Introduction

Macroautophagy (hereafter referred to as autophagy) is a well-conserved and lysosome-dependent catabolic process that recycles undesired or harmful cytosolic components for providing essential building blocks to maintain cellular homeostasis in mammals1,2,3,4. Autophagy necessitates de novo synthesis of double-membrane autophagosome through the sophisticated cooperation of a series of autophagy-related (ATG) proteins3,5,6. During the amino acid starvation-induced canonical autophagy, calcium transients trigger liquid-liquid phase separation of FAK family kinase-interacting protein of 200 kDa (FIP200) and the FIP200-containing Unc-51-like kinase (ULK) complex in the pre-autophagosomal structure (PAS) to initiate autophagosome formation7. Concurrently, the class III phosphatidylinositol-3-kinase complex I (PI3KC3-C1) is translocated to the PAS and is subsequently activated to phosphorylate phosphatidylinositol (PI) to generate phosphatidylinositol-3-phosphate (PI3P)8. Meanwhile, both the FIP200 subunit of the ULK complex and the WD repeat domain phosphoinositide-interacting protein 2 (WIPI2) that senses the PI3P signal from PI3KC3-C1 can regulate the targeting of the E3-like ATG12~ATG5-ATG16L1 complex (the ATG16L1 complex) to the PAS9,10,11,12. Subsequently, the ATG16L1 complex catalyzes the phosphatidylethanolamine (PE) lipidation of ATG8 family proteins to facilitate the elongation of PAS to form autophagosome13,14,15. Although the molecular mechanism underpinning the recruitment of the ATG16L1 complex by WIPI2 is well elucidated in previous studies12,16, how FIP200 specifically interacts with ATG16L1 to recruit the ATG16L1 complex remains elusive. Moreover, why autophagy deploys two distinct approaches to recruit the ATG16L1 complex is still enigmatic.

In addition to non-selective canonical autophagy, accumulating studies have uncovered considerable selective autophagy processes mediated by different types of autophagy receptors17,18,19,20,21,22,23. Currently identified autophagy receptors in mammals, such as p62, NBR1, Optineurin, CCPG1, NDP52, and TAX1BP1, all encompass a cargo-associating domain that can specially recognize certain types of autophagic cargoes as well as a distinct ATG8-interacting motif (AIM) that recognizes the ATG8 family proteins known as LC3A, LC3B, LC3C, GABARAP, GABARAPL1, and GABARAPL2 in mammals19,20,24,25,26. Moreover, recent studies revealed that in order to induce in situ autophagosome formation to envelope and sequester the targeting cargoes, many autophagy receptors can directly interact with FIP200 to recruit the ULK complex27,28,29,30. Particularly, some autophagy receptors can directly bind to the Claw domain of FIP200 through their respective FIP200-interacting region (FIR) for recruiting the ULK complex27,28,30,31. Since the sequence pattern of AIM bears a striking resemblance to that of FIR, the AIM motifs of many autophagy receptors can also function as FIR to interact with FIP200 Claw27,28. However, due to the high similarity, it is challenging to dissect the individual contribution of FIP200 and ATG8 family proteins for binding to these autophagy receptors. Interestingly, similar to aforementioned autophagy receptors, mammalian ATG16L1 can function as an adapter to specifically recognize invading pathogens or pathogen-containing vacuoles through its C-terminal WD40 repeats domain to mediate the antibacterial selective autophagy (also named as xenophagy)32,33,34,35. Moreover, a previous elegant study of the network organization of the human autophagy system well demonstrated that ATG16L1 can directly bind to ATG8 family proteins36. However, how ATG16L1 interacts with ATG8 family proteins and the detailed relationship between FIP200 and ATG8 family proteins in binding to ATG16L1 are largely unknown.

In this study, we discover that ATG16L1 contains a typical FIR motif for directly interacting with the Claw domain of FIP200. Importantly, in addition to binding to FIP200 Claw, ATG16L1 FIR can also serve as a noncanonical AIM motif to selectively recognize mammalian ATG8 family proteins. Moreover, our determined crystal structures of the ATG16L1 FIR/FIP200 Claw complex and the ATG16L1 FIR/GABARAPL1 complex not only uncover the detailed molecular mechanism governing the specific interactions of ATG16L1 with FIP200 and ATG8 family members, but also reveal that FIP200 and ATG8 family proteins are mutually exclusive in binding to ATG16L1. On this basis, we devise a specific ATG16L1 mutant that can interact well with ATG8 family proteins but not FIP200. Finally, using this ATG16L1 mutant together with relevant cell-based functional assays, we demonstrate that the interaction of ATG16L1 FIR with FIP200 is indispensable for the effective autophagic flux in canonical autophagy.

