The Inflammatory Bowel Disease–Associated Autophagy Gene Atg16L1T300A Acts as a Dominant Negative Variant in Mice

Autophagy is an evolutionarily conserved cellular process that facilitates the disposal of proteins, such as the soluble macromolecules (e.g., nucleic acids, proteins, carbohydrates, and lipids) and organelles (e.g., mitochondria, peroxisomes, and endoplasmic reticulum) that are released into the cytoplasm during cell damage. Thus, during autophagy, these cytosolic constituents become engulfed in double-membrane vesicles called autophagosomes that deliver their contents into lysosomes where they undergo degradation (1–3). This normal clearance function may account for the fact that autophagic dysfunction is associated with aging and human disease, including various types of cancer and neurodegenerative disorders (4). It should also be noted that autophagy participates in the clearance of intracellular bacteria and viruses from host cells and affects diverse immune system functions, such as Ag presentation, lymphocyte development, and cytokine secretion (4–7). This indicates that autophagic dysfunction can conceivably lead to disorders of host defense and to autoimmune inflammation.

The possible role of autophagy in inflammation was made evident by genome-wide association studies that identified an A > G single-nucleotide polymorphism (SNP) in the Atg16L1 gene encoding a Thr³⁰⁰-to-Ala mutation (T300A) in exon 9 that is associated with increased susceptibility to Crohn’s disease (CD) (8, 9). Atg16L1 interacts with Atg5 and Atg12 to form a large (∼800-kDa) complex that promotes the conjugation of LC3-I to phosphatidylethanolamine to form LC3-II, a protein necessary for elongation of the autophagic double membrane (10). The role of Atg16L1 risk polymorphisms in the causation of CD has been difficult to decipher. Initial studies in mice showed that ablation of the Atg16L1 gene caused increased inflammasome activity and IL-1β production (11). Later, this observation was extended to humans where it was shown that the presence of the Atg16L1 risk polymorphism led to increased IL-1β secretion when cells were stimulated with the NOD2 ligand but not with TLR ligands; however, in this case, the increased secretion may have been due to increased IL-1β synthesis rather than caspase 1–dependent inflammasome activity (12, 13). In another line of study, Atg16L1 gene deletion led to abnormalities in Paneth cell secretory function that were mirrored in CD patients with risk polymorphisms by morphologic abnormalities in Paneth cell granules (14). Most recently, it was demonstrated that Atg16L1T316A (Atg16L1 polymorphism equivalent to human Atg16L1T300A) sensitizes Atg16L1 to caspase 3–mediated degradation (15). Because caspase 3 is generated by cellular stress or TNF-α secretion, this study implies that the Atg16L1 risk polymorphism leads to autophagy protein inactivation during Crohn’s inflammation.

In this study, we define the functional defects in Atg16L1 knock-in (KI) mice bearing the CD T300A risk polymorphism on one allele (heterozygous KI mice) or on both alleles (homozygous KI mice). Surprisingly, we found that Atg16L1 heterozygous KI mice exhibited a range of autophagy defects that were very similar to those of homozygous KI mice; this included macrophage abnormalities affecting bacterial clearance and inflammasome cytokine production, as well as susceptibility to caspase 3 degradation. In addition, both types of KI mice exhibit similar defects affecting pathogen (Salmonella) invasion that lead to similar increases in the severity of Salmonella-induced colitis. These findings suggested that Atg16L1 bearing a T300A polymorphism on one allele acts as a dominant negative affecting the function of the wild-type (WT) Atg16L1 allele. Evidence in support of this conclusion came from studies showing that the phenotype of Atg16L1^KO/+ mice was the same as that of WT mice but was different from that of Atg16L1^T300A/+ mice.

