a crohn’s disease variant in atg16l1 enhances its degradation by caspase 3

ARTICLEdoi:10.1038/nature13044

A Crohn’s disease variant in Atg16l1enhances its degradation by caspase 3Aditya Murthy1, Yun Li1, Ivan Peng1, Mike Reichelt2, Anand Kumar Katakam2, Rajkumar Noubade1, Merone Roose-Girma3,Jason DeVoss1, Lauri Diehl2, Robert R. Graham4 & Menno van Lookeren Campagne1

Crohn’s disease is a debilitating inflammatory bowel disease (IBD) that can involve the entire digestive tract. Asingle-nucleotide polymorphism (SNP) encoding a missense variant in the autophagy gene ATG16L1 (rs2241880,Thr300Ala) is strongly associated with the incidence of Crohn’s disease. Numerous studies have demonstrated theeffect of ATG16L1 deletion or deficiency; however, the molecular consequences of the Thr300Ala (T300A) variantremains unknown. Here we show that amino acids 296–299 constitute a caspase cleavage motif in ATG16L1 and thatthe T300A variant (T316A in mice) significantly increases ATG16L1 sensitization to caspase-3-mediated processing. Weobserved that death-receptor activation or starvation-induced metabolic stress in human and murine macrophagesincreased degradation of the T300A or T316A variants of ATG16L1, respectively, resulting in diminished autophagy.Knock-in mice harbouring the T316A variant showed defective clearance of the ileal pathogen Yersinia enterocoliticaand an elevated inflammatory cytokine response. In turn, deletion of the caspase-3-encoding gene, Casp3, orelimination of the caspase cleavage site by site-directed mutagenesis rescued starvation-induced autophagy andpathogen clearance, respectively. These findings demonstrate that caspase 3 activation in the presence of a commonrisk allele leads to accelerated degradation of ATG16L1, placing cellular stress, apoptotic stimuli and impaired autophagyin a unified pathway that predisposes to Crohn’s disease.

Genome-wide association studies have contributed substantial insightinto genetically complex diseases such as cancer and chronic inflam-matory diseases1. Although disease-contributing polymorphisms providecorrelative evidence for the involvement of specific genes in pathology,mechanistic consequences of the majority of polymorphisms have yetto be unravelled. To date, over 150 susceptibility loci have been iden-tified for Crohn’s disease2. Crohn’s disease can affect the entire gastro-intestinal tract, with disease most commonly found in the distal smallintestine (terminal ileum) and ascending colon. Significant changes inlocal cytokine production, intestinal microflora, mucosal barrier functionand epithelial cell renewal are simultaneously observed3. Genome-wideassociation studies performed in geographically distant populationshave consistently identified SNPs in the genes Irgm, Card15/Nod2,Il23r, Lrrk2 and Atg16l1 as strongly associated variants in Crohn’sdisease2,4. More recently, cooperation between autophagy and the clin-ically relevant endoplasmic reticulum stress response gene Xbp1 wasshown to synergistically prevent ileal inflammation2,5. Ontologically,these findings implicate autophagy, cellular stress regulation and micro-bial pathogen sensing as key pathways in intestinal inflammation.

Macroautophagy (herein referred to as autophagy) is a cellular recyc-ling process where cytosolic cargo is engulfed in a double-membranevesicle and targeted for degradation by lysosomal fusion (reviewed inref. 6). Xenophagy is a targeted form of autophagy where intracellularpathogens are sequestered in autophagosomes for lysosomal degradation,and bridges pathogen sensors and the canonical autophagy machinery7.The autophagy gene Atg16l1 is a central adaptor required for the forma-tion of a mature autophagosome, and the A . G SNP (rs2241880, whereA 5 non-risk, G 5 risk for Crohn’s disease) encodes a Thr 300-to-Alavariant (Thr 5 non-risk and Ala 5 risk for Crohn’s disease) in exon 9.Since its identification, this SNP has remained one of the most clinically

significant variants in Crohn’s disease8. Numerous in vivo and in vitromodel systems have emerged to delineate the contribution(s) of ATG16L1loss-of-function to canonical autophagy, xenophagy and inflammat-ory signalling9–16. Despite intense investigation, definitive proof thatT300A is the causal allele on the risk haplotype is elusive, and the keymechanism(s) underpinning the consequences of T300A variant remainunresolved. This study demonstrates that the T300A variant sensitizesATG16L1 to caspase-3-mediated degradation, thereby revealing a func-tional connection between Crohn’s disease, caspase activation andautophagy.

T300A variant enhances ATG16L1 cleavageWe initiated our studies by performing multiple sequence alignment ofATG16L1 protein, focusing on the region proximal to Thr 300. ATG16L1is an adaptor consisting of an amino-terminal coiled–coil domain fol-lowed by seven WD-repeat domains. Five isoforms have been describedfor human ATG16L1 (NP_110430.5, NP_060444.3, NP_942593.2, NP_001177195.1 and NP_001177196.1). ATG16L1a and b are the majorisoforms expressed in intestinal epithelium8 and macrophages (currentstudy), and all isoforms encode exon 9, which contains Thr 300. Murinecells express detectable levels of additional variants (termed ATG16L1a9

and b9, described elsewhere13). Multiple sequence alignment analysisrevealed conserved putative caspase target sequences (DxxD) at aminoacids 296–299 and 319–322 of human ATG16L1 (NP_110430.5, Fig. 1a).In-vitro-translated ATG16L1bwas incubated with recombinant activecaspase 3, and the T300A variant showed significantly enhanced cleavage.Abolishing the 296–299 consensus caspase cleavage site by site-directedmutagenesis of D299E completely prevented ATG16L1b processing(Fig. 1b). Recombinant active caspase 1 and 6 also processed the T300Avariant of ATG16L1b, but with lower efficiency compared to caspase 3

1Department of Immunology, Genentech, Inc., 1 DNA Way, South San Francisco, California 94080, USA. 2Department of Pathology, Genentech, Inc., 1 DNA Way, South San Francisco, California 94080, USA.3Department of Molecular Biology, Genentech, Inc., 1 DNA Way, South San Francisco, California 94080, USA. 4ITGR Human Genetics, Genentech, Inc., 1 DNA Way, South San Francisco, California94080, USA.

0 0 M O N T H 2 0 1 4 | V O L 0 0 0 | N A T U R E | 1

Macmillan Publishers Limited. All rights reserved©2014

www.nature.com/doifinder/10.1038/nature13044

(Extended Data Fig. 1a). Transfection of HeLa cells with the sameconstructs demonstrated that death-receptor-mediated apoptosis bytumour necrosis factor (TNF)-a cleaves ATG16L1b; this is enhancedby the T300A variant and abrogated in the D299E mutant (ExtendedData Fig. 1b). Overexpression of ATG16L1 in cell lines results in degra-dation of excess protein, indicating that an upper limit of ATG16L1protein level is maintained by cells13 through constitutive ATG16L1bcleavage in addition to that induced by TNF-a stimulation (see 0 h timepoints, Extended Data Fig. 1b). The consensus caspase cleavage sitesflanking the ATG16L1 T300A SNP are highly conserved in human andmouse (Fig. 1a). We reconstituted ATG16L1-deficient murine bonemarrow progenitors with wild-type, T316A (corresponding to humanT300A) or D315E (human D299E) variants of ATG16L1b (ExtendedData Fig. 1c). TNFR1 (also known as TNFRSF1A) stimulation of thesecells demonstrated that, consistent with human ATG16L1, cleavage ofmurine ATG16L1b is enhanced by the T316A variant but abolished bythe D315E variant (Extended Data Fig. 1d). Importantly, these data alsoindicate that amino acids 319–322 do not constitute a caspase cleavagesite. We next assessed ATG16L1 processing in primary cells. First, peri-pheral blood monocyte-derived macrophages from human donorshomozygous for non-risk (wild-type) or T300A variant of ATG16L1were treated with increasing doses of TNF-a in the presence of cyclo-heximide to induce caspase 3 activation. Immunoblotting for ATG16L1indicated that the T300A variant significantly enhanced its processing,as indicated by the emergence of 36-kDa and 34-kDa caspase cleavageproducts. Densitometry analysis of immunoblots analysing the 36-kDaand 34-kDa fragments vs full-length ATG16L1 from a number of donorsrevealed that the T300A variant measurably increased the caspase cleav-age product in a TNF-a dose-dependent manner (Fig. 1c). Caspase 3

cleavage or the generation of cleaved PARP, an independent caspase 3substrate, was not increased in donor cells with the T300A variant(Extended Data Fig. 2), confirming that accelerated processing of theT300A variant is not due to increased caspase 3 activity. We generatedC57BL/6N mice with a T316A knock-in mutation to recapitulate theCrohn’s disease-associated SNP in vivo (Extended Data Fig. 3a, b).Mice were grossly normal, and no overt developmental (data not shown)or inflammatory phenotypes were observed by flow cytometry analysisof immune organs (Extended Data Fig. 4). Histological analysis showedno abnormal morphology of small intestine (Extended Data Fig. 5a) orcolon (data not shown). Quantification of small intestine Paneth cellgranules showed comparable morphology at baseline conditions betweenwild-type and T316A mice (Extended data Fig. 5b–d). Bone-marrow-derived macrophages from these mice were treated with TNF-a in thepresence of cycloheximide to assess ATG16L1 cleavage as in humanmacrophages. Murine ATG16L1 cleavage was greatly increased by theT316A mutation in a TNF-adose-dependent manner (Fig. 1d). Specificcomparison of thea andb isoforms showed that whereas ATG16L1a isexpressed at higher levels, ATG16L1a andb are processed equally afterTNF-a treatment (Extended Data Fig. 6a). Together, these data revealthat substitution of threonine with alanine at the P19 residue of a cas-pase cleavage site enhances processing of ATG16L1, whereas disrup-tion of the caspase cleavage site (D299E in human, D315E in mouse)abolishes its processing. Compatible with these findings, bioinformaticsand biochemical studies have indicated that alanine is preferred overthreonine as a P19 amino acid for caspase-mediated cleavage17,18. Finally,we tested whether the T300A or T316A variants inherently decreasedprotein stability of ATG16L1. In-vitro-translated ATG16L1 showedcomparable stability between all variants in the absence of caspases,

285

289

285289284

93

345

361

345360330156

P S L S H R S - D N S R R R S V N S F P A SQ D Y A D SQ G A G N Q E V R V P S A A V Y S F D A H D G E V N A V R F S P G S R L L A T G G M D R K- - - - - - - - - - - R R R S V N S F S S S P E N A E V - P S A C A D V R V P S T A L H I F D A H D G E V N A V IC R - - - - - - - - - - - - - -- - - - - - - - - - - R RA R F T A G E IA E N S R A S ID T L K A T G Y L G Q A - - - - - - N - - - P T K I LM K F E A H E N E S H A V R W S P V E R M V A T G G A D R K

- - - - - - - - - - - - R R S V S S F P V P Q D N V D T H P G S G K E V R V P A T A L C V F D A H D G E V N A V Q F S P G S R L L A T G G M D R RP L L G H H S S D A A R R R S V S S IP V P Q D IM D T H P A S G K D V R V P T T A S Y V F D A H D G E V N A V Q F S P G S R L L A T G G M D R R- - - - - - - - - - - - R R S V S S IP V P Q D V V D T H P A S G K D V R V P T T A S Y V F D A H D G E V N A V Q F S P G S R L L A T G G M D R R

