bacterial Infection in MEFs (A) Wild-type, Atg5/, or ATG16L1DCCD MEFs stimulated with rapamycin (25 mg/ml) for 2 hr and amounts of LC3, p62, and tubulin were analyzed by protein blot. (B) MEFs were infected with Shigella, Listeria,or Salmonella, and CXCL1 production was measured 16 hr after infection (n = 3). (C) Schematic depiction of the truncated ATG16L1 expressed in DCCD MEFs. (D) Expression of the DCCD fragment was conﬁrmed by protein blot. (E and F) Knockdown of the DCCD fragment in ATG16L1 DCCD MEFs transduced with ATG16L1 targeting shRNA was examined by protein blot (E) or qPCR (F, n = 6). (G) Control or DCCD knockdown ATG16L1DCCD MEFs were infected with Shigella, Listeria,or Salmonella, and CXCL1 was measured 16 hr after infection (n = 3, from a representative experiment). Protein blots in (A), (D), and (E) are representative of three independent experiments. Values shown are means + SEM. *p < 0.05, **p < 0.01 by one-way ANOVA and Tukey post-test (B) or t test (F and G).

experiencing a loss of ATG16L1 showed an increased IL-8 response to invasive Shigella infection (Figure 2G), indicating that human ATG16L1 also suppresses cytokine responses after bacterial infection. Thus, in line with the ﬁndings in the autophagy-deﬁcient DCCD MEFs, ATG16L1 has a regulatory role in bacteria-infected murine and human epithelial cells, and this role is independent from its function in macroautophagy.

ATG16L1 Regulates the Response to Cytosol- Exposed Pathogens We next sought to determine whether ATG16L1 was a general negative regulator of the response to bacterial infection or whether it regulated signaling within specific cellular compartments. Both Shigella and Listeria induce uptake into epithelial cells and subsequently escape from the entry endosome into the cytosol. To (Figure 2B), or Salmonella (Figure 2C), in line with the results determine whether cytosol-exposure was necessary for the obtained in MEFs (see Figure 1). As a control, we also observed increased cytokine response to Shigella or Listeria in that blocking autophagosome maturation with chloroquine ATG16L1-deficient ModeK cells, we performed infections with in ModeK cells did not increase cytokine responses to Shigella noninvasive Shigella or endosome-confined Listeria. Whereas infection, despite chloroquine-dependent accumulation of ATG16L1 expression regulated inflammatory responses to LC3-II (Figures 2D–2F). cytoinvasive strains, it had no impact on cytokine responses We also knocked down expression of ATG16L1 in human to noninvasive Shigella (Figure 3A) or the endosome-confined epithelial MDAMC cells and confirmed at the protein and RNA Listeria strain (Figure 3B) at either 8 or 16 hr after infection. level that ATG16L1 expression was silenced (Figures S1I Indeed, significant increases in the production of CXCL1 with and S1J). Similar to mouse epithelial cells, MDAMC cells cytoinvasion of Shigella or Listeria as compared to noninvasive

860 Immunity 39, 858–873, November 14, 2013 ª2013 Elsevier Inc. Immunity ATG16L1 Negatively Regulates Nod1 and Nod2

Figure 2. ATG16L1 Negatively Regulates the Epithelial Cell Response to Invasive Bacteria in an Autophagy-Independent Manner (A–C) Production of CXCL1 by ModeK cells transduced with the indicated shRNAs was measured 16 hr after infection with Shigella (A), Listeria (B), or Salmonella (C) (n = 3, values are expressed as an n-fold change relative to uninfected controls). (D) ModeK cells were treated with chloroquine (10 mM) and rapamycin (25 mg/ml) as indicated, and formation of LC3-II was monitored by protein blot. (E and F) Control or chloroquine-treated cells were infected with invasive Shigella, and production of CXCL1 (E) or CXCL2 mRNAs (F) was measured by qPCR (n = 4). (G) Control and ATG16L1 knockdown MDAMC cells were infected with Shigella for 16 hr, and secretion of IL-8 was measured by ELISA (n = 8). Values shown are means + SEM. *p < 0.05, **p < 0.01 by one-way ANOVA and Tukey post-test (A–C) or t test (G). See also Figure S1.

with Shigella and Listeria, ATG16L1 also regulated the response to Salmonella at 8hr(Figure 3C) in addition to 16 hr after infection (Figure 2C). These results suggest that the cytosolic exposure that occurs during the remodeling of the SCV by Salmonella is sufficient to trigger an ATG16L1-regulated cytokine response pathway. To directly test this hypothesis, we utilized DSPI-1/pInv, a SPI-1-deficient Salmonella strain expressing the Invasin protein from Yersinia. This strain is able to invade epithelial cells but does not trigger the same membrane damage, cytosolic exposure, or autophagy targeting as wild-type Salmonella (Birmingham et al., 2006; Tattoli et al., 2012a). In contrast to infection with wild-type Salmonella, infection of ATG16L1 knockdown cells with the DSPI-1/pInv strain did not lead to an enhanced cytokine response (Figure 3D). Together, these results demonstrate that ATG16L1 nega- or endosome-confined strains were observed in ATG16L1- tively regulates cytokine responses to invasive bacteria that knockdown cells but not in control or ATG5 knockdown cells are presented to the host cytosol. (Figure S2A). We next aimed to investigate whether ATG16L1-dependent In contrast to the cytosolic localization of Shigella and Listeria, blunting of cytokine secretion during cytoinvasive bacterial Salmonella replicates within a modified vacuole, termed the infection occurred at the transcriptional or posttranscriptional Salmonella-containing vacuole (SCV). Despite its vacuolar repli- level. We observed that knockdown of ATG16L1 led to cation site, early bacterial modification of the maturing endo- enhanced induction of CXCL1 and CXCL2 mRNA after infec- some and insertion of type III secretion systems into the vacuole tion with invasive but not noninvasive Shigella (Figure 3E), membrane result in partial cytosolic exposure of Salmonella or with wild-type but not endosome-confined Listeria (Figure 3F), Salmonella-derived molecules early during infection, leading to and with Salmonella (Figure 3G). Thus, ATG16L1 regulates the targeting by the autophagy machinery and activation of intra- cytokine responses to intracellular bacteria at the transcrip- cellular PRRs at early time points (Birmingham et al., 2006). As tional level.