The construct for targeting the C57BL/6 Atg16L1 locus in embryonic stem (ES) cells was made using a combination of standard molecular-cloning techniques. Briefly, the Atg16L1 genomic sequence was obtained from an Atg16L1 BAC clone (BACPAC Resources, Oakland, CA). A 7-kb fragment spanning Atg16L1 exon 9 was obtained by PCR from the BAC clone and subcloned into a pL253 vector (PL253-Atg16L1) by homologous recombination (Gene Bridges). A PCR fragment containing Atg16L1 exon 9 was cloned into a PGL451 vector using SacII and NotI sites. A second Loxp site was introduced in front of exon 9 by ligating a Loxp sequence via a SacII site into the PGL451. An Atg16L1T300A mutation was introduced into the cloned fragment in PGL451 by site-directed mutagenesis PCR (Stratagene), according to the manufacturer’s instructions. The mutation was confirmed by sequencing. A fragment spanning Loxp–exon9T300A–Loxp–FRT–bGHpA–Neo–PGK–FRT in the PL451 vector was further cloned into a PL253-Atg16L1 vector to replace the WT exon 9 fragment and to introduce a neomycin-resistance cassette flanked by two FRT sites via homologous recombination, according to the Gene Bridges kit instructions. A probe outside each homology arm was designed for monitoring successful incorporation of the Neo cassette. The targeting construct, as well as cloned probes, were sent to Ozgene. C57BL/6-derived ES cells were electroporated for generation of Atg16L1T300A-targeted ES cell clones, and the targeted cells were identified by Southern blot and PCR strategies. The T300A-KI region was sequenced to ensure integration at the appropriate site. Targeted ES cells were injected into blastocysts using standard techniques, and chimeras obtained were crossed with C57BL/6 females to obtain mice with germline transmission. Heterozygous mice were interbred to obtain WT, heterozygous, and homozygous mice. The following primers were used for genotyping, as indicated in Fig. 1A. P1 primer: forward: 5′-CCTGGAGCTGGGCAGTCAGGTTGGGCTCCATG-3′, P2 primer: reverse: 5′-GCTGCTTCCCTGTCAGTCAACTGTG-3′, and P3 primer: reverse: 5′-CGCATCGCCTTCTATCGCCTTCTTGACGAG-3′. PCR by P1, P2 primer generates a 260-bp fragment for the WT allele and generates a 350-bp fragment if there is a T300A-KI allele. For knockout (KO) detection, PCR by P1, P3 primer generates a 660-bp fragment. To obtain Atg16L1-KO mice, Atg16L1^f/f mice were crossed with LysMCre mice. All in vivo experiments were performed using age-matched littermate control mice. Our littermate breeding strategy was as follows: Atg16L1^T300A/+ × Atg16L1^T300A/+ to obtain WT, Atg16L1^T300A/+, and Atg16L1^T300A/T300A. Atg16L1^f/fLysMCre mice were obtained by Atg16L1^f/f × Atg16L1^f/fLysMCre+. Atg16L1^f/f mice were crossed with CMV-Cre mice to obtain Atg16L1^KO/+ mice and then Atg16L1^KO/+ mice were bred with Atg16L1^T300A/+ mice to obtain WT, Atg16L1^T300A/+, and Atg16L1^KO/+ offspring. Studies were carried out under animal care guidelines of the Institute of Microbiology, Chinese Academy of Sciences.

Bone marrow–derived macrophages (BMDMs) were obtained as previously described (16). Briefly, bone marrow cells flushed out of femurs were plated on sterile petri dishes and cultured for 1 wk in RPMI 1640 medium containing 10% FBS and 20% L929 cell–conditioned medium. Peritoneal macrophages were obtained by i.p. injection of 2.5 ml of 4% thioglycollate solution (Difco), followed by peritoneal lavage 4 d later. The cells obtained were washed and resuspended in DMEM supplemented with 10% FBS (17).

Salmonella typhimurium was grown in Luria broth at 37°C, whereas S. typhimurium expressed with mCherry (a gift from Feng Shao, National institute of Biological Sciences, Beijing, China) was grown in Luria broth containing tetracycline at 37°C. For in vitro studies of bacterial infection with these organisms, BMDMs or peritoneal macrophages were infected with bacteria at the indicated multiplicity of infection (moi), centrifuged at 1000 × g for 10 min at 25°C, and cultured for 30 min at 37°C. Then cells were washed extensively with 50 μg/ml gentamicin sulfate to remove free bacteria, cultured for an additional 3, 6, or 9 h, washed twice with PBS, and processed for real-time PCR, immunostaining, or intracellular bacterial counting. For intracellular bacterial counting, infected cells were lysed in 0.1% saponin, followed by plating serial dilutions of lysate on MacConkey agar, overnight incubation at 37°C, and counting the bacterial colonies on the agar plates. Following culture for 9 h, cytokine production in cell culture supernatants was assessed by ELISA.

Age- and sex-matched mice (6–8 wk old) were fasted for 5 h and orally infected with 1 × 10⁸ CFU Salmonella. At the indicated times postinfection, mice were euthanized for harvest of mesenteric lymph nodes (MLNs), liver, and spleen; these tissues were homogenized in 0.1% saponin, and the cell lysates obtained were plated on MacConkey agar overnight at 37°C. Bacterial colonies were counted the following day. Small intestine and colon were collected and analyzed for cytokine production by RT-PCR.