Homo sapiensMus musculus

Rattus norvegicusXenopus laevis

Danio rerioDrosophila melanogaster

Thr 300Caspase

cleavage site

Caspase

cleavage site

Thr 300

N coiled–coil

WD-repeats

N V5WT:

N V5

N V5

V5

T300A:

D299E:

T300-Cter:

ATG16L1 constructs

N

WT–

V5

T300

A–V5

D29

9E–V

5

T300

-Cte

r–V5

64

kDa

50

36

ATG16L1β

Cleaved

ATG16L1β

+ Active

caspase 3

Untreated

b

IB: Anti-V5

c

a

64

kDa

5036

3622

16

6450

0 2 5 20 0 2 5 20

T300A[TNF] ng ml–1

β

Caspase 3

Cleaved

WT

0 2 5 20

300

200

100

0

–100

–0.2

0.0

0.2

0.4

0.6

WTT300A

64

kDa

5036

362216

6450

0 2 5 20 0 2 5 20WT T316A

[TNF] ng ml–1 d

P = 0.0492

[TNF] ng ml–1

[TNF] ng ml–1

αCleaved β Cleaved α

βαCleaved β Cleaved α

caspase 3

Caspase 3

Cleavedcaspase 3

α-Tubulin

α-Tubulin

ATG16L1

ATG16L1

AT

G16L1 r

atio

(cle

aved

:full

leng

th)

AT

G1

6L

1 r

atio

(cle

aved

:fu

ll le

ng

th) P = 0.0475

WT–

V5

T300

A–V5

D29

9E–V

5

Lysa

te

Vect

or

Protein

0 2 5 20

WT

T316AP = 0.0014

P = 0.0047

P < 0.0001

64

50

36

TNF + CHX

TNF + CHX

Figure 1 | T300A variant sensitizes ATG16L1 to caspase-mediated cleavageat the D299–T300 scissile site. a, Multiple sequence alignment of ATG16L1(NP_110430.5). Protein accession numbers shown in Supplementary Table 3.b, Carboxy-terminal V5 immunoblot of full-length (68 kDa) and cleaved(36 kDa) ATG16L1b. Silver stain (bottom gels) depicts total protein. Datarepresent 3 independent experiments. IB, immunoblot. c, TNF-a-mediated

processing of human ATG16L1 in monocyte-derived macrophages. Scatterplot of immunoblot densitometry depicts a ratio of cleaved:full-lengthATG16L1 (Extended Data Fig. 2). n 5 7 donors. d, TNF-a-mediatedprocessing of murine ATG16L1 in macrophages. Immunoblot represents 5independent experiments. Scatterplot represents data pooled from 5independent experiments. WT, wild type.

RESEARCH ARTICLE

2 | N A T U R E | V O L 0 0 0 | 0 0 M O N T H 2 0 1 4


indicating that basal stability of the T300A protein is not compromised(Fig. 1b, untreated samples). We next tested stability of endogenousATG16L1 in primary murine macrophages using 35S-labelling pulse-chase. Immunoprecipitation of ATG16L1 using an antibody that pref-erentially detects the N-terminus of ATG16L1 (data not shown) showedcomparable turnover between wild-type and T316A variants of ATG16L1when chased over 2 hours (Extended Data Fig. 6b). Thus, the T300A/T316A variants do not compromise ATG16L1 protein stability underbasal conditions. Together these data demonstrate that cellular stressresulting in caspase activation is a requirement for accelerated degra-dation of the T300A/T316A variants of ATG16L1.

Caspase 3 cleaves ATG16L1We noted that human or murine ATG16L1 is cleaved in a caspase-dependent manner following induction of apoptosis by death-receptorengagement or staurosporine (Extended Data Fig. 7a, b). Given thatnumerous caspases recognize similar cleavage motifs in their substrates19,we aimed to identify the physiologically relevant caspase responsiblefor ATG16L1 processing. TNFR1 engagement by TNF-a induces type Iapoptosis whereby death receptor stimulation activates caspase 8, whichdirectly activates effector caspases 3 and/or 720. In type II apoptosis,caspase-8-mediated activation of pro-apoptotic Bcl-family proteins(for example, Bid, Bax) leads to the activation of executor caspases 3and/or 7. Short interfering RNA (siRNA)-mediated knockdown ofinitiator caspases 8 or 9 demonstrated that caspase 8 is required fordeath-receptor-, but not staurosporine-mediated cleavage of ATG16L1,whereas caspase 9 is dispensable for ATG16L1 processing (Fig. 2a, b).We further assessed ATG16L1 processing after induction of type IIapoptosis in colon carcinoma epithelial cells lacking the pro-apoptoticBcl-family member Bax. Activation of caspase 3 in Bax1/1 HCT116cells by TNF-a robustly induced ATG16L1 cleavage as measured byimmunoblotting; however, HCT116 cells lacking Bax failed to cleaveATG16L1 under these conditions (Extended Data Fig. 7c). To delin-eate the effector caspases responsible for ATG16L1 cleavage, we firstused siRNA-mediated knockdown of caspase 3 or 7 followed by TNF-aor staurosporine treatment. Knockdown of caspase 3 alone was suf-ficient to eliminate ATG16L1 cleavage following either TNF-a or staur-osporine treatment (Fig. 2c and Extended Data Fig. 7d). Induction ofapoptosis in caspase-3-deficient MCF-7 cells also showed that caspase 7activation was not sufficient to cleave ATG16L1 (Extended DataFig. 7e). Finally, we generated primary bone-marrow-derived macro-phages from Casp3-knockout mice and treated them with TNF-a orStaurosporine. Loss of Casp3 clearly abolished ATG16L1 cleavage under

both conditions, despite activation of effector caspase 7 (Fig. 2d). Thus,independent of cell type, caspase 3 is the effector caspase that directlycleaves ATG16L1, whereas caspase 8 and Bax are important upstreamfactors that promote ATG16L1 processing. Activation of the NLRP3inflammasome in bone-marrow-derived macrophages failed to cleaveATG16L1 (Extended Data Fig. 7f). These results demonstrate thatapoptotic and not inflammatory caspase activation is required forATG16L1 processing.

Defective autophagy via metabolic stressDuring the early stages of autophagosome maturation, ATG16L1 actsas an adaptor, stabilizing the interaction between ubiquitin-like proteinATG12 and the E3 ubiquitin ligase-like protein ATG5. This ‘ATG16Lcomplex’ acts in an E3-ligase-like manner to lipidate ATG8 (alsoknown as LC3 or Map1lc3a), a requirement for the elongation of thenascent autophagosomal membrane6. Nutrient deprivation (glucose oramino-acid starvation) is an established method to investigate canon-ical autophagy21–23, and has been reported to activate apoptotic caspasesin cells24–26. As the T300A variant facilitated ATG16L1 processing bycaspase 3, we asked whether this would result in decreased levels oftotal ATG16L1 during periods of nutrient starvation. Glucose-starvedmacrophages from T316A knock-in mice displayed a lower meanfluorescence intensity of ATG16L1, indicative of a ,50% decrease intotal ATG16L1 protein (Fig. 3a, b). We next determined if the SNP hasa functional consequence on starvation-induced autophagy13,14,27.Glucose or amino acid starvation followed by immunoblotting showedthat cells from T316A knock-in mice sensed metabolic stress (AMPKaphosphorylation) and initiated autophagy (emergence of LC3-II) (Fig. 3cand Extended Data Fig. 8a). By assessing the fluorescence intensity andtotal cellular area occupied by LC3-II punctae23 during glucose or aminoacid starvation, we were able to enumerate the LC3-II punctae (termedspot count) per cell. Glucose starvation resulted in a significant decreaseof cellular area occupied by lipidated LC3 in T316A knock-in macro-phages (Fig. 3d) and a small but measurable decrease in LC3-II spotcount (Extended Data Fig. 8b). Amino acid starvation similarly resultedin decreased LC3-II area in T316A macrophages (Extended Data Fig. 8c).Given the association between nutrient deprivation and induction ofapoptosis, we asked whether the observed decrease in ATG16L1 proteinand LC3-II area following glucose starvation correlated with an induc-tion of caspase activity. Either glucose or amino acid starvation inducedcaspase 3 activity which was comparable between wild-type and T316Amacrophages (Fig. 3e and Extended Data Fig. 8d), whereas treatmentwith the mTOR inhibitor rapamycin showed a similar increase in LC3-II

6450

36

362216

64

50

kDa

d

TNF

+ CHX

Staur

ospor

ine

Casp3+/+ Casp3–/–

TNF

+ CHX

Staur

ospor

ine

6450

36

22

50

36

22

16

Caspase 3

Cleaved

ATG16L1

caspase 3

Caspase 7

Cleavedcaspase 7

caspase 8 p43

Cleavedcaspase 8 p18

Cleaved

0 1 3 0 1 3

siCTL siCasp3

0 1 3

siCasp7

ATG16L15036

64

50

36

64

2216

p55p53p43p41

p18

Cleaved

caspase 8

Caspase 8

Time (h)kDa

36

2216

Caspase 7

Cleaved

caspase 7

36

2216

Caspase 3

Cleaved

caspase 3

6450

ca

0 1 3 0 1 3 0 1 3 0 1 3

siCTL siCasp8 siCTL siCasp8

5036

64

36

6450

Staurosporine TNF + CHX

kDa

0 1 3 0 1 3 0 1 3 0 1 3 Time (h)

siCTL siCasp9 siCTL siCasp9

Staurosporine TNF + CHX

Time (h)

Caspase 8

Caspase 9

ATG16L1

Cleaved caspase 950

36

64

64

36

6450

kDa

b

ATG16L1

TNF + CHX

Bone-marrow-derived macrophages

Unt

reat

ed

Unt

reat

ed



α-Tubulin

α-Tubulin


Cleaved β Cleaved α

β′ α′

β α

α-Tubulin

α-Tubulin

Figure 2 | Caspase 3 activation is required for ATG16L1 processing. a, HeLacells transfected with control siRNA (siCTL) or siCaspase 8 were stimulatedwith 2.0mM staurosporine or 20 ng ml21 TNF 1 10mg ml21 CHX. b, HeLacells transfected with control siRNA or siCaspase 9 were stimulated as ina. c, HeLa cells transfected with control, caspase-3- or caspase-7-specific

siRNAs were stimulated with 20 ng ml21 TNF 1 10mg ml21 CHX. d, Wild-type or caspase-3-knockout macrophages were stimulated with 2.0mMstaurosporine or 20 ng ml21 TNF 1 10mg ml21 CHX. Data in a–c represent2 independent experiments; data in d represent 4 independent experiments.

ARTICLE RESEARCH

0 0 M O N T H 2 0 1 4 | V O L 0 0 0 | N A T U R E | 3


area between wild-type and T316A macrophages (Extended Data Fig. 8e),and did not result in caspase 3 activation (Extended Data Fig. 8f).Finally, macrophages from Casp3-knockout mice showed significantlyincreased LC3-II area following glucose starvation in the presence ofbafilomycin A1 (Fig. 3f). These findings indicate that, in addition to theinhibition of mTOR signalling, nutrient starvation affects autophagythrough apoptotic caspase activation. Defective autophagy emerges incells with the T316A variant when caspase 3 activation is induced bymetabolic stress such as nutrient deprivation. Consistent with a loss-of-function caused by the T316A variant, macrophages lacking ATG16L1(LysMCre1Atg16l1loxp/loxp cells) showed a significant reduction of LC3-IIarea and spot count under glucose-starvation conditions (Extended DataFig. 8g, h).