Immunity 39, 858–873, November 14, 2013 ª2013 Elsevier Inc. 861 Immunity ATG16L1 Negatively Regulates Nod1 and Nod2

(legend on next page) 862 Immunity 39, 858–873, November 14, 2013 ª2013 Elsevier Inc. Immunity ATG16L1 Negatively Regulates Nod1 and Nod2

In contrast to CXCL1 and CXCL2, which can be activated In epithelial cells, we similarly observed that the knockdown of directly by PRR activation, ATF3 is a key stress-response- ATG16L1 in human MDAMC cells resulted in specific upregula- pathway component that can be induced by membrane damage tion of the response to Nod1 ligand but not to LPS or TNF-a (Fig- and which probably does not require PRR activation for its induc- ure 4C). In ModeK cells, which contrary to MEFs or MDAMC cells tion (Tattoli et al., 2012b). In order to determine whether express both Nod1 and Nod2 and respond to the Nod2 ligand ATG16L1 regulated all pathways downstream of membrane L18-MDP, we similarly observed that the knockdown of damage and cytosol exposure, we examined ATF3 induction in ATG16L1 resulted in specific upregulation of the response to ATG16L1-deficient ModeK cells (Figure 3H). As expected, infec- both Nod ligands but not to LPS or TNF-a (Figure 4D). As a tion with cytoinvasive strains of Shigella, Listeria, and Salmonella control, we also aimed to confirm that the effect of ATG16L1 triggered ATF3 activation 3 hr after infection; however, unlike the on Nod-driven signaling was independent of autophagy in proinflammatory cytokine response, silencing of ATG16L1 ModeK epithelial cells. Silencing of ATG5 did not alter responses expression did not result in increased ATF3 activation. Thus, to direct Nod stimulation (Figure S3A). Indeed, ATG5 knockdown ATG16L1 specifically suppressed transcriptional pathways lead- cells had a similar ligand response profile as control cells (Fig- ing to cytokine production after pathogen entry into the cytosol. ure S3B). Importantly, ATG16L1 knockdown cells had similar In addition, the increased cytokine responses in ATG16L1 CXCL1 production after TNF-a or LPS stimulation but had a knockdown cells did not depend on increased bacterial replica- greater response to Nod1 or Nod2 stimulation compared than tion; increased responses were still observed when replication of did ATG5 knockdown cells (Figure S3B). Similar to that of Shigella was blocked with chloramphenicol (Figures S2B and ATG5, silencing of ATG9a (Figures S1A and S1D) did not trigger S2C) or when a DphoP Salmonella mutant that invades epithelial increased responses to Nod1 or Nod2 stimulation (Figure S3C), cells but is attenuated in the intracellular environment was used despite functional inhibition of autophagy (Figure S1F). Finally, (Figure S2D). treatment with the autophagy inhibitor 3-MA did not alter cytokine production after Nod1 or Nod2 stimulation (Figure S3D). ATG16L1 Negatively Regulates Nod1- and Nod2-Driven To confirm that the changes in Nod responses were not due to Cytokine Responses off-target effects of the ATG16L1 murine or human shRNAs, we We next aimed to identify the mechanism through which generated additional shRNAs targeting murine ATG16L1 and ATG16L1 negatively regulates cytokine responses to invasive tested them in ModeK cells. Three of the shRNAs silenced bacterial pathogens. ATG16L1 interacts with Nod1 and Nod2 ATG16L1 expression at the protein and RNA levels (Figures to target the autophagic response to invading bacteria (Travas- S3E and S3F), whereas one did not successfully knock down sos et al., 2010). We speculated that this interaction could ATG16L1. In agreement, all three shRNAs that silenced also be responsible for the autophagy-independent modulation ATG16L1 resulted in increased Nod2 but not LPS responses in that we observed for cytokine responses in MEFs and epithelial comparison to those in control cells, whereas the inefficient cells. The truncated ATG16L1 found in autophagy-deficient shRNA failed to alter Nod2 responses (Figure S3G). ATG16L1 DCCD MEFs bound to Nod1 (Figure 4A left panels) Thus, ATG16L1 specifically inhibits Nod-dependent re- and Nod2 (Figure 4A right panels) with a similar efficiency as sponses in MEFs and epithelial cells. full-length ATG16L1. Furthermore, Nod1 and Nod2 also bound to the ATG16L1’s C-terminal region (amino acids 230–607) that ATG16L1 Modulates the Cytokine Response to Bacteria does not interact with ATG5 (data not shown), suggesting that through Nod Signaling the interaction occurred through the ATG16L1 WD40 repeat Infection with Shigella, Listeria,orSalmonella results in the region. activation of multiple PRRs, which could contribute to the In order to show that ATG16L1 limits Nod-dependent increased cytokine response in ATG16L1-deficient cells. We responses, we first stimulated ATG16L1 DCCD MEFs with the next determined whether ATG16L1 modulated host responses Nod1 ligand C12-iE-DAP. Knocking down the DCCD fragment to bacterial pathogens solely through the Nod1- or Nod2-Rip2 of ATG16L1 resulted in an increased cytokine response to the signaling axis. To do so, we knocked down ATG16L1 in primary Nod1 agonist but not LPS (Figure 4B), thus highlighting the MEFs from WT and Ripk2/ mice (Figure 4D). Although reduc- specificity of ATG16L1-dependent regulation of inflammatory tion of ATG16L1 expression led to an increased CXCL2 response signaling. Because the ATG16L1 DCCD MEFs are autophagy to invasive Shigella in WT MEFs (Figure 4E), this increase was deficient, these results also imply that ATG16L1 affects fully Rip2 dependent given that ATG16L1 silencing even Nod1-dependent signaling through autophagy-independent appeared to downregulate CXCL2 in infected Ripk2/ MEFs. pathways. Thus, ATG16L1-dependent blunting of proinflammatory

Figure 3. ATG16L1 Suppresses the Cytokine Response to Cytoinvasive Bacteria (A–H) White bars indicate scrambled shRNA cells, and black bars indicate responses in ATG16L1 knockdown cells. Control or ATG16L1 knockdown ModeK cells were infected with (A) invasive or noninvasive Shigella and with (B) Listeria or the endosome-conﬁned mutant, and CXCL1 was measured at 8 and 16 hr after infection. (C) At 8 hr after Salmonella infection, CXCL1 was measured in control and ATG16L1 knockdown cells. (A–C, n = 4). (D) Control or ATG16L1 knockdown ModeK cells were infected with wild-type Salmonella or the DSPI-1/pInv strain, and CXCL1 was measured 16 hr after infection (n = 9). (E–G) CXCL1 and CXCL2 mRNA was measured 3 hr after infection with invasive or noninvasive Shigella (E), wild-type or endosome-conﬁned Listeria (F), or Salmonella (G) (E–G, n = 6). (H) Control or ATG16L1 silenced ModeK cells were infected with the indicated strains, and production of ATF3 mRNA was analyzed by qPCR at 3 hr after infection (n = 4). All values are expressed as an n-fold change relative to each cell-type baseline and are shown as means + SEM. *p < 0.05, **p < 0.01 by t test. See also Figure S2.

Immunity 39, 858–873, November 14, 2013 ª2013 Elsevier Inc. 863 Immunity ATG16L1 Negatively Regulates Nod1 and Nod2