To detect starvation-induced Atg16L1 expression levels in BMDMs, the cells were starved by culture in Earle’s Balanced Salt Solution (EBSS) for 3 h and then detached from the plastic surface by incubation in trypsin for 5–10 min. The cells obtained were centrifuged, dispersed, fixed in Cytofix/Cytoperm buffer (BD Biosciences) for 20 min, and blocked by incubation in 1% BSA-PBS for 30 min at room temperature. The fixed cells were incubated with anti-Atg16L1 (Cell Signaling Technology) at a 1:50 dilution in Perm/Wash buffer for 1 h at 4°C, after which Alexa Fluor 488 goat anti-rabbit IgG was added at a 1:500 dilution for 30 min at 4°C.

To detect IL-17A and IFN-γ production, MLN and spleen cells from infected mice were cultured in growth medium containing 50 ng/ml PMA and 1 μM ionomycin at 37°C for 1 h, after which GolgiStop (BD Biosciences) was added, and the cells were cultured for an additional 3 h. Flow cytometric analysis of cells bearing intracellular stains was performed as described previously (18).

RNA isolation was performed using TRIzol Reagent (Invitrogen) in accordance with the manufacturer’s instructions. cDNA was generated using a High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). TNF-α, IL-1β, and IL-6 TaqMan quantitative PCR was performed. Probes for each of the genes are from the Universal Probe library (cat. no. 04684974001; Roche). Gene-expression values were calculated using 2^−△△Ct normalized to Hprt.

S. typhimurium enterocolitis was induced as previously described (19). Briefly, age- and sex-matched mice (6–8 wk old) of the indicated genotypes were fasted for 5 h and then administered streptomycin (20 mg) orally. Mice were infected 24 h later with 1 × 10⁵ CFU S. typhimurium by gavage and monitored daily for weight loss. Mice were euthanized at 8 d postinfection, at which point small intestine and colon were collected for histological staining [disease scores were calculated as previously described (20, 21)], and MLNs and spleen were collected for assay of IL-17A and IFN-γ production.

BMDMs or peritoneal macrophages were lysed in RIPA buffer with protease inhibitors and phosphatase inhibitors. Lysates obtained were subjected to SDS-PAGE and transferred to polyvinylidene difluoride membranes that were blocked by exposure to PBS containing 5% nonfat dry milk and 0.1% Tween 20 for 1 h at room temperature. The blocked membranes were probed by incubation with the indicated Abs at 4°C overnight. Proteins were detected with HRP-conjugated secondary Ab and visualized by ECL (GE). The following Abs were used: anti-actin (Cell Signaling Technology), anti-LC3 (Sigma), anti-Atg16L1 (Sigma and MBL).

Immunofluorescence microscopy assay was performed as previously described (22). Briefly, BMDMs were grown on glass coverslips and then treated with EBSS or infected by bacteria for the indicated times. The cells were fixed in 3.7% formaldehyde-PBS for 15 min and permeabilized with 0.2% Triton X-100–PBS containing 1 mg/ml DAPI for 10 min. Then cells were blocked by 1% BSA-PBS for 30 min, incubated with primary Ab (anti-LC3 diluted in 1% BSA–PBS) at 4°C for 12 h or at room temperature for 2 h, and washed three times with PBS. Finally, the cells were incubated with secondary Ab (Alexa Fluor 488 goat anti-rabbit IgG diluted in 1% BSA–PBS) at room temperature for 1 h and washed three times with PBS. The coverslips were mounted with PermaFluor Aqueous Mounting Medium and examined by confocal microscopy using a Leica SP8 microscope.

Small intestinal or colonic tissue obtained from euthanized mice was flushed with PBS, sectioned, fixed in 4% paraformaldehyde, embedded in wax, and stained with H&E.

Primary BMDMs and peritoneal macrophages were seeded in 12-well plates and allowed to grow for 24 h. Cells were stimulated as indicated. Cell supernatants were collected, and IL-1β, IL-6, and TNF-α were analyzed by ELISA in accordance with the manufacturer’s instructions (BD Biosciences).

GraphPad Prism v5 was used for statistical analyses. One/two-way ANOVA was initially done to determine whether an overall statistically significant difference existed before analysis with a Tukey/Bonferroni post hoc test for multiple comparisons. The p values < 0.05 were considered statistically significant.