We expanded our study of autophagy to human cells harbouringendogenous T300A mutations in a genetically heterogeneous back-ground by analysing Epstein–Barr virus (EBV)-transformed B-cell linesobtained from donors with Northern and Western European ancestryas part of the CEPH collection (CEU) of the International HapMapproject28 (Supplementary Table 1). We noted that GG homozygositywas observed at a frequency of 0.286 in the CEU cohort. Thus carriersfor the T300A SNP comprise a majority of the European Caucasianpopulation. Analysis of additional polymorphisms revealed SNPs inCard9 (rs10870077) and Il18rap (rs917997) were present in the major-ity of the population, whereas disease-associated polymorphisms inIrgm (also known as Irgm1; rs10065172), Il23r (rs11209026) or Lrrk2(rs11564258) were not found in the CEU cohort (Supplementary Table 2).Finally, cytogenetic analysis of selected individuals with non-risk orT300A variants of ATG16L1 showed normal karyotypes with no aneu-ploidies in any of the cell lines (data not shown).

Measuring ATG16L1 mean fluorescence intensity showed that T300AEBV-transformed B cells, in contrast to murine macrophages, haddecreased ATG16L1 levels when compared to non-risk (wild-type)controls at baseline (Fig. 4a). This is consistent with a significantlyhigher baseline caspase 3 activation in EBV-transformed B cells versusprimary murine bone-marrow-derived macrophages (,fivefold increase,compare Fig. 3e vs Fig. 4b, y axis). Similar to murine macrophages andcompatible with increased caspase 3 activation following metabolicstress, we observed a smaller LC3-II area in T300A compared to non-risk control (wild-type) cells following nutrient deprivation (Fig. 4c).Presence of early autophagosomes in these cells was independently

assessed at the ultrastructural level using transmission electron micro-scopy following bafilomycin A1 treatment. As with flow cytometryanalysis, we observed a decrease in early mature autophagosome-likestructures in T300A cells (red arrows) and increased accumulation ofmultivesicular structures (blue arrows) resembling autolysosomescompared to non-risk (wild-type) controls (Fig. 4d; graph quantifiescytosolic area occupied by autophagosomes per cell). Cumulatively, weuncovered a defect in starvation-induced autophagy conferred by theT300A variant, which is dependent on caspase 3 activation. The defectin starvation-induced autophagy induced by the SNP is conserved inmurine and human cells, and prevails in a genetically heterogeneouspopulation, underscoring a strong penetrance of this variant.

T300A/T316A variants compromise xenophagyDefective autophagy is emerging as a contributor to numerous humandiseases29,30. However, the specific association of the T300A SNP toCrohn’s disease is indicative of a unique role for autophagy in theintestinal environment. Both small and large intestines are colonizedby diverse microflora that have a critical role in mucosal health31,32. Todate, three studies have implicated the T300A variant in the invasionand survival of Salmonella typhimurium in fibroblast or epithelial celllines13,14,27. S. typhimurium predominantly colonizes the murine colonand caecum33, whereas Yersinia enterocolitica infects the ileum throughM cells into local Peyer’s patches where they invade macrophages andsubsequently enter mesenteric lymph nodes34. The T300A SNP isuniquely associated with ileal Crohn’s disease8,35, thus making Y. enter-ocolitica a physiologically relevant pathogen to investigate the effect ofthe murine T316A variant on inflammation of the small intestine.Infection of peripheral blood monocyte-derived macrophages fromhealthy donors expressing non-risk (wild-type) or T300A variants ofAtg16l1 demonstrated that the T300A variant resulted in defectiveclearance of Y. enterocolitica (Fig. 5a). We next assessed pathogen clear-ance in murine macrophages. Macrophage and dendritic cell apoptosisis known to be induced by Y. enterocolitica infection36–38; consistently,we observed caspase 3 activation in both wild-type and T316A macro-phages 6 hours after Y. enterocolitica infection (Extended Data Fig. 9a).As with human macrophages, murine T316A mutant macrophages wereunable to clear Y. enterocolitica as efficiently as wild-type cells (Fig. 5b).ATG16L1-deficient macrophages from LysMCre1Atg16l1loxp/loxp micealso showed a defect in bacterial clearance, indicating that the T316A

ATG16

L1

LC3

Mer

ge

Brig

htfie

ld

0306090

120

0 104 106

Fre

qu

en

cy

ATG16L1

intensity

WT

T316A

WT

T316A

WT

T316A

0

2

4

6

8

Mean A

TG

16L1

inte

nsity (x10

5)

P = 0.0002

– + ++

a b

d e

Glu. starve (4 h)Baf A1 (4 h)

7 μmWT

T316A– –

– + ++

Glu. starve (4 h)Baf A1 (4 h)– –

– + ++


– + ++


0.0

0.5

1.010

30

50

Mean

LC

3-I

I are

a (μ

m2)

Casp

ase 3

/7 a

ctivity

(RL

U ×

10

4)

Mean

LC

3-I

I are

a (μ

m2)

P < 0.0001

P < 0.0001

P < 0.0001

P = 0.0014

P = 0.0002

P = 0.0009

P = 0.0068

P < 0.0001

0

10

20

30

40

Casp3–/–

Casp3–/–

Casp3+/+

P = 0.018

0

2

4

6

16

6450 α-Tubulin

– + ++


– + ++

LC3-ILC3-II

kDa

WT T316A

– –– –

p-AMPKα(Thr 172)

AMPKα

6450

6450

c

f

Figure 3 | T316A variant reduces ATG16L1 levels and decreases autophagicflux upon nutrient starvation of murine macrophages. a, Intracellularstaining for LC3 and ATG16L1 following glucose starvation in the presence ofbafilomycin A1. White mask in LC3 channel illustrates punctate LC3 staining,representing LC3-II. Histogram depicts mean fluorescence intensity of 10,000cells per sample. b, Quantification of mean ATG16L1 fluorescence intensity ina, n 5 4 mice. Glu., glucose. c, Immunoblot of phosphorylated (p-) AMPKa

(Thr 172) and lipidated LC3 (LC3-II). d, Quantification of mean LC3-II area inmurine macrophages, n 5 4 mice. e, Caspase 3 activity in murine macrophagesfollowing glucose starvation, n 5 3 mice. RLU, relative light units.f, Quantification of mean LC3-II area in wild-type (Casp31/1) or Casp3-knockout (Casp32/2) macrophages, n 5 3 mice. Data in a–f represent 2independent experiments.

RESEARCH ARTICLE

4 | N A T U R E | V O L 0 0 0 | 0 0 M O N T H 2 0 1 4


mutation conferred a loss-of-function phenotype (Extended Data Fig. 9b).We further tested if the differences in bacterial load were due to analtered capacity of primary T316A mutant macrophages to clear invad-ing bacteria and observed that, whereas the T316A SNP significantlyimpaired bacterial clearance 6 hours after infection at several multipli-cities of infection (m.o.i.), Y. enterocolitica invasion was comparablebetween wild-type and T316A mutant macrophages (Extended DataFig. 9c, d), indicating that the SNP does not affect invasion of Y. enter-ocolitica in primary murine macrophages as opposed to a colon epithe-lial cell line14. Given our initial finding that Thr 300 (Thr 316 in mouse)resides at the P19 residue of a caspase-cleavage site, we asked whethercleavage of mutant ATG16L1 at this site indeed drives the increasedbacterial burden. Bone marrow progenitor cells (immortalized by expres-sion of oestrogen receptor–Hoxb8 fusion protein, ER–HoxB8) lacking

Atg16l1 were complemented with wild-type, T316A or D315E variantsof murine ATG16L1, differentiated into macrophages and infectedwith Y. enterocolitica. Cells lacking Atg16l1 (vector transduced controls)showed a significantly higher bacterial burden 6 hours after infectionwhen compared to cells expressing wild-type Atg16l1 (Fig. 5c). As withprimary murine macrophages, ER–HoxB8-immortalized macrophagesexpressing T316A mutant Atg16l1 maintained a high bacterial burden,similar to ATG16L1-deficient cells. Abolishing the caspase-3 cleavagesite by site-directed mutagenesis (D315E variant) restored bacterialclearance (Fig. 5c). These findings demonstrate that caspase-mediateddegradation of the T300A/T316A variants of ATG16L1 results indefective xenophagy by macrophages.

T316A variant enhances cytokine responseTo determine if defective bacterial clearance triggers a pro-inflammatory res-ponse by macrophages, we analysed cytokine response in Y. enterocolitica-infected cells. Levels of inflammatory cytokines TNF-a, IL-1b and IL-6were increased in cell culture media of T316A knock-in macrophagesat various time points following infection (Fig. 5d). Deletion of thecoiled–coil domain of ATG16L1 in murine macrophages has been shownto result in elevated IL-1b release following lipopolysaccharide (LPS)-endotoxin challenge9. To delineate whether elevated cytokine produc-tion was reflective of pathogen burden or enhanced sensitization ofT316A macrophages to bacterial components, we stimulated the cellswith heat-killed Y. enterocolitica (HKYe) or Toll-like receptor (TLR)and NOD2 ligands. TNF-a transcription and release were comparablebetween wild-type and T316A macrophages following stimulation withHKYe (Extended Data Fig. 10a). In contrast, transcription of Il1b (alsoknown as Il1b) was comparable, but levels of secreted IL-1b and itstarget IL-6 were elevated in T316A macrophages after treatment withHKYe (Extended Data Fig. 10b, c). Stimulation of specific TLRs showedcomparable TNF-a and IL-6 release following LPS (TLR4) or CpG-ODN (TLR9) treatment, and a modest increase of TNF-a and IL-6levels in T316A macrophages following Pam3CSK4 (TLR1/2) treat-ment (Extended Data Fig. 10d, e). NOD2 stimulation by muramyldipeptide (MDP) in the presence of LPS resulted in elevated IL-1brelease by T316A macrophages (Extended Data Fig. 10f). These dataindicate that enhanced TNF-a production in Y. enterocolitica-infectedT316A macrophages reflects increased pathogen burden, whereas ele-vated IL-1b release is compatible with a role for autophagy as a negativeregulator of inflammasome activation, consistent with previous inde-pendent observations9,39,40.

Given the relevance of Y. enterocolitica in ileitis, we next tested thephysiological effect of the T316A SNP in bacterial clearance andinflammation. On the basis of our in vitro findings, we asked whetherthe T316A mutation would result in defective pathogen clearance andenhance cytokine production in vivo. Oral ingestion of Y. enterocoli-tica resulted in specific inflammation of the ileum but not of the colon(Fig. 5e). The low level of intestinal inflammation induced by thispathogen prevented the identification of histological differences inileal inflammation between wild-type and T316A mice (ExtendedData Fig. 10g). Serum cytokine analysis identified elevated IL-1b levelsin T316A mice 48 h after Y. enterocolitica oral gavage (Extended DataFig. 10h). Measurement of cytokine transcription in the mesentericlymph nodes of infected mice showed increased mRNA levels of Tnfa, Il6,Il1a and Il1b in T316A compared to wild-type mice (Fig. 5f), consistentwith an increased bacterial burden and pro-inflammatory signalling inthe mesenteric lymph nodes of T316A mutant mice (Fig. 5g). Thus, theT300A/T316A variants confer an increased response to specific bac-terial components by macrophages and a defect in Y. enterocoliticaclearance, resulting in an elevated inflammatory cytokine response.