Figure 4. ATG16L1 Negatively Regulates Nod1- and Nod2-Driven Cytokine Responses (A) HEK293T cells were transfected with HA-Nod1 (left) or HA-Nod2 (right) and Flag-mCherry, Flag-ATG16L1, or Flag-DCCD. Interaction between the Flag-tagged proteins and Nod1 or Nod2 was assessed by immunoprecipitation. Input lysates and immunoprecipitated proteins were probed for Flag and HA (a representative of two independent experiments is shown). (legend continued on next page) 864 Immunity 39, 858–873, November 14, 2013 ª2013 Elsevier Inc. Immunity ATG16L1 Negatively Regulates Nod1 and Nod2

signaling in response to cyto-invasive bacteria occurred through 2011). HEK293T cells transiently expressing Nod1 were trans- the control of Nod-driven pathways. fected with ATG16L1 or a control vector, and cell lysates were In intestinal epithelial cells, we used a specific Rip2 inhibitor to separated by sucrose gradient ultracentrifugation after Nod1 determine whether ATG16L1 restricted Rip2-dependent stimulation. As expected, stimulation of control cells with Nod1 signaling during infection. As expected, treatment with 1 mMof ligand led to a shift of Rip2 signal into more dense fractions, the Rip2 inhibitor efficiently blocked CXCL1 production induced indicative of the incorporation of Rip2 into larger complexes (Fig- by the Nod ligands C12-iE-DAP and L18-MDP without blocking ure 5C). Interestingly, overexpression of ATG16L1 resulted in a the response to TNF-a or LPS (Figure S3H). In addition, treat- shift of Rip2 toward smaller fractions before and after Nod1 stim- ment with the Rip2 inhibitor did not block Shigella invasion or ulation (Figure 5C). It must be noted that ATG16L1 overexpres- replication (data not shown). In agreement with the data from sion affected the Nodosome even in the absence of the Nod1 fibroblasts, treatment with the Rip2 inhibitor in ATG16L1 knock- ligand C12-iE-DAP; this most likely reflects the fact that Nod1 down intestinal epithelial cells blocked the increased CXCL2 and overexpression is in itself sufficient to trigger formation of a No- CXCL1 mRNA response to invasive Shigella without impacting dosome and activation of downstream signaling, as previously the response to the noninvasive strain (Figures 4F and shown (Inohara et al., 2000). 4G). Thus, ATG16L1 specifically suppressed proinflammatory To confirm the above results by using an endogenous system, responses mediated by Nod1 and Nod2. we infected control or ATG16L1 knockdown ModeK cells with Shigella (Figure 5D). Knockdown of ATG16L1 led to a shift of ATG16L1 Negatively Regulates the Activation of Rip2 Rip2 into denser fractions after infection, consistent with the and Its Incorporation in Large Signaling Complexes increased Nod-driven signaling in these cells. Moreover, the We next sought to determine at which step of the Nod1 and accumulation of Rip2 into denser fractions was dependent on Nod2 signal transduction pathway ATG16L1 exerts its regulatory bacterial invasion; noninvasive Shigella did not trigger this shift function. Because stimulation of either Nod1 or Nod2 leads to in ATG16L1 knockdown cells (Figure 5E). the activation of NF-kB, we employed an NF-kB-luciferase re- Thus, ATG16L1 functions at the initial steps of the Nod1 and porter. Knockdown of ATG16L1 led to enhanced NF-kB activa- Nod2 signaling pathway to negatively regulate the activation of tion after stimulation with Nod1 or Nod2 ligands but not TNF-a Rip2 and its incorporation into Nodosomes. (Figure 5A), thus showing that ATG16L1 negatively regulated Nod1 and Nod2 signaling upstream of or at the level of NF-kB The Roles of ATG16L1 in Nod Regulation and Autophagy activation. Can Be Molecularly Decoupled Upstream of Nod-activated NF-kB, the activation of Rip2 is We next asked whether the Nod-regulatory function of ATG16L1 controlled by the addition of K63-linked polyubiquitin. Therefore, could be separated at a molecular level from its role in autophagy we asked whether the presence of ATG16L1 affected Nod- induction. To address this, we generated an ATG16L1 truncated induced ubiquitination of Rip2. In human embryonic kidney form lacking the first 84 amino acids. The 85–607-truncated form 293T HEK293T cells expressing both Nod1 and ubiquitin, of ATG16L1 lacks the N-terminal region necessary for ATG5 stimulation with the Nod1 ligand C12-iE-DAP triggered Rip2 binding (Mizushima et al., 2003), yet it is still able to bind Nod1 polyubqitination, detected as an increase in high-molecular- or Nod2 (Travassos et al., 2010)(Figure 6A). We first verified weight Rip2 and K63-ubiquitin after Rip2 immunoprecipitation this and found that, indeed, the 85–607-truncated ATG16L1 (Figure 5B; compare lanes 1 and 2, middle and right panels). mutant did not bind to ATG5 but still interacted with Nod1 and Importantly, overexpression of ATG16L1 blocked the Nod1- Nod2 (Figures S4A and S4B), in agreement with our data sug- dependent K63-linked poly-ubiquitination of Rip2 (Figure 5B; gesting that Nod proteins interacted with the C-terminal portion compare lanes 2 and 4, middle and right panels). Therefore, of ATG16L1 (see Figure 4). We next established a knockdown- ATG16L1 inhibits Nod-dependent signaling at the level of or up- reconstitution system in which endogenous ATG16L1 was effi- stream of Rip2 ubiquitination. ciently depleted with an shRNA targeting the 30-UTR of human Initiation of Nod signaling is thought to occur through recruit- ATG16L1 (Figure 6B, top panel), and we then re-expressed fragment of the Rip2 adaptor into a large Nod signaling complex, ments of ATG16L1 by lentiviral transduction. ATG16L1 knock- or Nodosome. We next asked whether ATG16L1 regulated the down cells were rescued with Flag-mCherry or with Flag-tagged incorporation of Rip2 into the Nodosome and used sucrose constucts of either full-length ATG16L1 or the 85–607 truncation. gradient centrifugation to track macromolecular complex forma- Rescue of ATG16L1 knockdown was confirmed at both the pro- tion, as was recently reported in the case of MAVS (Hou et al., tein and RNA levels (Figures 6B and 6C). We then examined the

(B) CXCL1 was measured 16 hr after stimulation of control and DCCD knockdown ATG16L1DCCD MEFs with 20 mg/ml C12-iE-DAP or 100 ng/ml LPS (n = 3 from a representative experiment). (C) IL-8 production by control or ATG16L1-knockdown MDAMC cells was measured after 16 hr stimulation with C12-iE-DAP (20 mg/ml), TNF-a (1 ng/ml), and LPS (100 ng/ml) (n = 3). (D) CXCL1 production was measured after 16 hr of stimulation of control and ATG16L1-knockdown ModeK cells with C12-iE-DAP (20 mg/ml), L18-MDP (20 mg/ml), TNF-a (10 ng/ml), or LPS (100 ng/ml) (n = 3–8). (E) ATG16L1 mRNA levels in control and knockdown wild-type and Ripk2/ knockout MEFs. (F) After 3 hr infection with Shigella, CXCL2 was measured in control and knockdown MEFs (n = 6). (G and H) CXCL2 (G) and CXCL1 (H) mRNAs were measured in control and ATG16L1 knockdown ModeK cells infected for 10 hr with invasive or noninvasive Shigella in the presence or absence of 1 mM of Rip2 inhibitor (a representative of two independent experiments is shown). Values for ModeK cells are presented as n-fold changes relative to unstimulated controls. Values shown are means + SEM. *p < 0.05, **p < 0.01 by t test. See also Figure S3.