To investigate the link between the Atg16L1T300A polymorphism and CD, we constructed a mouse line with an Atg16L1-KI mutation in the murine Atg16L1 gene that is equivalent to the Atg16L1 SNP associated with human CD, T300A (hereafter referred to as Atg16L1^T300A/T300A or Atg16L1^T300A/+ mice). However, it should be noted that the precise location of the KI mutation depends on which of the three splice isoforms of Atg16L1 present in mice is targeted, and it is only in the β isoform that the mutation is located at position 300. The Atg16L1 gene in these mice also had Loxp sites bracketing exon 9, so that Atg16L1 could be conditionally deleted when interbred with mice bearing Cre in various cells (Fig. 1A). Atg16L1^T300A/T300A and Atg16L1^T300A/+ mice develop normally and remain free of spontaneous inflammation after birth (data not shown).

To assess whether the presence of the T300A SNP affects autophagy, BMDMs from WT, T300A homozygous-KI (designated as Atg16L1^T300A/T300A), and T300A heterozygous-KI (designated as Atg16L1^T300A/+) mice were treated with rapamycin, an autophagy-inducing factor, after which whole-cell lysates were prepared and subjected to immunoblot to detect the presence of LC3-II. We found that LC3-II generation was decreased in Atg16L1^T300A/T300A and Atg16L1^T300A/+ mice compared with WT mice (Fig. 1B). To confirm this finding, BMDMs from WT, Atg16L1^T300A/T300A, Atg16L1^T300A/+, and Atg16L1^f/fLysMCre+ (Atg16L1-KO) mice were cultured under amino acid–starvation conditions (i.e., in EBSS media), another inducer of autophagy, in the absence or presence of bafilomycin A1 (Bafil.A1), a vacuolar H⁺ ATPase inhibitor that prevents a decrease in lysosomal pH and, thus, allows evaluation of autophagic flux. Cell lysates were then prepared and subjected to immunoblotting to detect LC3-II (Fig. 1C) and immunofluorescence staining to detect LC3-II puncta (Fig. 1D). As in the study with rapamycin, the accumulation of LC3-II detected by immunoblotting was decreased in amino acid–BMDMs from Atg16L1^T300A/T300A and Atg16L1^T300A/+ mice compared with those from WT mice (Fig. 1C). In addition, amino acid–starved BMDMs from Atg16L1^T300A/T300A and Atg16L1^T300A/+ mice exhibited decreased numbers of cells displaying LC3-II puncta, as well as decreased numbers of puncta per cell in puncta-expressing cells; in both cases, these decreases were similar to those in Atg16L1-KO cells (Fig. 1D, 1E). Finally, the studies in which Bafil.A1 was added to the cell culture indicated that the decreased autophagy was not due to changes in autophagic flux. Taken together, these data suggested that both the Atg16L1T300A polymorphism heterozygote and homozygote have defects in autophagy induction in macrophages.

Previous studies showed that Atg16L1- or Atg7-deficient macrophages secrete increased amounts of IL-1β in response to inflammasome stimulation, suggesting that basal autophagy regulates inflammasome-related inflammatory cytokine secretion (11). To determine whether the defective autophagy observed in BMDMs from T300A-KI mice also triggers an increased inflammasome response, cytokine responses were analyzed in BMDMs from WT, Atg16L1^T300A/+, and Atg16L1^T300A/T300A mice that were stimulated with LPS/ATP, LPS/MDP, and zymosan. BMDMs from Atg16L1^T300A/+ and Atg16L1^T300A/T300A mice (all figures hereafter show WT versus Atg16L1^T300A/+ versus Atg16L1^T300A/T300A) exhibited significantly increased IL-1β secretion compared with those from WT mice (Fig. 2A–C). In contrast, KI cells stimulated with the same stimuli elicited similar TNF-α and IL-6 responses as those of WT mice (Fig. 2A–C). Thus, defective autophagy in Atg16L1^T300A/+ and Atg16L1^T300A/T300A macrophages results in heightened inflammasome responses associated with normal noninflammasome cytokine responses.

In related studies, we determined cytokine responses of BMDMs from T300A-KI mice infected by S. typhimurium. As in the case of LPS/ATP or LPS/MDP stimulation, S. typhimurium infection–induced IL-1β secretion was significantly increased in BMDMs from Atg16L1^T300A/+, Atg16L1^T300A/T300A, and Atg16L1-KO mice compared with those from WT mice, whereas the production of IL-6 and TNF-α was equivalent to that in WT mice (Fig. 2D). However, this increased IL-1β secretion was not accompanied by increased IL-1β mRNA production, in keeping with the presumption that the increased secretion was due to inflammasome activation (Fig. 2E). To determine whether the increased cytokine production in infected Atg16L1T300A-KI BMDMs is due to increased bacterial burden, cytokine secretion was determined in BMDMs stimulated with heat-killed S. typhimurium. Again, IL-1β secretion was significantly increased in BMDMs from Atg16L1^T300A/+ mice, Atg16L1^T300A/T300A mice, and Atg16L1-KO mice compared with those from WT mice (Fig. 2F), whereas IL-6 and TNF-α production by BMDMs from these mice groups was equivalent. Thus, the increased Salmonella-induced IL-1β production described above was not due to increased bacterial burden.