DiscussionIn a healthy intestine, the turnover of ATG16L1 is dependent on basalcaspase 3 activity. However, in the presence of the T300A SNP, thepersistence of apoptotic stimuli in the form of metabolic or endoplasmic

60

40

20

0

30

20

10

0

3

4

2

1

0

– + ++

a

WT

T300A


d

– –

– + ++

Glu. starve (4 h)

Baf A1 (4 h)– –– + +

+Glu. starve (4 h)

Baf A1 (4 h)– –

bP = 0.0005P = 0.0013

P = 0.0112

P = 0.0012

P < 0.0001

P = 0.00318

P = 0.0031

0.00

0.01

0.02

0.03

0.04

Ratio

AV

:cyto

so

l are

a

TEM autophagosome

quantification

c

500 nm

500 nm2 μm

2 μm

T300A

WT

T300A

WT

T300A

WTT300A

WT

Mean

LC

3-I

I are

a (μm

2)

Casp

ase 3

/7 a

ctivity

(RLU

×10

4)

Mean A

TG

16L1

inte

nsity (×

10

5)

Figure 4 | Human T300A variant reduces autophagic flux upon nutrientstarvation of genetically heterogeneous B cells. a, Quantification ofATG16L1 mean fluorescence intensity by image-based flow cytometry of10,000 cells, n 5 4 (wild-type, WT), 5 (T300A). Data represent 3 independentexperiments. b, Quantification of caspase 3 activity upon glucose starvationof 2 3 104 cells, n 5 5. Data represent 2 independent experiments.c, Quantification of mean LC3-II area of samples in a. Data represent 3independent experiments. d, Visualization of early autophagosomes bytransmission electron micrographs (TEM). Red arrows, early autophagosomes,characterized by vesicular structures with a double-membrane. Blue arrows,multivesicular lysosomal structures, identified by degraded intravesicularcargo. Micrographs representative of 10–12 images per donor, n 5 4 (WT),n 5 3 (T300A). A total of 29 (WT) and 26 (T300A) images containingautophagosomes was analysed. Histogram quantifies autophagic vesicles as aratio of autophagosome:cytosolic area.

ARTICLE RESEARCH

0 0 M O N T H 2 0 1 4 | V O L 0 0 0 | N A T U R E | 5


reticulum stress, death receptor activation or pathogen infection sig-nificantly enhances ATG16L1 cleavage, thereby diminishing basal autop-hagy. This in turn triggers inflammatory cytokine production, therebygenerating a chronic inflammatory state in the intestine (illustrated inFig. 6). Recent work2 implicates autophagy as a mechanism to preventapoptosis upon endoplasmic reticulum stress. Our study now in turnstrongly supports a role for apoptotic caspases in the regulation of autop-hagy through degradation of ATG16L1.

Our findings uncover a functional consequence of the T300A vari-ant, namely its sensitization to caspase-3-mediated cleavage. This is acritical initial step leading to defects observed in pathogen handlingand stress response by cells of the intestine. T300A is a common variantof ATG16L1, and its high prevalence in independent populations sup-ports the observations that the T300A genotype does not confer com-plete loss-of-function of autophagy. Our in vitro and in vivo findingsare consistent with the phenotypes associated with ATG16L1 loss-of-function2,9–12,16,41, and consistent with our understanding that agenetically complex disease such as Crohn’s disease is influenced bymultiple risk loci, each contributing only partly to the disease pheno-type31. Importantly, the observation that the T300A variant increasessusceptibility to proteolytic degradation opens up the possibility thattherapeutic inhibition of pathways that lead to caspase 3 activationmay restore autophagy and gut homeostasis in part by stabilizingATG16L1.

METHODS SUMMARYDonors. Healthy human donors were genotyped for the T300A variant (rs2241880).Written informed consent was obtained from study participants and the studyprotocols were reviewed and approved by the Western Institutional Review Board.Mice. Atg16l1 T316A knock-in mice were generated using C57BL/6 embryonicstem cells (Supplementary Fig. 5a). Deletion of Atg16l1 was generated by crossingLysMCre1 (macrophage-specific, described in ref. 42) with Atg16l1loxp/loxp mice(Extended Data Fig. 3c, d). Caspase 3-knockout mice43 were obtained from JacksonLaboratories. All in vivo experiments were performed using age-matched mice andlittermate controls. All mice were bred into the C57BL/6 background, and allprotocols were approved by the Genentech Institutional Animal Care and UseCommittee.Macrophage ATG16L1 cleavage assay. Macrophages were stimulated with indi-cated doses of recombinant TNF-a (Peprotech) 1 10mg ml21 cycloheximide (CHX,

macrophages reconstituted with

ATG16L1 variants

c.f

.u.

per

1.2

5 ×

10

5 c

ells

c.f

.u.

per

1.2

5 ×

10

5 c

ells

c.f

.u.

per

2.5

0 ×

10

5 c

ells

[TN

F] p

g m

l–1

[IL

-1β]

pg

ml–

1

[IL

-6] p

g m

l–1

Human peripheral blood-derived

macrophages

Thioglycollate-elicited

macrophages

101

0

102

103

104

102

103

104

105

102

101

103

104

105

102

101

103

104

105

VectorWT–V5T316A–V5

WTT316A

WTT316A

WT

T316A

WTT316A

WTT300A D315E–V5

P = 0.0101

P = 0.0089

P = 0.0089

P = 0.0103

f

a c

P = 0.0026 P = 0.0009

0 3 6 90

200

400

600

800

1,000 P = 0.0056P = 0.0011

P = 0.0056

0 3 6 90

200

400

600P = 0.0074

P = 0.0132P = 0.0055

Time (h) Time (h) 0

c.f

.u.

per

gra

m t

issu

e

P = 0.0208

Mesenteric lymph node

eb

d g

0.00

0.02

0.04

0.06

0.08

0.000

0.002

0.004

0.006

0.008

0.010P = 0.0275 P = 0.0197 P = 0.0105P = 0.0417

0.03

0.02

0.01

0.00

Tn

fa/G

ap

dh

exp

ressio

n

– + – + – + – + – + – +– + – +

Colon

Ileum

PBS

Y. enterocolitica

oral gavage (48 h)

Y. enterocolitica oral gavage (48 h)

0 3 6 90

2,000

4,000

6,000 P = 0.0102P = 0.0138

Time (h)

Il6

/Gap

dh

exp

ressio

n

Il1

a/G

ap

dh

exp

ressio

n

0.3

0.2

0.1

0.0

Il1

b/G

ap

dh

exp

ressio

n

200 μm

ATG16L1-deficient immortalized

200 μm

100 μm 100 μm

Y. enterocolitica

Figure 5 | T300A/T316A variants of ATG16L1 confer defective clearance ofY. enterocolitica and elevated cytokine production. a–c, Y. enterocoliticacolony forming units (c.f.u.) in human and murine macrophages infected for6 h (20 m.o.i.). Data in a are pooled from 4 independent cohorts of 3 or 4 donorsper genotype, total n 5 16 (non-risk, WT), 15 (T300A, risk). Data in brepresent 3 independent experiments, n 5 4 mice. Data in c represent 3independent experiments, n 5 6. d, TNF-a, IL-1b and IL-6 levels in culturemedia of WT and T316A macrophages infected with 20 m.o.i. Y. enterocolitica.Data represent 2 independent experiments, n 5 4 mice. e, Ileum and colon

histology (haematoxylin and eosin staining) of WT mice. Images arerepresentative of 3 (PBS) and 8 (Y. enterocolitica) treated male mice.f, Measurement of cytokine mRNA transcripts in mesenteric lymph nodes.Data are pooled from 2 independent experiments of 3 mice (PBS) and 7 mice(Y. enterocolitica) per experiment. Transcripts are normalized to Gapdh(22DCt). g, Y. enterocolitica c.f.u. in mesenteric lymph nodes of WT and T316Aknock-in mice 48 h following oral gavage. Data represent 3 independentexperiments, n 5 9 male mice.

T300A

Caspase 3 activation

Autophagy

Stress

response

Xenophagy

Pathogen

clearance

Wild-type ATG16L1

Healthy Crohn’s disease

Sustained cellular stress

Pathogen expansion

Inflammation

T300

Cellular/metabolic stress

Infection

Death-receptor activation

Caspase activators

ATG16L1 cleavage

Defective autophagy

and xenophagy

Disease progression

Figure 6 | Mechanism of T300A variant contribution to Crohn’s diseasepathogenesis. An inflamed intestinal environment induces cellular stressand caspase activation, thereby enhancing caspase-3-mediated cleavage ofthe T300A variant of ATG16L1. This results in defective stress-inducedautophagy and bacterial clearance by xenophagy, establishing a chronicinflammatory state.

RESEARCH ARTICLE

6 | N A T U R E | V O L 0 0 0 | 0 0 M O N T H 2 0 1 4


Sigma) or 2.0mM staurosporine to induce ATG16L1 processing. Full-length andcleavage products were visualized by immunoblotting.Nutrient starvation assays. All cells were cultured for 4 hours in glucose-deficientDMEM 1 GlutaMAX without serum (glucose starvation), or in amino-acid defi-cient RPMI without GlutaMAX or serum (amino acid starvation). Bafilomycin A1was added at 400 nM where indicated (Sigma).Bacterial protection assay. Yersinia spp. enterocolitica (ATCC 27729) grown at25 uC was added at an m.o.i. of 20 and cells were spin-infected at ,300g for10 min. After further incubation for 30 min, cells were washed in macrophageculture media containing 100mg ml21 gentamycin (Gibco). Cells were lysed in0.1% TX-100 1 PBS and plated for colony formation.In vivo Y. enterocolitica studies. Mice were fasted overnight before infection.2 3 107 colony forming units (c.f.u.) Y. enterocolitica was administered directlyfrom frozen stock by oral gavage. c.f.u. in mesenteric lymph nodes were analysed48 h after infection. For cytokine analysis, bacteria was grown overnight at 25 uCon tryptic soy agar with sheep blood, and 2 3 109 c.f.u. bacteria was administered.

Online Content Any additional Methods, Extended Data display items and SourceData are available in the online version of the paper; references unique to thesesections appear only in the online paper.

Received 30 August 2013; accepted 17 January 2014.

Published online 19 February 2014.

1. Manolio, T. A. Bringing genome-wide association findings into clinical use. NatureRev. Genet. 14, 549–558 (2013).

2. Adolph, T. E. et al. Paneth cells as a site of origin for intestinal inflammation. Nature503, 272–276 (2013).

3. Abraham, C. & Cho, J. H. Inflammatory bowel disease. N. Engl. J. Med. 361,2066–2078 (2009).

4. Gardet, A. & Xavier, R. J. Common alleles that influence autophagy and the risk forinflammatory bowel disease. Curr. Opin. Immunol. 24, 522–529 (2012).

5. Kaser, A. et al. XBP1 links ER stress to intestinal inflammation and confers geneticrisk for human inflammatory bowel disease. Cell 134, 743–756 (2008).