Immunity 39, 858–873, November 14, 2013 ª2013 Elsevier Inc. 865 Immunity ATG16L1 Negatively Regulates Nod1 and Nod2

Figure 5. ATG16L1 Regulates the Formation of the Nod Signaling Complex (A) NF-kB activation in control and ATG16L1-knockdown ModeK cells followed a 4.5 hr stimulation with C12-iE-DAP (10 or 50 mg/ml), L18-MDP (10 mg/ml), or TNF-a (10 ng/ml) (n = 3-5). Values are normalized to their respective unstimulated controls and are presented as means + SEM. *p < 0.05 by t test. (B) Rip2 was immunoprecipitated from Nod1- and ubiquitin-overexpressing HEK293T cells also transfected with ATG16L1 or pcDNA3 and stimulated with C12- iE-DAP (1 mg/ml) for 30 min. Ubiquitinated Rip2 was revealed by blots for Rip2 and K63-ubiquitin, and high-molecular-weight Rip2 and K63 are indicated with a bracket. (C and D) Rip20s incorporation into large complexes was assessed by sucrose-gradient ultracentrifugation after stimulation of ATG16L1- or control-cotransfected Nod1-expressing HEK293T cells with C12-iE-DAP (1 mg/ml) for 30 min (C), infection of control and ATG16L1-knockdown ModeK cells with Shigella for 30 min (D), and infection of ATG16L1-knockdown ModeK cells with invasive or non-invasive Shigella for 30 min (E). Fractions were collected and blotted for Rip2. Arrows indicate the most dense fraction with Rip2 signal. Data presented in (B)–(E) are representative images from 2–3 independent experiments.

866 Immunity 39, 858–873, November 14, 2013 ª2013 Elsevier Inc. Immunity ATG16L1 Negatively Regulates Nod1 and Nod2

ability of these ATG16L1 fragments to rescue autophagy induc- ical for the early targeting of intracellular Shigella and Salmonella tion. As expected, rescue with full-length ATG16L1 restored tar- to autophagosomes for degradation in the lysosome compart- geting of intracellular Shigella to autophagosomes (Figures 6D ment (Cooney et al., 2010; Travassos et al., 2010). Moreover, and 6E). In contrast, re-expression of the 85–607 truncation, direct stimulation of Nod1 and Nod2 by muramyl peptides or which is unable to interact with ATG5 (see Figure S4A), rescued peptidoglycan can initiate autophagosome formation (Cooney neither autophagy targeting of Shigella (Figures 6D and 6E) nor et al., 2010; Homer et al., 2010; Travassos et al., 2010). Here, rapamycin-induced accumulation of LC3 puncta (data not we have shown that the interaction of ATG16L1 with Nod1 or shown). We then characterized how these fragments affected Nod2 negatively regulates their signaling. We showed that cytokine responses after invasive bacterial infection. Re-expres- ATG16L1 functions at an upstream step of the Nod signaling sion of the full-length ATG16L1 reduced the enhanced cytokine pathway by negatively regulating the activation of Rip2 and the response observed in ATG16L1 knockdown cells after Shigella incorporation of Rip2 into large Nodosome complexes. Unex- infection by 32.8% ± 7.0% (Figure 6F). The partial rescue pectedly, this Nod-regulatory capacity of ATG16L1 was inde- observed is most likely explained by the heterogenous re- pendent of its role in canonical autophagy. Finally, we observed expression of ATG16L1 by transient transduction in knockdown that the capacity of ATG16L1 to modulate Nod signaling was cells. In contrast, re-expression of a general autophagy-compe- impaired with the expression of the CD-associated allele of tent N-terminal fragment of ATG16L1 containing amino acids ATG16L1. Thus, our results on ATG16L1 unveil an unanticipated 1–230 was unable to rescue the cytokine response (data not nonautophagic function of a protein from the core autophagic shown). Finally, the 85–607 fragment was able to reduce the machinery and support an emerging paradigm for distinct exacerbated cytokine response by 55.6% ± 5.2%, despite its biological processes (see also below). Because NOD2 and being unable to rescue autophagy induction (Figure 6F). Thus, ATG16L1 are susceptibility genes for CD, these observations although the N-terminal region of ATG16L1 (ATG5-binding and will also contribute to the understanding of the etiology of this coiled-coil domains) is required for autophagy induction, the disease. C-terminal region containing the WD40 repeats mediates the A growing body of evidence shows that the critical function of modulation of Nod signaling. Nod1 and Nod2 occurs at the point where membrane integrity is initially breached and peptidoglycan fragments enter the cytosol, Nod-Driven Inflammatory Responses Are Dysregulated either through the actions of type III or type IV secretion systems in Cells Expressing the 300A* Variant of ATG16L1 or through bacterial invasion (Philpott and Girardin, 2010). A SNP with a threonine-to-alanine substitution at position 300 of ATG16L1 also localizes in proximity to the plasma membrane ATG16L1 is associated with the development of CD. Although through its interaction with the heavy chain of clathrin (Raviku- this polymorphism has reproducibly been linked to CD, the two mar et al., 2010). Importantly, Nod1 and Nod2 direct autophagy alleles are present at approximately equal proportions in healthy by recruiting ATG16L1 to the plasma membrane at the site of controls, indicating that this polymorphism probably plays a bacterial entry (Travassos et al., 2010). Our findings extend this modifying role in CD rather than being a primary driver of disease model by demonstrating that significant regulation of the Nod- development. To investigate the effect of the 300A SNP during driven response also occurs at an early time point, most likely bacterial infection, we again used a knockdown-reconstitution at this membrane-signaling node within the cell. ATG16L1 spe- system to re-express either ATG16L1 300T or 300A variants in cifically regulated the cytokine response to cytosol-exposed the human epithelial MDAMC cells (Figures 6A and 6B). Both strains of Shigella and Listeria without affecting cytokine produc- variants of ATG16L1 were able to rescue anti-bacterial auto- tion triggered by infection with noninvasive Shigella or endo- phagy (Figures 6C and 6D), in agreement with a previous report some-confined Listeria strains. In addition, ATG16L1 regulated (Fujita et al., 2009). Having uncovered a regulatory role for the cytokine response to wild-type Salmonella infection that ATG16L1 in the Nod-driven cytokine response, we next asked triggers cytosol exposure during SCV remodeling, but it did not whether cells expressing the 300A risk variant had altered Nod regulate the response to an invasive yet SPI-1-deficient strain signaling in response to bacterial infection. Re-expression of that does not trigger cytosol exposure. In agreement with an the 300T variant resulted in a partial rescue of the enhanced early regulatory role for ATG16L1, blocking Shigella replication IL-8 response induced by Shigella in ATG16L1 knockdown cells 30 min after infection did not affect the dysregulation of (26.9% ± 4.0%) (Figures 7E and 7F), in line with the results Nod-driven signaling in ATG16L1-deficient cells, and changes obtained above (see Figure 6F). Interestingly, similar reconstitu- in the activation of Rip2 and complex formation were observed tion with the CD-associated variant 300A failed to rescue the as early as 30 min after infection or stimulation, in agreement exaggerated IL-8 expression induced by Shigella in ATG16L1 with our previous studies (Girardin et al., 2001). These findings knockdown cells. Together, these observations suggest that suggest that Nod signaling is initiated and regulated within this the ATG16L1 300A variant displays a blunted capacity to early time point after infection. In addition, Nod1 and Nod2 are mitigate Nod-dependent proinflammatory signaling, which could able to recruit ATG16L1 to the plasma membrane without coloc- contribute to heightened inflammation in CD-affected individuals alization of ATG5 (Travassos et al., 2010) and are able to interact carrying this SNP. with ATG16L1 constructs that lack ATG5 binding (Figure S4). Overall, the results presented here suggest that there exists, DISCUSSION prior to recruitment of the autophagy machinery, an early step at which ATG16L1 efficiently limits the activation of the Nod1 In addition to their function in driving proinflammatory cytokine and Nod2 signaling pathway independently of its subsequent production, Nod1 and Nod2 interact with ATG16L1, which is crit- role in targeting autophagosome formation. Importantly, and as