Overall, these data are consistent with previous findings showing that autophagy negatively regulates inflammasome-mediated IL-1β secretion. In addition, they show that Atg16L1^T300A/+ and Atg16L1^T300A/T300A macrophages exhibit responses very similar to those of Atg16L1-KO cells.

Autophagy was shown to play an important role in the recognition and degradation of intracellular pathogens (21, 23, 24). In initial studies to determine whether this is the case for T300A-KI mice, we infected cells from KI mice (as well as KO mice) and WT mice with S. typhimurium, washed the cells to remove uningested bacteria, and quantitated residual bacteria within cells over time. We found that BMDMs and peritoneal macrophages from Atg16L1^T300A/+ and Atg16L1^T300A/T300A mice displayed significantly less efficient clearance of ingested bacteria than did WT cells (Fig. 3A, 3B). In addition, this deficiency was similar to that displayed by cells from Atg16L1-KO mice (Fig. 3A, 3B).

In further studies to determine whether the above defect in bacterial clearance was an artifact of altered bacterial entry of the KI BMDMs, we infected cells with fluorescence-labeled S. typhimurium that constitutively express mCherry and washed the cells to remove uningested bacteria. We then evaluated the cells by immunofluorescence to determine their capacity to take up labeled bacteria at zero time to measure bacterial entry and again at 3 and 6 h after washing to measure subsequent bacterial clearance. We found that bacterial entry was comparable in BMDMs from WT, Atg16L1^T300A/+, Atg16L1^T300A/T300A, and Atg16L1-KO mice (Fig. 3C). In contrast, bacterial clearance by BMDMs from Atg16L1^T300A/+ as well as Atg16L1^T300A/T300A andAtg16L1-KO mice was significantly compromised compared with WT cells (Fig. 3C, 3D).

Next, we determined the capacity of T300A-KI mice to clear bacteria in vivo. To this end, various organs of mice infected orally with S. typhimurium were assessed for residual bacterial burden 7 d postinfection (see 2Materials and Methods). We found that the MLNs from infected Atg16L1^T300A/+ and Atg16L1^T300A/T300A mice were significantly larger than were those from WT mice (Fig. 4A); in addition, bacterial burdens in spleens and livers of KI mice were significantly increased compared with WT mice (Fig. 4B). Finally, IL-1β, IL-6, and TNF-α mRNA expression in the small intestine (Fig. 4C) and IL-1β and TNF-α secretion in peritoneal macrophages (Fig. 4D) were increased significantly in Atg16L1^T300A/+ and Atg16L1^T300A/T300A mice compared with WT mice after S. typhimurium administration.

Collectively, these data provide further evidence that Atg16L1^T300A/+ and Atg16L1^T300A/T300A mice have a very similar phenotype in vivo and, thus, complement the in vitro data provided above.

To explore the mechanism of the compromised autophagy, defective bacterial clearance, and increased inflammatory cytokine response exhibited by T300A-KI mice, BMDMs from WT, Atg16L1^T300A/+, and Atg16L1^T300A/T300A mice were cultured under starvation or starvation/Bafil.A1 conditions, after which Atg16L1 expression was measured by real-time RT-PCR and flow cytometry. We found that Atg16L1 mRNA levels are comparable among WT, Atg16L1^T300A/+, and Atg16L1^T300A/T300A mice (Fig. 5A), whereas Atg16L1^T300A/+ and Atg16L1^T300A/T300A mice display slightly decreased amounts of Atg16L1 protein, as measured by flow cytometry, compared with WT mice (Fig. 5B).