6. He, C. & Klionsky, D. J. Regulation mechanisms and signaling pathways ofautophagy. Annu. Rev. Genet. 43, 67–93 (2009).

7. Baxt, L. A., Garza-Mayers, A. C.& Goldberg,M.B.Bacterial subversion ofhost innateimmune pathways. Science 340, 697–701 (2013).

8. Hampe, J. et al. A genome-wide association scan of nonsynonymous SNPsidentifies a susceptibility variant for Crohn disease in ATG16L1. Nature Genet. 39,207–211 (2007).

9. Saitoh, T. et al. Loss of the autophagy protein Atg16L1 enhances endotoxin-induced IL-1b production. Nature 456, 264–268 (2008).

10. Cadwell, K. et al. A key role for autophagy and the autophagy gene Atg16l1 inmouse and human intestinal Paneth cells. Nature 456, 259–263 (2008).

11. Cadwell, K. et al. Virus-plus-susceptibility gene interaction determines Crohn’sdisease gene Atg16L1 phenotypes in intestine. Cell 141, 1135–1145 (2010).

12. Cooney, R. et al. NOD2 stimulation induces autophagy in dendritic cellsinfluencing bacterial handling and antigen presentation. Nature Med. 16, 90–97(2010).

13. Fujita, N. et al. Differential involvement of Atg16L1 in Crohn disease and canonicalautophagy: analysis of the organization of the Atg16L1 complex in fibroblasts.J. Biol. Chem. 284, 32602–32609 (2009).

14. Messer, J. S. et al. The Crohn’s disease: associated ATG16L1 variant andSalmonella invasion. BMJ Open 3, e002790 (2013).

15. Sorbara, M. T. et al. The protein ATG16L1 suppresses inflammatory cytokinesinduced by the intracellular sensors Nod1 and Nod2 in an autophagy-independent manner. Immunity 39, 858–873 (2013).

16. Conway, K. L. et al. Atg16l1 is required for autophagy in intestinal epithelial cellsand protection of mice from Salmonella infection. Gastroenterology 145,1347–1357 (2013).

17. Stennicke, H. R., Renatus, M., Meldal, M. & Salvesen, G. S. Internally quenchedfluorescent peptide substrates disclose the subsite preferences of humancaspases 1, 3, 6, 7 and 8. Biochem. J. 350, 563–568 (2000).

18. Rawlings, N. D., Barrett, A. J. & Bateman, A. MEROPS: the database of proteolyticenzymes, their substrates and inhibitors. Nucleic Acids Res. 40, D343–D350(2012).

19. McStay,G.P., Salvesen,G.S.&Green,D.R.Overlappingcleavagemotif selectivityofcaspases: implications for analysis of apoptotic pathways. Cell Death Differ. 15,322–331 (2008).

20. Jost, P. J. et al. XIAP discriminates between type I and type II FAS-inducedapoptosis. Nature 460, 1035–1039 (2009).

21. McAlpine, F., Williamson, L. E., Tooze, S. A. & Chan, E. Y. Regulation of nutrient-sensitive autophagy by uncoordinated 51-like kinases 1 and 2. Autophagy 9,361–373 (2013).

22. Mizushima, N., Yoshimori, T. & Levine, B. Methods in mammalian autophagyresearch. Cell 140, 313–326 (2010).

23. Klionsky, D. J. et al. Guidelines for the use and interpretation of assays formonitoring autophagy. Autophagy 8, 445–544 (2012).

24. Bialik, S. et al. The mitochondrial apoptotic pathway is activated by serum andglucose deprivation in cardiac myocytes. Circ. Res. 85, 403–414 (1999).

25. Caro-Maldonado, A. et al. Glucose deprivation induces an atypical form ofapoptosis mediated by caspase-8 in Bax-, Bak-deficient cells. Cell Death Differ. 17,1335–1344 (2010).

26. Altman, B. J. & Rathmell, J. C. Metabolic stress in autophagy and cell deathpathways. Cold Spring Harb. Perspect. Biol. 4, a008763 (2012).

27. Kuballa, P., Huett, A., Rioux, J. D., Daly, M. J. & Xavier, R. J. Impaired autophagy of anintracellular pathogen induced by a Crohn’s disease associated ATG16L1 variant.PLoS ONE 3, e3391 (2008).

28. Altshuler, D. M. et al. Integrating common and rare genetic variation in diversehuman populations. Nature 467, 52–58 (2010).

29. Mizushima, N., Levine, B., Cuervo, A. M. & Klionsky, D. J. Autophagy fights diseasethrough cellular self-digestion. Nature 451, 1069–1075 (2008).

30. Choi, A. M., Ryter, S. W. & Levine, B. Autophagy in human health and disease.N. Engl. J. Med. 368, 651–662 (2013).

31. Jostins, L. et al. Host–microbe interactions have shaped the genetic architecture ofinflammatory bowel disease. Nature 491, 119–124 (2012).

32. Brown, E. M., Sadarangani, M. & Finlay, B. B. The role of the immune system ingoverning host-microbe interactions in the intestine. Nature Immunol. 14,660–667 (2013).

33. Grassl, G. A., Valdez, Y., Bergstrom, K. S., Vallance, B. A. & Finlay, B. B. Chronicenteric Salmonella infection in mice leads to severe and persistent intestinalfibrosis. Gastroenterology 134, 768–780 (2008).

34. Dube, P. H., Revell, P. A., Chaplin, D. D., Lorenz, R. G. & Miller, V. L. A role for IL-1a ininducing pathologic inflammation during bacterial infection. Proc. Natl Acad. Sci.USA 98, 10880–10885 (2001).

35. Prescott, N. J. et al. A nonsynonymous SNP in ATG16L1 predisposes to ilealCrohn’s disease and is independent of CARD15 and IBD5. Gastroenterology 132,1665–1671 (2007).

36. Mills, S. D. et al. Yersinia enterocolitica induces apoptosis in macrophages by aprocess requiring functional type III secretion and translocation mechanisms andinvolving YopP, presumably acting as an effector protein. Proc. Natl Acad. Sci. USA94, 12638–12643 (1997).

37. Erfurth, S. E. et al. Yersinia enterocolitica induces apoptosis and inhibits surfacemolecule expression and cytokine production in murine dendritic cells. Infect.Immun. 72, 7045–7054 (2004).

38. Denecker, G. et al. Yersinia enterocolitica YopP-induced apoptosis of macrophagesinvolves the apoptotic signaling cascade upstream of bid. J. Biol. Chem. 276,19706–19714 (2001).

39. Nakahira, K. et al. Autophagy proteins regulate innate immune responses byinhibiting the release of mitochondrial DNA mediated by the NALP3inflammasome. Nature Immunol. 12, 222–230 (2011).

40. Shi, C. S. et al. Activation of autophagy by inflammatory signals limits IL-1bproduction by targeting ubiquitinated inflammasomes for destruction. NatureImmunol. 13, 255–263 (2012).

41. Marchiando, A. M. et al. A deficiency in the autophagy gene Atg16L1enhances resistance to enteric bacterial infection. Cell Host Microbe 14, 216–224(2013).

42. Clausen, B. E., Burkhardt, C., Reith, W., Renkawitz, R. & Forster, I. Conditional genetargeting in macrophages and granulocytes using LysMcre mice. Transgenic Res.8, 265–277 (1999).

43. Kuida, K. et al. Decreased apoptosis in the brain and premature lethality in CPP32-deficient mice. Nature 384, 368–372 (1996).

Supplementary Information is available in the online version of the paper.

Acknowledgements The authors would like to thank M. Zepeda for coordinatinghuman donors, R. A. Flavell for providing Casp3-knockout mice, J. E. Cupp, W. Ortmann,J. Borneo, J. Ruan, J. Ting and L. Rangell for technical assistance, D. Holmes,N. Kayagaki, C. J. Spooner, M. E. Keir, A. Ashkenazi and T. W. Behrens for criticalevaluation of the manuscript.

Author Contributions A.M. and M.v.L.C. conceptualized the study and designedexperiments; A.M. and Y.L. conducted experiments; I.P. and J.D. performed in vivoadministration of Y. enterocolitica; M.R. and A.K.K. performed electron microscopy; R.N.performed 35S pulse-chase assays on ATG16L1; M.R.-G. designed the T316A knock-inconstruct and coordinated generation of the mutant mouse; L.D. performedhistological analysis; R.R.G. provided EBV-transformed cells, guided HapMap analysisand discussed the study; A.M. and M.v.L.C. wrote the manuscript.

Author Information Reprints and permissions information is available atwww.nature.com/reprints. The authors declare competing financial interests: detailsare available in the online version of the paper. Readers are welcome to comment onthe online version of the paper. Correspondence and requests for materials should beaddressed to M.v.L.C. ([email protected]).

ARTICLE RESEARCH

0 0 M O N T H 2 0 1 4 | V O L 0 0 0 | N A T U R E | 7




www.nature.com/reprints


mailto:[email protected]

METHODSAntibodies and reagents. ATG16L1 was detected using mouse monoclonal(M150-3, immunoblotting) or rabbit polyclonal (PM040, image-based flow cyto-metry) antibodies from MBL international. LC3 was detected with a mouse mono-clonal antibody (M152-3, MBL international). V5 epitope was detected usinganti-V5 mouse monoclonal antibody (M167-3, MBL international). Fas activationwas induced with anti-Fas (CD95) mouse monoclonal antibody, clone CH11 (SY-001,MBL international). Anti-caspase 3 (catalogue no. 9662), 7 (catalogue no. 9492), 8(catalogue no. 8592), 9 (catalogue no. 9509), anti-phospho-AMPKa (T172, cata-logue no. 2535), anti-AMPKa (catalogue no. 2793), anti-PARP (catalogue no.9546) and horseradish peroxidase-conjugated anti-a-tubulin (catalogue no.12351) antibodies were obtained from Cell Signaling. Inflammasome activationwas confirmed by immunoblotting for active caspase 1 (clone 4B4, Genentech,Inc.). Recombinant active human caspase 1, 3, 6 and 7 were used for in vitrocleavage assays (MBL international). Pan-caspase inhibitor zVAD-fmk was usedat a final concentration of 20mM (Promega). Human and murine M-CSF, murineGM-CSF and TNF-a were obtained from Peprotech, Inc. Cycloheximide (CHX),Staurosporine,b-oestradiol and 4-hydroxytamoxifen (4-OHT) were obtained fromSigma. G418 was obtained from Clontech.Mice. The construct for targeting the C57BL/6 ATG16L1 locus in embryonicstem cells was made using a combination of recombineering as well as standardmolecular cloning techniques. The ATG16L1 targeting vector was linearized withNotI, and C57BL/6 embryonic stem cells were targeted using standard methods(electroporation, G418 positive selection and ganciclovir negative selection). Thepositive clones were identified using PCR strategy described in Extended DataFig. 3. The T316A knock-in region was sequenced to ensure integration at theappropriate site. TaqMan analysis was subsequently performed to ensure singlecopy of integration. The targeted embryonic stem cells were transfected with a Flpplasmid to remove neomycin resistance cassette. T316A knock-in embryonicstem cells were then injected into blastocysts using standard techniques and germ-line transmission was obtained after crossing resulting chimaeras with C57BL/6females. Further breeding into C57BL/6 was performed for the generation ofsufficient numbers of mice for this study. Heterozygous mice were interbred toobtain homozygous mice to get both wild-type and knock-in mice. The followingprimers were used for genotyping and confirmation of the knock-in sequence:forward 39 AGGAGACGCTCTGTCTCTTC; reverse 59 ACCTCCATCAGGAACTTCAC. Deletion of Atg16l1 was generated by crossing LysMCre1 mice (macrophage-specific, described in ref. 42) with Atg16l1loxp/loxp mice (Extended Data Fig. 3c, d).Caspase 3-knockout mice43 were obtained from Jackson Laboratories. All in vivoexperiments were performed using age-matched mice and littermate controls. Allmice were bred into the C57BL/6 background, and all protocols were approved bythe Genentech Institutional Animal Care and Use Committee.EBV-transformed B-cell lines. EBV-transformed B-cell lines from the CEPHcollection (CEU) of the International HapMap project were obtained from theCoriell Cell Repository (http://www.ccr.coriell.org). Cell lines were cultured inRPMI 1 15% FBS 1 2 mM L-glutamine 1 antibiotics. The following cell lines wereused (ATG16L1 genotype; A 5 non-risk allele; G 5 risk allele): NA07031 (GG),NA07037 (GG), NA11893 (GG), NA11894 (GG), NA11917 (AA), NA11919 (GG),NA12272 (AA), NA12282 (AA), NA12400 (AA), NA12489 (AA).Bacteria. Yersinia spp. enterocolitica was obtained from ATCC (strain 27729) andmaintained at 25 uC on tryptic soy agar II containing sheep blood (BD Biosciences).Human macrophage culture. Donor genotype for the A . G variant in Atg16l1was determined using a TaqMan SNP Genotyping Assay for SNP ID rs3828309(ref. 44) (Life Technologies, catalogue no. 4351379). Primary human macrophageswere differentiated from peripheral blood monocytes. Leukocytes were separatedfrom total blood via a Ficoll gradient (GE healthcare) using Leucosep tubes (FisherScientific). Monocytes were enriched using negative selection (monocyte isolationkit, Miltenyi Biotec) and purity confirmed by flow cytometry analysis of CD141