Immunity 39, 858–873, November 14, 2013 ª2013 Elsevier Inc. 867 Immunity ATG16L1 Negatively Regulates Nod1 and Nod2

Figure 6. The Nod-Regulatory Function of ATG16L1 Can Be Separated from Its Role in Autophagosome Formation (A) Schematic depiction of ATG16L1 and the 85–607, 1–230, and DCCD truncations, The sites ampliﬁed by ATG16L1 PCR 1 and 2 are shown. (B) ATG16L1 levels in control and ATG16L1 30-UTR shRNA knockdown MDAMC cells were examined by protein blot (top). Rescue with the indicated fragments was analyzed by Flag blot (bottom). (C) Expression of the ATG16L1 N-terminal (PCR 1), and coiled-coil domains (PCR 2) in control and rescued cells was analyzed by qPCR. Data presented in (B) and (C) are representative of three independent experiments. (legend continued on next page) 868 Immunity 39, 858–873, November 14, 2013 ª2013 Elsevier Inc. Immunity ATG16L1 Negatively Regulates Nod1 and Nod2

detailed below, it is likely that these early ATG16L1-dependent CD results from the dysregulation of the mucosal immune sys- events are those that are improperly regulated by the ATG16L1 tem and the consequent hyperactivation of inflammatory path- CD-associated variant. ways. Clearly, multiple triggers, both environmental and genetic, Previous reports have suggested an induced-proximity Nod are responsible for the initiation of disease. Our results demon- signaling model in which oligomerization of Nod1 or Nod2 results strate that the link between two CD-susceptibility genes, in the recruitment of Rip2 (Inohara et al., 2000; Ogura et al., NOD2 and ATG16L1, goes beyond the targeting of microbes 2001b). Our results from sucrose gradient fractionations provide for degradation or antigen presentation and directly affects the biochemical evidence in support of this model. In both overex- inflammatory response to infection. Several defects associated pression and knockdown conditions, ATG16L1 was found to with the T300A allele of ATG16L1, including defective MDP- negatively regulate this process. Currently, it remains unclear induced autophagy in DCs (Cooney et al., 2010) and lympho- whether the ubiquitination of Rip2 occurs prior to or subsequent blasts (Travassos et al., 2010), have been reported in human to its recruitment to a large Nod-dependent scaffold and whether cells. In in vitro reconstitution systems, some groups have ATG16L1 negatively regulates the incorporation of Rip2 into the reported defective targeting of Salmonella in cells expressing complex or acts indirectly by controlling the recruitment of a the 300A allele (Kuballa et al., 2008), whereas others have tertiary factor required for the activation and/or retention of reported no difference (Fujita et al., 2009). In our reconstitution Rip2 within a signaling complex. Disentangling this process will system, both the 300T and 300A alleles of ATG16L1 equally provide insight both into the Nod signaling pathway and into rescued targeting of Shigella to autophagosomes. The discrep- the fine regulatory mechanisms dependent on ATG16L1. ancies in these findings might result from differences in the Autophagy is emerging as one of the critical effector mecha- sensitivity of the reconstitution systems to endogenously ex- nisms of the immune response. Indeed, autophagy contains pressed ATG16L1, as well as from differences between initiating intracellular bacteria and delivers them to the lysosome to pro- autophagy with direct stimulation of Nod2 or infection with bac- mote clearance from infected cells. In addition, autophagy can teria that may trigger autophagy through multiple pathways. influence other processes of the immune response and immune Our results support the notion that the CD-associated homeostasis, including antigen presentation (Cooney et al., ATG16L1 variant causes an improper control of inflammatory 2010; Lee et al., 2010), inflammasome activation (Harris signaling dependent on Nod1 and Nod2 activation given that et al., 2011; Nakahira et al., 2011; Saitoh et al., 2008; Zhou we observed that the ATG16L1 300A variant displayed an altered et al., 2011), and Paneth cell function (Cadwell et al., 2008; Witt- capacity to downregulate Nod-dependent proinflammatory kopf et al., 2012), among others (Deretic, 2011). In line with a signaling. These findings provide a molecular mechanism in critical function of autophagy in immunity, the interaction of support of the observations by Plantinga et al., who reported, ATG16L1 with Nod1 and Nod2 allows for the targeting of in human PBMCs from donors homozygous for the T300A allele intracellular bacteria to autophagosomes. Unexpectedly, we of ATG16L1, increased cytokine production specifically in observed that the Nod-regulatory function of ATG16L1 to response to Nod2 but not TLR2 or TLR4 ligand stimulation (Plan- suppress inflammation was independent of its canonical role in tinga et al., 2011). Importantly, our results provide further autophagy induction. These findings support a growing body evidence that Nod2, which is strongly linked to CD development, of evidence that, in the course of initiating autophagy, autophagy and ATG16L1 are functionally linked in cell signaling cascades proteins can have ancillary functions that are independent from that are improperly regulated in CD. In agreement, recent work their canonical roles in autophagosome formation. Indeed, the has demonstrated that ATG16L1 hypomorphic mice have a ATG5-12 conjugate regulates RIG-I signaling (Jounai et al., hyperimmune phenotype that protects them from Citrobacter 2007), and ATG9a regulates the association of STING with infection and that the protective hyperimmune phenotype is TBK1 (Saitoh et al., 2009). An autophagosome-independent lost in mice doubly deficient for Nod2 and Atg16L1 (Marchiando function of ATG16L1 in hormone secretion has also been identi- et al., 2013). However, the present observations suggest that, fied (Ishibashi et al., 2012). Furthermore, recent work has contrary to previous expectations, it might be the Nod-depen- implicated ATG7 and the ATG16L1:ATG5-ATG12 complex in dent signaling function rather than the autophagy-targeting an autophagy-independent function negatively regulating property that would be improperly modulated by the CD-associ- murine norovirus replication, thus further demonstrating that ated ATG16L1 variant. subsets of autophagy proteins can also be co-opted for nonau- In summary, we have identified ATG16L1 as a negative regu- tophagosome functions in the immune response (Hwang et al., lator of the Nod1 and Nod2 signaling pathway and have shown 2012). In this regard, it is interesting to notice that only the that it functions in an autophagosome-independent manner to N-terminal region of ATG16L1 is required for core autophagy negatively regulate the activation of Rip2. This regulation is regulation, as evidenced by the fact that yeast ATG16 lacks essential for modulating the response to both Nod stimulation the C-terminal WD40 repeat domain. It is thus possible that the and infection with invasive intestinal pathogens, suggesting presence of the WD40 repeat domain in the ATG16 ortholog of that it might play a critical role at the level of the intestinal higher eukaryotes conferred additional functions, such as the mucosa. These results strengthen the connection between regulation of Nod signaling in mammalian cells, to this protein. Nod proteins and ATG16L1 and support a model in which

(D and E) Targeting of Shigella (DAPI, blue) to autophagosomes (GFP-LC3, green) 2 hr after infection was assessed in control and rescued cells. Representative images are shown in (D), and targeting was quantiﬁed (E) (n = 3). (F) IL-8 mRNA was measured 4.5 hr after infection with Shigella in control and rescued cells (n = 7). Values shown are means + SEM. *p < 0.05, **p < 0.01 by t test. See also Figure S4.

