Unconventional autophagy mediated by the WD40 domain of ATG16L1 is derailed by the T300A Crohn disease risk polymorphism

ABSTRACT

A coding polymorphism of the critical autophagic effector ATG16L1 (T300A) increases the risk of Crohn disease, but how this mutation influences the function of ATG16L1 has remained unclear. In a recent report, we showed that the A300 allele alters the ability of the C-terminal WD40 domain of ATG16L1 to interact with proteins containing a specific amino acid motif able to recognize this region. This defect impairs the capacity of the motif-containing transmembrane molecule TMEM59 to induce the unconventional autophagic labeling of the same single-membrane vesicles where this protein is located. Such alteration derails the intracellular trafficking of TMEM59 and the xenophagic response against bacterial infection. In contrast, canonical autophagy remains unaffected in the presence of ATG16L1^T300A. These data argue that the T300A polymorphism impairs the unconventional autophagic activities carried out by the WD40 domain, a region of ATG16L1 whose function has remained poorly understood.

Crohn disease is a chronic inflammatory condition of the digestive tract. Several genome-wide association studies have identified a number of single-nucleotide polymorphisms that increase the risk of suffering this pathology. One of these alleles (rs2241880) imposes a single amino acid change from T to A at position 300 of ATG16L1, but how this mutation favors the onset of the disease has remained elusive. Intriguingly, ATG16L1 includes a C-terminal domain containing 7 WD40-type repetitions (the WD40 domain, WDD) that seems unnecessary for conventional autophagy. Previous studies showed that the WDD provides a docking surface for an amino acid motif present in the intracellular region of the transmembrane protein TMEM59 and other proteins. This interaction promotes LC3 labeling of the same single-membrane vesicles where TMEM59 is located, an unconventional autophagic activity that is involved in xenophagy against bacterial infection. Thus, ATG16L1 participates in atypical autophagic processes induced by adaptor molecules that, like TMEM59, promote LC3 lipidation in alternative localizations other than phagophores, the precursors to autophagosomes.

In a recent publication, we used ATG16L1-deficient cells reconstituted with different versions of ATG16L1 to explore both the contribution of the WDD to the autophagic activity of the molecule and the functional consequences of the A300 allele. We found that cells simultaneously expressing separated N-terminal (residues 1–299) and C-terminal (300–607, including the WDD) regions of ATG16L1 exhibit unaltered basal and rapamycin-induced autophagy. In contrast, TMEM59-induced autophagy is blunted. These results suggest that ATG16L1 includes 2 functional platforms: a C-terminal domain that provides an interaction site for molecules containing a WDD-binding motif to induce unconventional autophagy, and an N-terminal region that suffices for canonical autophagy and is required for all autophagic activities of the molecule. Such wide requirement for the N-terminal domain is expected, as this region recruits the LC3-lipidating complex formed by ATG12–ATG5.

Using cells restored with T300 or A300 versions of ATG16L1, we found that the A300 allele does not alter the basal autophagic flux or autophagy stimulated by rapamycin, indicating that the conventional pathway sustained by the N-terminal domain is unaffected. However, TMEM59-induced autophagy is impaired in cells expressing ATG16L1^T300A, thus causing reduced LC3 labeling of the same endosomes where TMEM59 is located. This defect provokes accumulation of TMEM59-positive vacuoles, likely due to a slowed intracellular trafficking through the endocytic route. These data indicate that the A300 allele specifically alters the unconventional activities of ATG16L1 activated through engagement of the WDD.

The T300A mutation increases susceptibility of ATG16L1 to CASP3-mediated processing at a neighboring site. However, we excluded an involvement of caspases in the observed phenotypes, because the same alterations occur in the presence of the pan-caspase inhibitor zVAD.fmk or in cells expressing a caspase-insensitive ATG16L1^T300A mutant (D299A). We found that these phenotypes likely arise from an impaired binding of the WDD to TMEM59, as the T300A mutation prevents the interaction between ATG16L1 and TMEM59 in both cellular and in vitro experimental systems. Surprisingly, other motif-containing peptides bind normally to the pathological protein, suggesting that only a fraction of the molecules that harbor a motif have their physiological activities derailed by the risk allele. This result argues that the T300A mutation might cause a mild structural alteration of the WDD that shifts its binding specificity, rather than completely inhibiting its ability to interact with the motif.

Notably, this binding defect has an impact on the xenophagic response against Staphylococcus aureus infection, a known TMEM59-dependent activity. Thus, we found that the association between TMEM59 and ATG16L1 that is induced early upon infection is blocked by the T300A mutation, without detectable CASP3 activity being present in the infected cells. This impairment is accompanied by reduced labeling of S. aureus-containing phagosomes with LC3 and increased intracellular survival of the bacteria, indicating defective xenophagy. Additional studies confirmed that a substantial portion of the xenophagic response in this setting is mediated by the WDD and TMEM59. These results show an important role of unconventional, A300-sensitive, ATG16L1-dependent autophagic events in restraining S. aureus proliferation.

Previous publications reported that caspase-mediated processing of ATG16L1^T300A during bacterial infection reduces the levels of the full-length protein and inhibits xenophagy. However, our data showing that cells expressing split ATG16L1 (where both fragments correspond to caspase-cleaved ATG16L1) have unaffected canonical autophagy suggest that even complete cleavage of the whole ATG16L1 pool would not derail the conventional pathway. Consequently, caspase processing of ATG16L1 probably impairs xenophagy by inhibiting the unconventional activities carried out by the WDD, likely because it uncouples this domain from the N-terminal effector region.

Therefore, taken together, our data indicate that the T300A mutation primarily alters the function of ATG16L1 activators that include a WDD-binding motif, either by impairing their ability to recognize the WDD or by inhibiting their function after caspase cleavage of the A300 allele (Fig. 1). This scenario suggests that future efforts to investigate the ATG16L1 dysfunctions that promote Crohn disease should focus on the atypical autophagic functions of this protein.

Figure 1.

The T300A mutation derails the function of the WD40 domain of ATG16L1 via 2 possible mechanisms. (A) The T300A mutation impairs the ability of the WDD to recognize certain versions of the generic WDD-binding motif previously identified in TMEM59. This impairment leads to defective unconventional autophagy executed by the N-terminal region of ATG16L1. (B) The T300A mutation increases susceptibility of ATG16L1 to caspase-mediated processing under stressful situations where caspases are activated. This cleavage event uncouples the N- and C-terminal domains of the molecule, leading to impaired unconventional autophagy triggered by motif-containing proteins. The WDD-binding motif version present in TMEM59 is able to bind normally the C-terminal portion of ATG16L1 that is generated after caspase processing. Both scenarios result in the same functional outcome: defective unconventional autophagy initiated by motif-containing molecules.

Notably, yeast Atg16 does not include a WDD. Our results argue that the WDD was added during evolution to allow the involvement of Atg16 in unconventional autophagic activities through the action of motif-containing adaptors that promote LC3 lipidation in membrane sites unrelated to autophagosome formation.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

Funding was obtained from the Ministerio de Ciencia e Innovación (Ref. SAF2011-23714) and the Ministerio de Economía y Competitividad (Ref. SAF2014-53320-R) of the Spanish Government, the Broad Medical Research Program (IBD-0369), the Junta de Castilla y León local government (Department of Education (CSI001A10-2, FIC016U14) and Department of Health (SAN11-FXP)) and the Fundación Solórzano (FS/1-2009 and FS/18-2014). Additional funding was received from the FEDER program of the European Union.