It was reported recently that Atg16L1 protein resulting from a Atg16L1 gene that bears a T316A SNP in Atg16L1 (i.e., an SNP equivalent to the T300A SNP studied in this article), is more sensitive to TNF-α–induced caspase 3 activation and cleavage than is WT Atg16L1 protein (15, 25). These studies focused on mice bearing a mutated Atg16L1 gene on both alleles. To investigate whether Atg16L1 in Atg16L1T300A heterozygous cells is also more sensitive to TNF-α–induced caspase 3 cleavage than is Atg16L1 in WT mice, BMDMs from WT, Atg16L1^T300A/+, and Atg16L1^T300A/T300A mice were treated with TNF-α in the presence of cycloheximide, after which extracts were subjected to immunoblot analysis. Atg16L1 in Atg16L1^T300A/+ and Atg16L1^T300A/T300A cells exhibited enhanced cleavage compared with that in WT cells (Fig. 5C). Because TNF-α treatment can induce cell apoptosis (26, 27), it is possible that the above difference in cleavage was due to protein degradation associated with apoptosis. To examine this possibility, we cultured BMDMs from WT and T300A-KI mice in the presence of TNF-α and measured apoptosis by flow cytometric analysis of annexin V/propidium iodide staining (Fig. 5D). The results showed that apoptosis in BMDMs from Atg16L1^T300A/+ and Atg16L1^T300A/T300A mice was equivalent to that in WT mice; therefore, increased apoptosis was not the cause of the increased cleavage.

Overall, these findings are consistent with the previous report showing that the murine T316A polymorphism of Atg16L1 is more sensitive to caspase 3 cleavage. However, we provide evidence that cells from Atg16L1^T300A/+ heterozygous mice are also more sensitive to caspase 3 cleavage than are those from Atg16L1^T300A/T300A mice.

It was shown that mice with defective autophagy manifest increased intestinal invasion by commensal and pathogenic organisms. We next explored the development of Salmonella-induced colitis in Atg16L1^T300A-KI mice. Mice were treated with streptomycin to partially reduce the number of normal bacteria in the intestinal lumen (28) and were administered S. typhimurium 24 h later. The mice were monitored for weight loss for 8 d, after which they were evaluated for histology and cytokine response. During the 8-d period of observation, Atg16L1^T300A/+ and Atg16L1^T300A/T300A mice suffered greater weight loss than did WT mice (Fig. 6A) and exhibited more intense inflammation in the small intestine and colon (Fig. 6B, 6C). However, the numbers of CD4⁺ IL-17⁺ cells and CD4⁺ IFN-γ⁺ T cells were not significantly different among the various groups (Fig. 6D). These results provide strong evidence that mice bearing an Atg16L1T300A polymorphism, either as homozygotes or heterozygotes, are more sensitive to infection with invasive bacteria.

The above studies documenting the presence of abnormalities in mice bearing the Atg16L1^T300A polymorphism are compatible with the possibility that the mutant allele acts as a dominant-negative variant that inhibits the function of the WT allele. Alternatively, they are compatible with the possibility that the WT allele is haploinsufficient, in that it is unable to provide sufficient function on its own. To distinguish between these possibilities, we compared the function of cells from Atg16L1T^300A/+ mice with that of cells from Atg16L1^KO/+ mice, recognizing that the latter mice should exhibit haploinsufficiency if the Atg16L1 gene is inadequate when present on only one allele; however, if this is not the case, the Atg16L1T300A polymorphism is a dominant negative. To this end, we bred our Atg16L1^f/f mice with CMV-Cre mice to allow acquisition of Atg16L1^KO/+ mice. We then bred Atg16L1^KO/+ mice with Atg16L1^T300A/+ mice to obtain WT, Atg16L1^KO/+, and Atg16L1^T300A/+ mice among the offspring. Atg16L1 expression levels in peritoneal macrophages and BMDMs from WT, Atg16L1^T300A/+, Atg16L1^T300A/T300A, and Atg16L1^KO/+ mice were determined by RT-PCR or flow cytometry, as described for some of these cells in Fig. 5A and 5B. As expected, in repeated studies, Atg16L1 in Atg16L1^KO/+ macrophages was ∼50% lower than that of WT and KI mice at the mRNA and protein levels (Fig. 7A, 7B). A similar result was obtained for the protein level in Western blot studies (data not shown).