Immunity 39, 858–873, November 14, 2013 ª2013 Elsevier Inc. 869 Immunity ATG16L1 Negatively Regulates Nod1 and Nod2

(legend on next page)

870 Immunity 39, 858–873, November 14, 2013 ª2013 Elsevier Inc. Immunity ATG16L1 Negatively Regulates Nod1 and Nod2

perturbation of this axis, either in terms of autophagy induction or ATG16L1, Flag-85-607, or Flag-DCCD. AFter 16 hr, cells were lysed with cytokine responses, can contribute to dyseregulation of the cold RIPA buffer. Lysates were incubated overnight at 4 C with 50 ml Protein m mucosal immune response. G beads (Thermo Scientific) and 1 l mouse anti-Flag M2 antibody (Sigma). For Rip2 polyubiquitination assays, HEK293T cells in 10 cm dishes were transfected with 50 ng of HA-Nod1 and 1.5 mg Flag-ATG16L1 or 1.5 mg pcDNA3. EXPERIMENTAL PROCEDURES After 16 hr, cells were stimulated with 1 mg/ml C12-iE-DAP and lysed in cold RIPA buffer 30 min after stimulation. Lysates were incubated overnight at Cell Culture and Transductions 4C with 50 ml Protein A beads (Thermo Scientific) and 1 ml rabbit anti-Rip2 / Primary wild-type and Ripk2 MEFs were prepared as previously described antibody (Santa Cruz Biotechnology). After three washes, immunoprecipitated / (Travassos et al., 2010). Primary MEFs, wild-type MEFs, Atg5 MEFs, proteins were analyzed by protein blot. ATG16L1 DCCD immortalized MEFs, ModeK cells, MDAMCs that stably express GFP-LC3, and HEK293T cells were grown in DMEM supplemented Sucrose-Gradient Ultracentrifugation with 10% FBS (Wisent). As further described in the Supplemental Experimental For sucrose-gradient experiments, HEK293T cells were transfected and Procedures, in knockdown experiments cells were seeded in 6-well plates and stimulated as described for Rip2 immunoprecipitations (see above). For transduced with 1 ml of lentivirus preparation, and 24 hr after transduction cells ModeK sucrose-gradient experiments, cells were seeded at 3 3 105 cells/ were split into media containing puromycin. Knockdown cells were used well in 6-well plates and infected with invasive or noninvasive Shigella as 5 days after transduction. For rescue experiments, cells were transduced described above. Three wells from a 6-well plate were pooled for each sample. with rescue lentivirus 20 hr prior to use (as described in the Supplemental Lysates were loaded on a 3%–30% discontinuous sucrose gradient in 50 mM Experimental Procedures). Tris-HCL and 150 mM NaCL. Samples were centrifuged at 100,000 3 g for 16 hr at 4C, and fractions were collected and analyzed by protein blot. Infections ModeK cells were seeded at a density of 1.2 3 105 cells/well in 24-well plates Statistical Analysis (ELISAs) or 3 3 105 cells per well in 6-well plates (RNA) and infected the For Figures 1B and 2A–2C and Figures S3A, S3C, and S3G, a one-way ANOVA following day. MEFs were seeded at a density of 8 3 104 cells/well in 24- followed by a Tukey post-test was used; otherwise, a Student’s t test was well plates. MDAMC cells were seeded at 1 3 105 cells/well in 24-well plates. used. The significance of rescue percentages was determined by comparison Bacteria were grown overnight as described in the Supplemental Experimental of the observed rescue with a mean rescue of 0% via a one-sample t test. An Procedures. For infections, overnight cultures were diluted, and bacteria were asterisk indicates p < 0.05, and two asterisks indicate p < 0.01. Unless used in mid-to-late log phase at an OD650 nm of 0.5 for Shigella, 0.8 for Listeria, otherwise indicated, all experiments were performed in at least three indepen- and 2.3–2.6 for Salmonella. For cytokine production assays, Shigella and dent replicates. Listeria were added to the cells at 50 moi (multiplicity of infection) and spun at 2,000 rpm for 10 min. Salmonella was added to the cells at 25 MOI and spun at 1,000 rpm for 5 min. After centrifugation, infected cells were incubated SUPPLEMENTAL INFORMATION at 37C for 20 min, cells were then washed three times with PBS, and media containing 50 mg/ml gentamycin was added. In experiments in which bacterial Supplemental Information include Supplemental Experimental Procedures, replication was blocked after invasion, medium containing 50 mg/ml of genta- four figures, and three tables and can be found with this article online at mycin and 5 mg/ml of chloroamphenicol was added. In experiments in which http://dx.doi.org/10.1016/j.immuni.2013.10.013. autophagosome maturation was blocked, medium containing 50 mg/ml gentamycin and 10 mM chloroquine (Invivogen) was added 30 min after invasion. ACKNOWLEDGMENTS

Stimulations This work was supported by grants from the Canadian Institutes of Health ModeK or MEF cells were stimulated with 10 or 20 mg/ml C12-iE-DAP (Invivo- Research (CIHR) (to D.J.P and S.E.G.) and the Crohn’s and Colitis Foundation Gen), 10 or 20 mg/ml L18-MDP (InvivoGen), 10 ng/mL TNF-a (InvivoGen), or of Canada (to N.L.J., S.E.G., and D.J.P.). M.T.S. was supported by fellowships 100 ng/ml LPS (InvivoGen). For inhibition of Rip2, cells were pretreated with from the Natural Sciences and Engineering Research Council (CGS-M), the 1 mM Rip2 inhibitor (GSK’214) for 30 min and then infected in the presence Ontario Genealogical Society, and CIHR (CGS-D) and by CIHR Strategic of inhibitor for 10 hr. MDAMC cells were stimulated with 20 mg/ml C12-iE- Training Fellowship STP-53877. We would like to thank Glaxo-Smith Kline DAP (InvivoGen), 1 ng/ml TNF-a (InvivoGen), or 100 ng/ml LPS (InvivoGen). for generously providing the Rip2 inhibitors.

ELISAs Received: May 4, 2012 Murine CXCL1 and human IL-8 ELISA assays were performed according to the Accepted: September 3, 2013 manufacturer’s protocols (RND Biosystems). Published: November 14, 2013 See Supplemental Experimental Procedures for protein blot, immunoﬂuore- sence, qPCR, and luciferase assay procedures. REFERENCES

Immunoprecipitations Birmingham, C.L., Smith, A.C., Bakowski, M.A., Yoshimori, T., and Brumell, For interaction experiments, HEK293T cells in 10 cm dishes were transfected J.H. (2006). Autophagy controls Salmonella infection in response to damage with 3 mg of HA-Nod1 or HA-Nod2 and with 500 ng of Flag-Cherry, Flag- to the Salmonella-containing vacuole. J. Biol. Chem. 281, 11374–11383.