cells. Monocytes were plated at 0.5 3 106 cells ml21 in macrophage culture media(RPMI1640 1 15% FBS, GlutaMAX, non-essential amino acids, sodium pyruvate,penicillin/streptomycin, all from Gibco) containing 50 ng ml21 recombinant humanM-CSF (Peprotech). Differentiation was induced for 5–6 days before experimentation.Murine macrophage culture. Thioglycollate-elicited macrophages were obtainedby intraperitoneal administration of 1.0 ml thioglycollate suspension for 4 daysfollowed by peritoneal lavage. Cells were grown in macrophage culture media. Bonemarrow-derived macrophages were differentiated from total bone marrow flushedout of femurs, grown in macrophage culture medium containing 50 ng ml21

recombinant murine M-CSF (Peprotech). Macrophage differentiation was inducedfor 5–6 days before experimentation.Immortalized macrophage progenitors. Total bone marrow obtained fromLysMCre1Atg16l1loxp/loxp mice was subjected to lineage depletion to obtain bonemarrow progenitors (mouse lineage depletion kit, Miltenyi Biotec). These cellswere immortalized using retroviral transduction with ER–HoxB8, provided by

Y. Qu (method described in ref. 45). Atg16l1 deficient cells were reconstitutedwith retroviral vectors (MSCV-IRES-hCD4, provided by C. Spooner) containingwild-type, T316A or D315E variants of murine ATG16L1 (NP_001192321.1)with a C-terminal V5 tag. Retrovirus was generated using Lipofectamine 2000CD (Invitrogen) transfection of Plat-E packaging cells (Cell Biolabs). 72 h afterretroviral transduction, cells were stained for human CD4 (hCD4-FITC, eBioscience)and sorted by magnetic separation (anti-FITC microbeads, Miltenyi Biotec).Polyclonal populations of reconstituted bone marrow progenitors were maintainedin macrophage culture media 1 20 ng ml21 murine GM-CSF 1 2mMb-oestradiol1 200 nM 4-OHT 1 200mg ml21 G418. Macrophage differentiation was inducedby removal of b-oestradiol, 4-OHT and G418 for 7–10 days.HeLa cell culture and transfection, MCF-7 cell culture. HeLa and MCF-7 cellswere obtained from an in-house cell-line banking program (gCell), where celllines are routinely tested for mycoplasma contamination. Cells were grown inDMEM (high glucose) 1 10% FBS. HeLa cells were transiently transfected withpcDNA6.1 vectors containing wild-type, T300A or D299E variants of human Atg16l1(isoform 1, NM_030803.6) with a C-terminal V5 tag by using Lipofectamine 2000CD. siRNA knockdown of caspase genes was done using Lipofectamine RNAiMAX(Invitrogen); siRNA sequences are provided in Supplementary Table 5.HCT116 cells. Colon carcinoma cell line HCT116 of Bax1/1 and Bax2/2 geno-types were obtained from Horizon, Inc. and grown in DMEM (high glucose) 1

10% FBS.RAW264.7 cells. Cells were obtained from an in-house cell-line banking pro-gram (gCell), where cell lines are routinely tested for mycoplasma contamination.Cells were grown in RPMI 1 15% FBS 1 GlutaMAX 1 non-essential amino acids1 sodium pyruvate 1 penicillin, streptomycin).Cellular ATG16L1 cleavage assays. ATG16L1 cleavage was induced by caspase 3activation in culture. (1) Cells were treated with recombinant TNF-a at indicateddoses and time points in the presence of 10 mg ml21 CHX. Where no time point isindicated, cells were treated for 3 h. (2) 2.0mM staurosporine was added for indi-cated time points. (3) 0.5mg ml21 anti-Fas agonist antibody (clone CH11) forindicated time points. Cells were directly lysed in 13 SDS loading buffer contain-ing 2% b-mercaptoethanol and reduced by boiling. Caspase 3 activation andATG16L1 cleavage were assessed by immunoblot. ImageJ was used for densito-metry analysis, and total band intensity (area under curve) of cleaved ATG16L1aand b was divided by total intensity of full-length ATG16L1a and b to obtaincleaved:full-length ATG16L1 ratio (arbitrary units). Densitometry analysis wasgated on samples that demonstrated TNF-dependent caspase 3 cleavage.Inflammasome activation. Bone marrow macrophages were primed with100 ng ml21 ultrapure LPS (upLPS, Escherichia coli 0111:B4, InvivoGen) over-night, followed by treatment with 5 mM ATP or 5mg ml21 Nigericin (both fromInvivoGen) for 1 h in RPMI1640 1 0.5%FBS. Supernatant was collected andcleaved caspase 1 detected by immunoblotting.In vitro ATG16L1 cleavage. In vitro translation of human ATG16L1 was per-formed using a TNT Quick Coupled Transcription/Translation kit (Promega).pcDNA6.1 constructs of wild-type, T300A, D299E and T300-C-terminal trun-cation variants of human Atg16l1 with C-terminal V5 tag were translated in 50 mlvolume. 2-ml samples were incubated with 2 ml recombinant active caspase 1, 3, 6or 7 (1 Uml21) in 4 ml of 23 caspase activity buffer 1 100 mM DTT (MBLInternational.) for a final volume of 8ml. Samples were incubated at 37 uC for1 h to induce ATG16L1 cleavage. Samples were reduced by addition of 100ml 43

SDS loading buffer and a final concentration of 50 mM DTT and boiled for 5 min.ATG16L1 cleavage was detected by running reduced samples on a 4–20% SDSPAGE gel and transfer to nitrocellulose membrane, followed by immunoblottingfor V5 tag. Total protein was determined by silver stain (Silver Quest silverstaining kit, Invitrogen).Nutrient-deprivation-induced autophagy. Bone marrow-derived macrophageswere grown in RPMI1640 lacking amino acids or GlutaMAX, without serum, orDMEM plus GlutaMAX lacking glucose and serum for 4 h. Lysosomal degrada-tion was inhibited by addition of 400 nM bafilomycin A1 for 4 h to accumulateLC3-II. Cells were lysed directly in 13 SDS lysis buffer. LC3-lipidation andAMPKa phosphorylation were assessed by immunoblotting.Image-based flow cytometry. Bone marrow-derived macrophages were grownon low attachment plates (non-tissue culture treated plates, Thermo Scientific).EBV transformed cells were grown in RPMI 1 15% FBS in round bottom 96-wellplates. Cells were starved of glucose or amino acid as above and macrophagesrapidly detached by incubation in 103 TrypLE Select (Gibco) for 2–5 min. Followingcentrifugation, cells were fixed in Cytofix/Cytoperm buffer (BD Biosciences) for10 min at 25 uC. Cells were then blocked in MaxBlock buffer (Active Motif) at 4 uCovernight, followed by staining for ATG16L1 and LC3 at 1:100 dilution in 13

Perm/Wash buffer for 1 h at 4 uC. Secondary antibodies were added at 1:1,000dilution for 30 min at 25 uC to detect ATG16L1 (donkey anti-rabbit IgG conju-gated to Alexa488) and LC3 (goat anti-mouse IgG2a conjugated to Alexa647).