We next conducted studies evaluating the function of Atg16L1^KO/+ and other mice. Accordingly, BMDMs from WT, Atg16L1^T300A/+, Atg16L1^T300A/T300A, and Atg16L1^KO/+ mice were cultured under amino acid–starvation conditions (i.e., in EBSS media) in the absence or presence of Bafil.A1. Cells were then prepared and subjected to immunofluorescence staining to detect LC3-II puncta. LC3-II puncta accumulation in Atg16L1^KO/+ cells was similar to that of WT cells; in contrast, LC3-II puncta accumulation in Atg16L1^T300A/+ cells was decreased compared with that in Atg16L1^KO/+ cells but was similar to that in Atg16L1^T300A/T300A cells (Fig. 7C). In further studies, peritoneal macrophages and BMDMs from WT, Atg16L1^T300A/+, Atg16L1^T300A/T300A, and Atg16L1^KO/+ mice were infected with Salmonella, and bacterial clearance was quantitated as in Fig. 3. We found that both peritoneal macrophages and macrophages from WT and Atg16L1^KO/+ mice displayed similar bacterial clearance capacity, whereas macrophages from Atg16L1^T300A/+ and Atg16L1^T300A/T300A mice exhibited significantly less efficient clearance; in addition, this decreased clearance was similar to that displayed by macrophages from Atg16L1-KO mice (Fig. 7D). Similar findings were obtained with studies of BMDMs (data not shown). In a final study, Salmonella colitis was induced in WT, Atg16L1^T300A/+, Atg16L1^T300A/T300A, and Atg16L1^KO/+ mice, as in Fig. 6. Then the mice were monitored for weight loss for 6 d, after which they were evaluated for histology. The weight loss (Fig. 7E) and histology (Fig. 7F) of Atg16L1^KO/+ mice were the same as for WT mice, whereas the weight loss and severity of inflammation in Atg16L1^T300A/+ mice were greater than those in Atg16L1^KO/+ mice (Fig. 7G).

Taken together, these data provide strong evidence that autophagic function in Atg16L1^KO/+ mice is equivalent to that in WT mice, despite their heterozygosity; therefore, the Atg16L1T300A polymorphism acts as a dominant-negative variant rather than a haploinsufficient variant.

In the initial studies establishing that a threonine-to-alanine substitution in Atg16L1 (T300A SNP) was a risk polymorphism associated with CD, it was found that, although the data in support of this association were robust, the actual increase in risk was relatively modest: 1.65-fold in homozygotes with all forms of CD and 2.2-fold in homozygotes with ileal disease (29). This was due to the fact that, although a high percentage of CD patients were homozygous for the risk allele (59–60%), a high percentage of unaffected individuals were also homozygous for this allele (52–53%) (8, 29). Heterozygosity for the risk allele among patients (and controls) was also high; although it did not confer increased risk for all CD patients, it conferred a marginally increased risk for patients with ileal disease, despite the fact that the association with CD was said to follow a recessive model. In the current study, we establish that KI mice that are homozygous or heterozygous for the risk allele manifest severe abnormalities in a wide range of autophagy functions. Therefore, these data in mice suggest that the presence of an Atg16L1 risk polymorphism on only one chromosome may have more impact on autophagy function in humans than one would surmise from the only marginal increase in risk for ileal CD conveyed by heterozygous carriage of the risk allele. Evidence that this may be the case comes from a study by Strisciuglio et al. (30) showing that pediatric patients with CD who are homozygous and heterozygous for the T300A risk polymorphism exhibit impairment of Ag uptake and processing, as well as compromised interactions between dendritic cells and the intestinal epithelium. One way to explain a possible discrepancy between compromised autophagy function in heterozygotes and a risk for CD in humans is to assume that increased risk for CD conveyed by the Atg16L1 risk polymorphism only occurs when the polymorphism is concomitantly associated with a second risk polymorphism and that homozygosity is more efficient than heterozygosity in facilitating this compound genetic risk. This could occur because the underlying autophagy abnormality is somewhat more severe in humans who are homozygous for the risk allele than in those who are heterozygous for this allele or because of other undefined factors. In any case, the data reported in this article highlight the importance of additional studies of autophagy function in homozygous and heterozygous patients with CD, as well as in individuals without CD.

In light of the mouse data presented in this article showing that heterozygous mice have profound autophagy defects and susceptibility to caspase 3 cleavage, it appeared likely that this genetic abnormality is manifesting haploinsufficiency or a dominant-negative effect. To distinguish between these possibilities, we conducted studies in which we evaluated heterozygous mice carrying a WT Atg16L1 allele and a null allele. These mice exhibited normal function by several criteria, indicating that carriage of one normal Atg16L1 allele is sufficient for normal Atg16L1 function. Thus, the explanation for the abnormality in Atg16L1^T300A/+ mice is not due to haploinsufficiency but rather to a dominant-negative effect. The question then arises: How could this occur? One possibility is that, because Atg16L1 is part of a complex that facilitates activation of LC3, the cleaved molecule arising from the chromosome encoding the Atg16L1 bearing the risk polymorphism might still be capable of participating in the formation of the complex, but that the Atg16L1^T300A/+ complex containing the cleaved molecule does not function well. Studies to examine this and other possibilities are underway.