Figure 7. The 300A Allele of ATG16L1 Is Defective in Its Regulation of the Shigella-Induced IL-8 Response (A and B) ATG16L1-knockdown MDAMC cells were rescued with mCherry, ATG16L1 300T, or ATG16L1 300A. Expression of the rescue proteins was assessed by qPCR examining expression of the coiled-coil domain (PCR 2) of ATG16L1 (A) or by protein blot (B). Data presented in (A) and (B) are representative of at least three independent experiments. (C and D) Rescued MDAMC cells were infected with Shigella, and targeting of GFP-LC3 (green) to intracellular Shigella (DAPI, blue) was examined. The number of GFP-LC3+ Shigella 2 hr after infection was quantiﬁed (n = 3). (E) IL-8 mRNA was measured 4.5 hr after infection with Shigella in rescued cells (representative data from one experiment of six independent experiments are shown). (F) Percent rescue of the enhanced IL-8 response after Shigella infection of ATG16L1-knockdown cells was calculated for the 300T and 300A variants (n = 6). Values shown are means + SEM. *p < 0.05 by t test.

Immunity 39, 858–873, November 14, 2013 ª2013 Elsevier Inc. 871 Immunity ATG16L1 Negatively Regulates Nod1 and Nod2

Birmingham, C.L., Canadien, V., Gouin, E., Troy, E.B., Yoshimori, T., Cossart, Inohara, N., Koseki, T., Lin, J., del Peso, L., Lucas, P.C., Chen, F.F., Ogura, Y., P., Higgins, D.E., and Brumell, J.H. (2007). Listeria monocytogenes evades and Nu´ n˜ ez, G. (2000). An induced proximity model for NF-kappa B activation in killing by autophagy during colonization of host cells. Autophagy 3, 442–451. the Nod1/RICK and RIP signaling pathways. J. Biol. Chem. 275, 27823–27831. Cadwell, K., Liu, J.Y., Brown, S.L., Miyoshi, H., Loh, J., Lennerz, J.K., Kishi, C., Inohara, N., Ogura, Y., Fontalba, A., Gutierrez, O., Pons, F., Crespo, J., Fukase, Kc, W., Carrero, J.A., Hunt, S., et al. (2008). A key role for autophagy and the K., Inamura, S., Kusumoto, S., Hashimoto, M., et al. (2003). Host recognition of autophagy gene Atg16l1 in mouse and human intestinal Paneth cells. Nature bacterial muramyl dipeptide mediated through NOD2. Implications for Crohn’s 456, 259–263. disease. J. Biol. Chem. 278, 5509–5512. Chamaillard, M., Hashimoto, M., Horie, Y., Masumoto, J., Qiu, S., Saab, L., Ishibashi, K., Uemura, T., Waguri, S., and Fukuda, M. (2012). Atg16L1, an Ogura, Y., Kawasaki, A., Fukase, K., Kusumoto, S., et al. (2003). An essential essential factor for canonical autophagy, participates in hormone secretion role for NOD1 in host recognition of bacterial peptidoglycan containing diami- from PC12 cells independently of autophagic activity. Mol. Biol. Cell 23, nopimelic acid. Nat. Immunol. 4, 702–707. 3193–3202. Cooney, R., Baker, J., Brain, O., Danis, B., Pichulik, T., Allan, P., Ferguson, Jounai, N., Takeshita, F., Kobiyama, K., Sawano, A., Miyawaki, A., Xin, K.Q., D.J., Campbell, B.J., Jewell, D., and Simmons, A. (2010). NOD2 stimulation Ishii, K.J., Kawai, T., Akira, S., Suzuki, K., and Okuda, K. (2007). The Atg5 induces autophagy in dendritic cells influencing bacterial handling and antigen Atg12 conjugate associates with innate antiviral immune responses. Proc. presentation. Nat. Med. 16, 90–97. Natl. Acad. Sci. USA 104, 14050–14055. Deretic, V. (2011). Autophagy in immunity and cell-autonomous defense Kuballa, P., Huett, A., Rioux, J.D., Daly, M.J., and Xavier, R.J. (2008). Impaired against intracellular microbes. Immunol. Rev. 240, 92–104. autophagy of an intracellular pathogen induced by a Crohn’s disease associated ATG16L1 variant. PLoS ONE 3, e3391. Dupont, N., Lacas-Gervais, S., Bertout, J., Paz, I., Freche, B., Van Nhieu, G.T., Lee, H.K., Mattei, L.M., Steinberg, B.E., Alberts, P., Lee, Y.H., Chervonsky, A., van der Goot, F.G., Sansonetti, P.J., and Lafont, F. (2009). Shigella phagocytic Mizushima, N., Grinstein, S., and Iwasaki, A. (2010). In vivo requirement for vacuolar membrane remnants participate in the cellular response to pathogen Atg5 in antigen presentation by dendritic cells. Immunity 32, 227–239. invasion and are regulated by autophagy. Cell Host Microbe 6, 137–149. Lee, J., Kim, H.R., Quinley, C., Kim, J., Gonzalez-Navajas, J., Xavier, R., and Fujita, N., Itoh, T., Omori, H., Fukuda, M., Noda, T., and Yoshimori, T. (2008). Raz, E. (2012). Autophagy suppresses interleukin-1b (IL-1b) signaling by acti- The Atg16L complex specifies the site of LC3 lipidation for membrane biogen- vation of p62 degradation via lysosomal and proteasomal pathways. J. Biol. esis in autophagy. Mol. Biol. Cell 19, 2092–2100. Chem. 287, 4033–4040. Fujita, N., Saitoh, T., Kageyama, S., Akira, S., Noda, T., and Yoshimori, T. Marchiando, A.M., Ramanan, D., Ding, Y., Gomez, L.E., Hubbard-Lucey, V.M., (2009). Differential involvement of Atg16L1 in Crohn disease and canonical Maurer, K., Wang, C., Ziel, J.W., van Rooijen, N., Nun˜ ez, G., et al. (2013). A autophagy: analysis of the organization of the Atg16L1 complex in fibroblasts. deficiency in the autophagy gene Atg16L1 enhances resistance to enteric J. Biol. Chem. 284, 32602–32609. bacterial infection. Cell Host Microbe 14, 216–224. Girardin, S.E., Tournebize, R., Mavris, M., Page, A.L., Li, X., Stark, G.R., Bertin, Mizushima, N., Kuma, A., Kobayashi, Y., Yamamoto, A., Matsubae, M., Takao, J., DiStefano, P.S., Yaniv, M., Sansonetti, P.J., and Philpott, D.J. (2001). T., Natsume, T., Ohsumi, Y., and Yoshimori, T. (2003). Mouse Apg16L, a novel CARD4/Nod1 mediates NF-kappaB and JNK activation by invasive Shigella WD-repeat protein, targets to the autophagic isolation membrane with the flexneri. EMBO Rep. 2, 736–742. Apg12-Apg5 conjugate. J. Cell Sci. 116, 1679–1688. Girardin, S.E., Boneca, I.G., Viala, J., Chamaillard, M., Labigne, A., Thomas, Nakahira, K., Haspel, J.A., Rathinam, V.A., Lee, S.J., Dolinay, T., Lam, H.C., G., Philpott, D.J., Sansonetti, P.J., Labigne, A., Zahringer, U., et al. (2003a). Englert, J.A., Rabinovitch, M., Cernadas, M., Kim, H.P., et al. (2011). Nod2 is a general sensor of peptidoglycan through muramyl dipeptide Autophagy proteins regulate innate immune responses by inhibiting the (MDP) detection. J. Biol. Chem. 278, 8869–8872. release of mitochondrial DNA mediated by the NALP3 inflammasome. Nat. Girardin, S.E., Boneca, I.G., Carneiro, L.A., Antignac, A., Je´ hanno, M., Viala, J., Immunol. 12, 222–230. Tedin, K., Taha, M.K., Labigne, A., Za¨ hringer, U., et al. (2003b). Nod1 detects a Ogawa, M., Yoshimori, T., Suzuki, T., Sagara, H., Mizushima, N., and unique muropeptide from gram-negative bacterial peptidoglycan. Science Sasakawa, C. (2005). Escape of intracellular Shigella from autophagy. 300, 1584–1587. Science 307, 727–731. Hampe, J., Franke, A., Rosenstiel, P., Till, A., Teuber, M., Huse, K., Albrecht, Ogura, Y., Bonen, D.K., Inohara, N., Nicolae, D.L., Chen, F.F., Ramos, R., M., Mayr, G., De La Vega, F.M., Briggs, J., et al. (2007). A genome-wide asso- Britton, H., Moran, T., Karaliuskas, R., Duerr, R.H., et al. (2001a). A frameshift ciation scan of nonsynonymous SNPs identifies a susceptibility variant for mutation in NOD2 associated with susceptibility to Crohn’s disease. Nature Crohn disease in ATG16L1. Nat. Genet. 39, 207–211. 411, 603–606. Harris, J., Hartman, M., Roche, C., Zeng, S.G., O’Shea, A., Sharp, F.A., Lambe, Ogura, Y., Inohara, N., Benito, A., Chen, F.F., Yamaoka, S., and Nunez, G. E.M., Creagh, E.M., Golenbock, D.T., Tschopp, J., et al. (2011). Autophagy (2001b). Nod2, a Nod1/Apaf-1 family member that is restricted to monocytes controls IL-1beta secretion by targeting pro-IL-1beta for degradation. and activates NF-kappaB. J. Biol. Chem. 276, 4812–4818. J. Biol. Chem. 286, 9587–9597. Philpott, D.J., and Girardin, S.E. (2010). Nod-like receptors: sentinels at host Homer, C.R., Richmond, A.L., Rebert, N.A., Achkar, J.P., and McDonald, C. membranes. Curr. Opin. Immunol. 22, 428–434. (2010). ATG16L1 and NOD2 interact in an autophagy-dependent antibacterial Plantinga, T.S., Crisan, T.O., Oosting, M., van de Veerdonk, F.L., de Jong, D.J., pathway implicated in Crohn’s disease pathogenesis. Gastroenterology 139, Philpott, D.J., van der Meer, J.W., Girardin, S.E., Joosten, L.A., and Netea, 1630–1641, 1641 e1–2. M.G. (2011). Crohn’s disease-associated ATG16L1 polymorphism modulates Hou, F., Sun, L., Zheng, H., Skaug, B., Jiang, Q.X., and Chen, Z.J. (2011). pro-inflammatory cytokine responses selectively upon activation of NOD2. MAVS forms functional prion-like aggregates to activate and propagate anti- Gut 60, 1229–1235. viral innate immune response. Cell 146, 448–461. Ravikumar, B., Moreau, K., Jahreiss, L., Puri, C., and Rubinsztein, D.C. (2010). Hugot, J.P., Chamaillard, M., Zouali, H., Lesage, S., Ce´ zard, J.P., Belaiche, J., Plasma membrane contributes to the formation of pre-autophagosomal struc- Almer, S., Tysk, C., O’Morain, C.A., Gassull, M., et al. (2001). Association of tures. Nat. Cell Biol. 12, 747–757. NOD2 leucine-rich repeat variants with susceptibility to Crohn’s disease. Saitoh, T., Fujita, N., Jang, M.H., Uematsu, S., Yang, B.G., Satoh, T., Omori, H., Nature 411, 599–603. Noda, T., Yamamoto, N., Komatsu, M., et al. (2008). Loss of the autophagy Hwang, S., Maloney, N.S., Bruinsma, M.W., Goel, G., Duan, E., Zhang, L., protein Atg16L1 enhances endotoxin-induced IL-1beta production. Nature Shrestha, B., Diamond, M.S., Dani, A., Sosnovtsev, S.V., et al. (2012). 456, 264–268. Nondegradative role of Atg5-Atg12/ Atg16L1 autophagy protein complex in Saitoh, T., Fujita, N., Hayashi, T., Takahara, K., Satoh, T., Lee, H., Matsunaga, antiviral activity of interferon gamma. Cell Host Microbe 11, 397–409. K., Kageyama, S., Omori, H., Noda, T., et al. (2009). Atg9a controls