RESEARCH ARTICLE


http://www.ccr.coriell.org

Cells were resuspended in 70ml PBS following staining and analysed using theAmnis ImageStream X with the 488 nm laser power set to 20 mW, and 761 nmlaser power set to 150 mW. 10,000 events were collected with a lower size limit of50mm. Single cells in focus were used for downstream analysis. LC3-II area wascalculated by generating a mask for channel 11 where minimum pixel intensity wasempirically established to specifically mask bright, punctate signal. The cellulararea occupied by this mask was obtained and the mean area per sample was usedfor data analysis. Protocols measuring LC3 bright detail intensity and spot countare detailed elsewhere46.Caspase 3/7 activity assay. 2 3 104 cells per well were plated in an opaque 96-wellplate in a culture volume of 50 ml. Following indicated treatments, 50 ml of caspase3/7 activity assay buffer was added directly to wells (caspase-Glo 3/7 assay, Promega).Plates were shaken at 500 r.p.m. for 30 s and incubated for 30 min before measuringluminescence. Caspase 3/7 activity was measured as relative light units (RLU).Lysozyme staining. Immunohistochemistry (IHC) was performed on 4-mm thickformalin-fixed, paraffin-embedded tissue sections mounted on glass slides. AllIHC steps were carried out on the Ventana Discovery XT automated platform(Ventana Medical Systems; Tucson, AZ). Sections were treated with mild CellConditioner 1 and then incubated in the pre-diluted primary antibody, anti-Lysozyme (Rabbit polyclonal, catalogue no. 760-2656, Ventana Medical Systems;Tucson, AZ), for 4 min at 37 uC. Specifically bound primary antibody was detectedby incubating sections in OmniMap anti-Rabbit–horseradish peroxidase (VentanaMedical Systems; Tucson, AZ) for 16 min, followed by ChromoMap DAB (VentanaMedical Systems; Tucson, AZ). The sections were counterstained with haematox-ylin II (Ventana Medical Systems; Tucson, AZ), dehydrated and coverslipped.Transmission electron microscopy. All samples were treated with 400 nM bafi-lomycin A1 for 4 h before fixation. Cells were fixed in modified Karnovsky’sfixative (2% paraformaldehyde and 2.5% glutaraldehyde in 0.1 M sodium caco-dylate buffer, pH 7.2) and then post-fixed in 1% aqueous osmium tetroxide for 2 hfollowed by overnight incubation in 0.5% uranyl acetate at 4 uC. The samples werethen dehydrated through a series of ethanol (50%, 70%, 90%, 95%, 100%) fol-lowed by propylene oxide (each step was for 15 min) and embedded in Eponate12 (Ted Pella). Ultrathin sections (80 nm) were cut with an Ultracut microtome(Leica), stained with 0.2% lead citrate and examined in a JEOL JEM-1400 trans-mission electron microscope (TEM) at 120 kV. Digital imaged were captured witha GATAN Ultrascan 1000 CCD camera. Quantification of autophagic vacuolestructures was performed manually in Adobe Photoshop CS6 by calculating thefollowing cellular pixel areas: total cellular area (cell membrane), nucleus area(nuclear membrane), total area occupied by autophagosomes (autophagosomesidentified as structures similar to those depicted in Fig. 4d, red arrows). Scale barlengths were measured in pixels and back-calculated to nm. This was used toconvert pixel areas to nm2 measurements. Cytosolic area was generated by sub-tracting nuclear area from total cellular area. Finally, a ratio of cellular areaoccupied by autophagic vacuoles was generated by dividing total autophagicvacuole area by cytosolic area. This normalized the cellular space occupied byautophagosome structures to cytosolic space, thereby eliminating variability con-ferred by presence of nucleus. Each point in the scatterplot of Fig. 4d represents anindividual cell.Cytogenetics. Metaphase spreads were generated and karyotype analysis performedon contract at the Van Andel Research Institute, Grand Rapids, Michigan, USA.Bacterial infection and gentamycin protection assays. Macrophages from indi-cated sources were grown in 24- or 48-well plates to a density of 2.5 3 105 or1.25 3 105 cells per well, respectively. Peripheral blood monocyte-derived macro-phages from human donors were primed with 100 ng ml21 upLPS for 18 h inantibiotic-free macrophage culture media. Y. enterocolitica was directly thawedfrom frozen stock and added at indicated m.o.i. in antibiotic-free macrophageculture media and cells were spin-infected at ,300g for 10 min at 25 uC. Infectionwas continued for 30 min at 37 uC, followed by extensive washing of bacteria with100mg ml21 gentamycin sulphate (Gibco). Bacterial clearance assay was initiatedby continued culture for indicated time points in 100mg ml21 gentamycin sul-phate. Media was collected for cytokine ELISA, and cells were washed twice withPBS followed by lysis and spread onto TSA II agar containing sheep blood (BDbiosciences) and bacterial c.f.u. counted after 48 h of culture at 25 uC. Cells in 24-well plates were lysed in 1.0 ml 0.1% Triton-X 100 in PBS, and 48-well cultureswere lysed in 0.5 ml 0.1% Triton-X 100 in PBS. c.f.u. was calculated and depictedas log10 c.f.u. per number of cells plated.TLR and NOD2 stimulation. 0.5 3 106 thioglycollate-elicited macrophages wereplated in 300ml macrophage media and stimulated with 100 ng ml21 ultra-pureLPS from E. coli 0111:B4 (InvivoGen, tlrl-3pelps), 100 ng ml21 Pam3CSK4(InvivoGen, tlrl-pms), 10 mg ml21 CpG-ODN 1826 (InvivoGen, tlrl-1826) for24 h. For NOD2 stimulation, macrophages were treated with 10 mg ml21 MDP(InvivoGen, tlrl-mdp) for 24 h in the presence or absence of 100 ng ml21 ultra-pure LPS. Media was collected and ELISA for TNF-a (R&D Systems), IL-6 (BD

Biosciences) and IL-1b (R&D Systems) performed. All cytokine concentrationswere normalized to macrophage protein in untreated samples quantified usingstandard bicinchoninic acid protein assay (Pierce).In vivo bacterial infection. Age- and sex-matched mice (7–10 weeks old) ofindicated genotypes were fasted overnight and inoculated orally with 2 3

107 c.f.u. Y. enterocolitica obtained directly from frozen stock in 300ml PBS. At48 h post-infection, mice were euthanized and mesenteric lymph nodes, smalland large intestines were collected for downstream analysis. Entire small intestineand colon were rolled and fixed in Formalin overnight, followed by storage in 70%ethanol before paraffin embedding and processing for haematoxylin and eosinstaining. All mesenteric lymph nodes were weighed before homogenization in1 ml 0.1% Triton X-100 1 PBS. Homogenates were serially diluted in TSA IIplates containing defibrinated sheep blood for 48 h at 25 uC, and bacterial coloniescounted. c.f.u. counts were normalized to organ weight. For cytokine analysis,mice were inoculated orally with 2 3 109 c.f.u. Y. enterocolitica from stock grownovernight at 25 uC on TSA II plates containing sheep blood. Serum was collectedand stored directly at 280 uC before cytokine measurement by ELISA. IL-1bELISA was performed by following manufacturer’s instructions (R&D Systems).Mesenteric lymph nodes were stored in RNAlater (Qiagen) at room temperatureovernight before RNA isolation in RLT buffer (Qiagen). Following generation ofcDNA by reverse-transcription, cytokine transcript levels of Tnfa, Il6, Il1a and Il1bwere measured using TaqMan probes (Applied Biosystems). A sample size ofn 5 7–8 infected and 3 uninfected animals of each genotype were used for eachindependent experiment. Data shown in Fig. 5f and Extended Data Fig. 10h ispooled from 2 independent experiments with 3 uninfected (PBS) and 7–8 infected(Y. enterocolitica) mice used per experiment. Male mice were used in experiment 1;female mice were used in experiment 2. No blinding or randomization method wasused to place animals in uninfected or infected groups. No animals were excludedfrom analysis.Flow cytometry. Immune cell populations were identified using standard flowcytometry protocols and the following antibody clones: anti-CD11b (M1/70);anti-CD11c (N418); anti-F4/80 (BM8); anti-Ly6C (HK1.4); anti-Ly6G (1A8);anti-Siglec F (E50-2440); anti-CD103 (2E7); anti-CD3e (145-2C11); anti-CD4(RM4-4); anti-CD8 (53-6.7); anti-CD19 (1D3); anti-NK1.1 (PK136); anti-FoxP3(FJK-16 s); anti-CD69 (H1.2F3). Viability was evaluated using Sytox Blue(Invitrogen). FoxP3 staining was performed on fixed and permeabilized lympho-cytes using the FoxP3 fixation/permeabilization concentrate and diluent as permanufacturer’s instructions (eBioscience). Viability in fixed cells was evaluatedby a fixable viability dye eFluor 506 (eBioscience).[35S]Methionine metabolic labelling. Bone marrow-derived macrophages werewashed and incubated with 25 mCi ml21 [35S]methionine (Perkin Elmer) inDMEM high glucose medium without methionine (Life Technologies) supplemen-ted with 50 U ml21 penicillin, 50mg ml21 streptomycin, 2 mM of L-glutamine,sodium pyruvate and 10% FBS. After 1 h of labelling, cells were either collected(for time 0) or incubated further for the indicated time in macrophage culturemedia with non-radiolabelled methionine. Cells were washed three times with coldPBS, and then lysed in lysis buffer (50 mM Tris pH 7.5, 150 mM NaCl, 1 mMEDTA, 1% Triton X-100) containing protease and phosphatase inhibitor cocktail(Halt protease and phosphatase inhibitor cocktail, Pierce). Lysates were pre-clearedwith control agarose beads (Pierce) for 1 h and incubated with anti-ATG16L1antibody overnight (PM040, MBL Intl.). Immunoprecipitates were captured usingProtein A/G ultralink resin, eluted with sample buffer and separated by SDS–PAGE. Gels were dried using a gel drier and exposed to X-ray film.qPCR analysis. RNA isolation was performed using the RNeasy Mini kit (Qiagen).cDNA was generated from RNA using the iScript cDNA Synthesis kit (Bio-Rad).Tnfa, Il1a, Il1b and Il6 qPCR was performed using TaqMan Gene ExpressionAssays (Applied Biosystems). The following primer-probesets were used: Tnfa,Mm00443260_g1; Il1b, Mm00434228_m1; Il1a, Mm00439620_m1; Il6, Mm0046190_m1 and Mm00439620_m1; Gapdh, catalogue no. 4308313. Gene expression valueswere calculated by using 22DCt method, normalizing individual transcript levels toGapdh endogenous control.Multiple sequence alignment. Bioinformatic analysis of ATG16L1 protein fromindicated genomes was performed using NCBI alignment tool COBALT andillustrated using Jalview.Atg16l1 genotyping. All cell lines were genotyped for the A . G variant inAtg16l1 using a TaqMan SNP Genotyping Assay for SNP ID rs382830940 (LifeTechnologies, catalogue no. 4351379). Genotypes are listed in SupplementaryTable 4.Densitometry. All densitometry analysis of immunoblots was performed usingImageJ.Statistical analysis. All statistical analyses were performed using GraphPadPrism software, unpaired two-sided t-test assuming equal variance was used todetermine if the values in 2 sets of data differ. All scatterplot bars and bar graphs

ARTICLE RESEARCH


depict means of data. Given the increased cytokine production by T316A macro-phages in culture following Y. enterocolitica infection, a one-sided t-test assumingequal variance was used to determine if cytokine expression in mesenteric lymphnodes or serum IL-1b levels were greater in T316A mice vs. WT controls 48 hfollowing Y. enterocolitica oral gavage (Fig. 5f, Extended Data Fig. 10h). A P valueof ,0.05 was considered significant.

44. Barrett, J. C. et al. Genome-wide association defines more than 30 distinctsusceptibility loci for Crohn’s disease. Nature Genet. 40, 955–962 (2008).

45. Wang, G. G. et al. Quantitative production of macrophages or neutrophils ex vivousing conditional Hoxb8. Nature Methods 3, 287–293 (2006).

46. de la Calle, C., Joubert, P.-E., Law, H. K. W., Hasan, M. & Albert, M. Simultaneousassessment of autophagy and apoptosis using multispectral imaging cytometry.Autophagy 7, 1045–1051 (2011).

RESEARCH ARTICLE


Extended Data Figure 1 | T300A enhances whereas D299E abolishescaspase-3-mediated ATG16L1 processing. a, In vitro translated variantsof V5-tagged ATG16L1 (wild-type, T300A, D299E) were incubated withrecombinant active caspase 1, 3, 6 or 7 for 1 h followed by assessment ofATG16L1 cleavage by V5 immunoblot. b, HeLa cells were transfected withindicated C-terminal V5-tagged ATG16L1 construct, followed by treatmentwith 20 ng ml21 TNF 1 10mg ml21 cycloheximide (CHX) for indicatedtimes. Immunoblot was performed for V5. *non-specific band.c, LysMCre1Atg16l1loxp/loxp bone marrow progenitors were immortalized withER–HoxB8 and reconstituted with the indicated murine ATG16L1 retroviralconstructs. Macrophage differentiation was induced by removal of b-oestradiol

for 10 days. Immunoblotting with an ATG16L1 antibody indicates completedeletion of endogenous protein in vector-transfected cells and successfulreconstitution with the listed ATG16L1 variants. Expression of the truncatedsplice variant ATG16L1b9 (52 kDa) is observed along with full-lengthATG16L1b (68 kDa). d, ER–HoxB8 immortalized progenitors transduced withabove constructs were differentiated into macrophages, and cells were treatedwith indicated doses of TNF 1 10mg ml21 CHX. C-terminal fragment ofcleaved ATG16L1 was detected by immunoblotting with a V5 antibody.Data in a are representative of 3 independent experiments; data in b–d arerepresentative of 2 independent experiments.