The increased inflammasome activity associated with abnormalities in Atg16L1 function was noted in previous studies of mice with total deletion of Atg16L1, as well as in one previous study of a murine T300A-KI abnormality (11, 15). The mechanism of this negative effect of intact autophagy on inflammasome activity is poorly understood, but it may involve the ability of autophagic activity to minimize or inhibit the release of intracellular inflammasome inducers, such as reactive oxygen species arising from dysfunctional mitochondria (31, 32). Whatever mechanism is operative, it is clear that defective autophagy leads to increased NLRP3 activity and release of IL-1β, a cytokine that was shown to intensify intestinal inflammation. Thus, although it is easy to assume that Atg16L1 defects lead to CD because they cause decreased bacterial clearance and increased penetration of the commensal organisms, the more important result of the Atg16L1 defect may be its effect on inflammasome activity and IL-1β release by innate immune effector cells, such as macrophages and dendritic cells. This view is supported by the fact that IL-1β is an important proinflammatory effector cytokine, as well as a potent inducer of Th17 responses, especially in humans (33).

It is commonly assumed that a risk polymorphism that occurs at a high frequency in the general population must operate in tandem with other genetic defects to actually cause disease. This assumption has special relevance to the Atg16L1T300A risk polymorphism because, as we have seen, autophagy abnormalities due to this polymorphism are seen in cells from heterozygous individuals. In addition, it is unlikely that the loss of the bacterial-clearance mechanism in KI mice or in humans with the Atg16L1 risk polymorphism is sufficient to cause inflammatory disease by itself, because KI mice do not develop spontaneous gut inflammation. As discussed above, the Atg16L1 risk polymorphism is likely to require the presence of second risk polymorphism to result in CD. A possible second risk polymorphism is the NOD2 risk polymorphism, because several studies showed that Atg16L1 and NOD2 function are sometimes associated. In one such study, NOD2 was shown to participate in the recruitment of Atg16L1 to the plasma membrane and, thus, the initiation of autophagy during bacterial entry and uptake by autophagosome membranes (34). In a more recent and somewhat related study, it was shown that Atg16L1 and NOD2 are mechanistically involved in the uptake of organisms: in this case the uptake of Bacteroides fragilis that induce regulatory T cells that produce IL-10 and, thus, are involved in gut homeostasis (35). However, despite this functional association of Atg16L1 and NOD2, there is no evidence from population studies that polymorphisms involving the genes encoding these proteins act together to magnify the risk for CD (29). On the contrary, it appears that the Atg16L1 risk polymorphism acts independently of NOD2 or any other risk polymorphism.

Another risk polymorphism that might be associated with the Atg16L1T300A polymorphism is one involving Xbp1, a gene encoding a protein regulating the unfolded protein response and epithelial cell stress. This possibility derives from studies showing that mice with epithelial cells deficient in Xbp1 and Atg16L1 manifest Paneth cell dysfunction and spontaneous small intestinal inflammation (36). Further population studies are necessary to assess whether polymorphisms involving Atg16L1 and Xbp1 genes are in fact interactive; in any case, it should be borne in mind that polymorphisms of the Xbp1 gene are not sufficiently common to qualify as regulators of Atg16L1 autophagy abnormalities in human populations.

In our study, oral administration of Salmonella to KI mice led to increased uptake of Salmonella organisms and increased dissemination of this pathogen to various internal sites. We demonstrated for the first time, to our knowledge, that KI mice exhibit increased Salmonella-induced colitis compared with WT mice. This susceptibility to Salmonella infection is likely to arise from the defective macrophage-clearance mechanism alluded to above that also governs commensal organisms, except in this case the organism involved is much more likely to survive and proliferate in the mucosa because of its inherent virulence properties. One should not extrapolate from this characteristic of Atg16L1^T300A-KI mice that the Atg16L1T300A risk polymorphism leads to the development of CD, because this polymorphism results in an inability to handle Salmonella infection or infection with another intestinal pathogen; this follows from the fact that CD is not characterized by an increased incidence of such infection either prior to or during the course of disease. In contrast, it is possible that the risk polymorphism impairs the ability to clear a semipathogenic organism, such as the adherent invasive Escherichia coli organisms that were cited as aggravating factors in ileal CD (37).

We thank Xiaolan Zhang for expert technical support.

The authors have no financial conflicts of interest.