872 Immunity 39, 858–873, November 14, 2013 ª2013 Elsevier Inc. Immunity ATG16L1 Negatively Regulates Nod1 and Nod2

dsDNA-driven dynamic translocation of STING and the innate immune stalling pre-autophagosomal structures. EMBO J. Published online October response. Proc. Natl. Acad. Sci. USA 106, 20842–20846. 25, 2013. http://dx.doi.org/10.1038/emboj.2013.234. Sorbara, M.T., and Philpott, D.J. (2011). Peptidoglycan: a critical activator of the mammalian immune system during infection and homeostasis. Immunol. Travassos, L.H., Carneiro, L.A., Ramjeet, M., Hussey, S., Kim, Y.G., Rev. 243, 40–60. Magalha˜ es, J.G., Yuan, L., Soares, F., Chea, E., Le Bourhis, L., et al. (2010). Nod1 and Nod2 direct autophagy by recruiting ATG16L1 to the plasma Tattoli, I., Philpott, D.J., and Girardin, S.E. (2012a). The bacterial and cellular membrane at the site of bacterial entry. Nat. Immunol. 11, 55–62. determinants controlling the recruitment of mTOR to the Salmonella-containing vacuole. Biol. Open 1, 1215–1225. Wittkopf, N., Gu¨ nther, C., Martini, E., Waldner, M., Amann, K.U., Neurath, M.F., Tattoli, I., Sorbara, M.T., Vuckovic, D., Ling, A., Soares, F., Carneiro, L.A., and Becker, C. (2012). Lack of intestinal epithelial atg7 affects paneth cell Yang, C., Emili, A., Philpott, D.J., and Girardin, S.E. (2012b). Amino acid star- granule formation but does not compromise immune homeostasis in the gut. vation induced by invasive bacterial pathogens triggers an innate host defense Clin. Dev. Immunol. 2012, 278059. program. Cell Host Microbe 11, 563–575. Tattoli, I., Sorbara, M.T., Yang, C., Tooze, S.A., Philpott, D.J., and Girardin, Zhou, R., Yazdi, A.S., Menu, P., and Tschopp, J. (2011). A role for mitochondria S.E. (2013). Listeria phospholipases subvert host autophagic defenses by in NLRP3 inﬂammasome activation. Nature 469, 221–225.

Immunity 39, 858–873, November 14, 2013 ª2013 Elsevier Inc. 873