ARTICLE RESEARCH


Extended Data Figure 2 | ATG16L1 cleavage is enhanced in primary humanmacrophages with T300A variant during TNF-a-mediated apoptosiswhereas caspase 3 and PARP cleavage remain unaltered in both groups.Representative immunoblots depicting cleavage of ATG16L1,

caspase 3 and PARP in non-risk (WT) and T300A cells following TNF-a 1

CHX treatment. Scatter plots are gated on donor macrophages demonstratingTNF-a-mediated caspase 3 cleavage.

RESEARCH ARTICLE


Extended Data Figure 3 | Generation of T316A knock-in mutant mice.a, Atg16l1 T316A knock-in construct design. b, Genomic PCR analysis ofembryonic stem cell clone used for microinjection to generate colony (cloneH4-1-B2-1), wild-type and knock-in mice. WT 5 470bp; knock-in 5 436bp.c, Atg16l1loxp/loxp construct design. d, Atg16l conditional knockout mice were

generated by crossing Atg16l1loxp/loxp mice with LysMCre1 mice. Western blotsdemonstrate protein levels of ATG16L1b/a and a-tubulin in lysates fromthioglycollate-elicited macrophages from LysMCre1Atg16l1loxp/loxp andAtg16l1wt/wt mice.

ARTICLE RESEARCH


Extended Data Figure 4 | Relative abundance of myeloid and lymphoid cellsubsets in ATG16L1 T316A knock-in mice and WT controls under basalconditions. a, Flow cytometry analysis of indicated myeloid cell populations inbone marrow, spleen, mesenteric and peripheral (inguinal and brachial) lymphnodes. n 5 5 female mice aged 6–10 weeks per genotype were analysed.

Scatterplot bars represent mean. b, Flow cytometry analysis of indicatedlymphoid cell populations in spleen, mesenteric and peripheral lymph nodes.T-cell activation was measured by CD69 staining. Five female mice aged 6–10weeks per genotype were analysed. Scatterplot bars represent mean.

RESEARCH ARTICLE


Extended Data Figure 5 | Histological analysis and quantification of smallintestine Paneth cell granules in ATG16L1 T316A knock-in and WT controlmice under basal conditions. a, Haematoxylin and eosin staining depictingnormal villus and crypt architecture, with comparable Paneth cell abundancein WT and T316A female mice aged 6–10 weeks. b–d, Similar size andmorphology of Paneth cell granules identified by lysozyme staining in WT and

T316A mice. Graph in c represents quantification of lysozyme-positive granuleallocation patterns as described10. Data are mean 6 S.D., n 5 5 female mice pergenotype, aged 6–10 weeks. A total of 701 (WT) and 891 (T316A) Panethcells were quantified. Arrows indicate specific Paneth cell morphologies asdescribed in reference 10. Error bars depict standard deviation from mean.

ARTICLE RESEARCH


Extended Data Figure 6 | Enhanced cleavage of murine ATG16L1 a and bisoforms in macrophages harbouring the T316A genotype; comparableATG16L1 stability under baseline conditions. a, Multiple exposures ofATG16L1 immunoblots reveal increased processing of ATG16L1b/aharbouring T316A variant. Graphs quantify cleavage of ATG16L1a and

b isoforms. Data are pooled from 5 independent experiments. b, 35S pulse-chaseillustrating comparable turnover of wild-type and T316A variants of ATG16L1.Densitometry analysis of ATG16L1 levels is depicted in the graph using datafrom 3 independent experiments.

RESEARCH ARTICLE


Extended Data Figure 7 | ATG16L1 processing requires caspase 3. a, HeLacells were pre-treated with vehicle (DMSO) or pan-caspase inhibitor (zVAD-fmk), followed by stimulation with 20 ng ml21 TNF 1 10mg ml21 CHX,0.5mg ml21 Fas-agonist antibody 1 10mg ml21 CHX or 2.0mM staurosporinefor indicated times. b, RAW 264.7 cells were pre-treated with DMSO or zVAD-fmk followed by treatment with 20 ng ml21 TNF-a 1 10mg ml21 CHX forindicated times. c, Bax1/1 or Bax2/2 colon carcinoma epithelial cells(HCT116) were pre-treated with DMSO or 20mM zVAD-fmk for 3 h, followedby treatment with 20 ng ml21 TNF 1 10mg ml21 CHX for indicated times.d, HeLa cells transfected with control, caspase-3- or caspase 7-specific siRNAs

were stimulated with 2.0mM staurosporine for indicated times. e, MCF-7 cellswere pre-treated with DMSO or zVAD-fmk followed by stimulation with20 ng ml21 TNF 1 10mg ml21 CHX, 0.5mg ml21 Fas-agonist antibody 1

10mg ml21 CHX or 2.0mM staurosporine for indicated times. Data in a–e arerepresentative of 2 independent experiments. Atg16l1 genotypes of cell linesused are listed in Supplementary Table 4. f, Bone-marrow-derivedmacrophages from WT mice were treated as indicated to induce canonicalNLRP3 inflammasome activation. ATG16L1, caspase 1 and IL-1b processingwere assessed by immunoblotting. upLPS 5 ultrapure LPS. Data arerepresentative of 2 independent experiments.

ARTICLE RESEARCH


Extended Data Figure 8 | Reduced autophagic flux induced by nutrientstarvation of bone-marrow-derived macrophages from T316A mice or micewith a macrophage-specific deletion of Atg16l1 (LysMCre1Atg16l1loxp/loxp).a, Immunoblot analysis of LC3-lipidation following amino acid and serumstarvation for 4 h. 400 nM bafilomycin A1 was added where indicated topromote accumulation of lipidated LC3 (LC3-II). b, Cells were nutrient starvedby culture in glucose-free medium without serum for indicated time points.400 nM bafilomycin A1 was added where indicated to promote accumulationof lipidated LC3 (LC3-II). LC3-II spots enumerated by counting the number ofpunctae identified by LC3 bright detail intensity analysis in Fig. 3. 10,000 cellsper sample were analysed, n 5 4 mice. Scatterplot bars represent mean.c, Cells were nutrient starved by culture in amino-acid depleted mediumwithout serum for indicated time points. 400 nM bafilomycin A1 was addedwhere indicated to promote accumulation of lipidated LC3 (LC3-II). LC3-II

area was quantified using image-based flow cytometry. d, Cells were culturedas in c (2 3 104 cells per well), followed by caspase 3 activity analysis usinga luciferase-based caspase 3 substrate cleavage assay. e, Cells were treatedwith 2.5mg ml21 rapamycin for 4 h. 400 nM bafilomycin A1 was addedwhere indicated. LC3-II area was quantified as in c. f, Cells were culturedas in e (2 3 104 cells per well), followed by caspase 3 activity analysisas in d. g, h, Bone-marrow-derived macrophages from control(LysMCre1Atg16l1wt/wt) or LysMCre1Atg16l1loxp/loxp mice were starvedof glucose in the presence or absence of bafilomycin A1 for 4 h, andimage-based flow cytometry used to measure the area occupied by punctateLC3 (LC3-II area, g) and LC3-II spots (h) for 10,000 cells per sample,n 5 3 mice. Scatter plot bars represent mean. Data in a–h are representativeof 2 independent experiments.

RESEARCH ARTICLE


Extended Data Figure 9 | Defective Y. enterocolitica clearance in T316Amutant or ATG16L1 deficient macrophages; comparable Y. enterocoliticainvasion and caspase 3/7 activation in wild type and T316A macrophages.a, Thioglycollate-elicited peritoneal macrophages (2 3 104 cells per well) wereinfected with 20 m.o.i. Y. enterocolitica for 30 min followed by 6 h of culturein gentamycin sulphate. Cells were lysed and caspase 3 activity measured usinga luciferase-based caspase 3 activity assay. 2.0mM staurosporine treatmentfor 3 h was used as a positive control for caspase 3 activation. zVAD-fmk wasadded where indicated to abolish caspase 3 activity. b, Thioglycollate-elicitedperitoneal macrophages from LysMCre1Atg16l11/1 or Atg16l1loxp/loxp mice

(1.25 3 105 cells per well) were infected with Y. enterocolitica for 30 minfollowed by 6 h of culture in gentamycin sulphate. Bacterial c.f.u. werecalculated following cell lysis. c, Thioglycollate-elicited peritoneal macrophagesfrom WT and T316A mutant mice (1.25 3 105 cells per well) were infected withindicated m.o.i. of Y. enterocolitica for 30 min followed by 6 h of culture ingentamycin sulphate to measure pathogen clearance. n 5 4 mice, scatterplotbars represent mean. d, Macrophages were infected as in c and immediatelywashed with gentamycin sulphate followed by cell lysis and plating to measurebacterial invasion. n 5 4 mice, scatterplot bars represent mean. Data in a–d arerepresentative of 2 independent experiments.

ARTICLE RESEARCH


Extended Data Figure 10 | Tnfa, Il1b and Il6 transcription and secretionby murine macrophages following stimulation with heat-killedY. enterocolitica, TLR ligands and muramyl dipeptide (MDP, NOD2ligand); ileal inflammation and systemic IL-1b release induced by oralgavage of Y. enterocolitica. a–c, Thioglycollate-elicited peritonealmacrophages from WT or T316A mice were stimulated with heat-killedY. enterocolitica (bacteria heated to 90 uC for 30 min) for indicated time points.a, Tnfa transcript levels were analysed by quantitative PCR, and TNF-a proteinrelease was measured by ELISA. Cytokine transcript levels are normalized toGapdh (22DCt). b, Il1b transcript levels and IL-1b protein release measuredas in a. c, Il6 transcript levels and IL-6 protein release measured as in a.d, e, TNF-a (d) and IL-6 (e) release from thioglycollate-elicited macrophagesmeasured by ELISA following stimulation with indicated ligands

for 24 h. f. IL-1b release from thioglycollate-elicited macrophages measured byELISA following stimulation with LPS, MDP or both for 24 h. Data in a–fare representative of 2 independent experiments. g, Histological analysis(haematoxylin and eosin) depicting ileal but not colonic inflammation in WTand T316A mice 48 h following oral gavage of Y. enterocolitica. Images arerepresentative of 3 PBS-treated and 7–8 Y. enterocolitica-infected mice. Allmice were males aged 7–10 weeks. h, Serum ELISA of murine IL-1b 48 hfollowing oral gavage of Y. enterocolitica. Data are pooled from 2 independentexperiments of 3 PBS-treated and 7–8 Y. enterocolitica infected mice, all aged7–10 weeks. Male mice were used in the first experiment; female mice were usedin the second experiment. Dotted line depicts lower detection limit of theELISA. Scatterplot bars depict means.

RESEARCH ARTICLE


a crohn’s disease variant in atg16l1 enhances its degradation by caspase 3

Documents