Results

Biochemical characterizations of the interaction of ATG16L1 with FIP200 Claw

Previous studies uncovered that the ATG16L1(229–242) region adjacent to the WIPI2-binding site 1 (WBS1) of ATG16L1 is essential for the interaction between ATG16L1 and FIP200 (Fig. 1a)9,10. Consistently, using size-exclusion chromatography (SEC)-based assays, we demonstrated that the highly conserved central region of ATG16L1, the ATG16L1(78–247) fragment (Supplementary Fig. 1), can readily interact with FIP200(1490–1594) (hereafter referred to as FIP200 Claw) (Fig. 1b). Further multi-angle light-scattering analysis revealed that the purified ATG16L1(78–247)/FIP200 Claw complex forms a stable 2:2 stoichiometric complex in solution (Fig. 1c). Strikingly, detailed sequence alignment analysis of the FIP200-binding region of ATG16L1 with that of NAP1, SINTBAD, CCPG1, NDP52, p62, NBR1, Optineurin and TNIP1, all of which were demonstrated to directly interact with FIP200 Claw27,28,29,30,31,37,38,39,40, unraveled that the ATG16L1(235–247) region conforms to the criteria for a FIP200 Claw-binding FIR motif27, likely representing a putative FIR motif (hereafter referred to as ATG16L1 FIR) (Fig. 1d). Indeed, further quantitative fluorescent polarization (FP)-based assay revealed that ATG16L1 FIR can specifically interact with FIP200 Claw with a Kd value of about ~1.33 μM (Fig. 1e). In contrast, the removal of ATG16L1 FIR totally disrupted the association of ATG16L1(78–247) with FIP200 Claw (Fig. 1f). Taken together, all these data clearly demonstrated that ATG16L1 contains a conserved FIR motif to directly interact with the Claw domain of FIP200.

Fig. 1: Biochemical characterizations of the interaction of ATG16L1 with FIP200.

a A schematic diagram showing the domain organizations of FIP200, ATG16L1, and mammalian ATG8 family proteins. In this drawing, the interactions of ATG16L1 with FIP200 and mammalian ATG8 family proteins are highlighted and indicated by two-way arrows. b Size-exclusion chromatography-based analysis of the interaction of FIP200 Claw with Trx-tagged ATG16L1(78–247). In this panel, the “Sum” stands for the theoretical sum of Trx-tagged ATG16L1(78–247) and FIP200 Claw profiles, while “Mixture” stands for the Trx-tagged ATG16L1(78–247) and FIP200 Claw mixture sample. c Multi-angle light-scattering analysis of the purified ATG16L1(78–247)/FIP200 Claw complex showing the relative light-scattering signals as a function of elution volume. The molecular mass error is the fitted error obtained from the data analysis software. d Sequence alignment analysis of the FIR of ATG16L1 with the currently known FIP200 Claw-binding regions of NAP1, SINTBAD, CCPG1, NDP52, p62, NBR1, Optineurin, and TNIP1 from the human species. In this alignment, the highly conserved acidic residues (Asp, Glu, or potentially phosphorylated Ser residue) and the following two conserved hydrophobic residues are boxed and highlighted with black triangles and stars, respectively. e The fluorescence polarization (FP)-based assay measuring the binding affinity of FIP200 Claw with FITC-labeled ATG16L1 FIR. The Kd value is the fitted dissociation constant with standard errors obtained by using the one-site binding model to fit the FP data. f Size-exclusion chromatography-based analysis of the interaction of FIP200 Claw with Trx-tagged ATG16L1(78–235). In this panel, the “Sum” stands for the theoretical sum of Trx-tagged ATG16L1(78–235) and FIP200 Claw profiles, while “Mixture” stands for the Trx-tagged ATG16L1(78–235) and FIP200 Claw mixture sample.

The overall structure of the ATG16L1 FIR/FIP200 Claw complex

Then, we intended to determine the ATG16L1 FIR/FIP200 Claw complex structure to uncover how ATG16L1 FIR binds to the Claw domain of FIP200. After numerous attempts, we finally managed to solve the crystal structure of ATG16L1 FIR in complex with FIP200 Claw at 1.61 Å resolution (Supplementary Table 1). The determined ATG16L1 FIR/FIP200 Claw complex structure is composed of two ATG16L1 FIR molecules and a FIP200 Claw dimer, forming a distinct 2:2 stoichiometric hetero-tetramer (Fig. 2a), in line with our aforementioned multi-angle light-scattering result (Fig. 1c). In the complex structure, except for a distinct N-terminal α-helix that was only found in one of the two FIP200 Claw domains owing to crystal packing (Supplementary Fig. 2a), the two monomeric FIP200 Claw domains adopt a highly similar core architecture assembled by six β-strands and one α-helix (Supplementary Fig. 2b). In parallel, the two ATG16L1 FIR molecules in the complex structure mainly form two short β-strands that symmetrically augment the β4-strand of two FIP200 Claw domains in an anti-parallel manner (Fig. 2a). Further structural comparison analyses showed that the overall structure of the monomeric FIP200 Claw domain in the ATG16L1 FIR/FIP200 Claw complex is highly akin to that of the apo-form FIP200 Claw domain (PDB ID: 6DCE) (Supplementary Fig. 2c), whereas the binding of ATG16L1 FIR to FIP200 Claw induces an obvious conformational rearrangement of the FIP200 Claw dimer (Fig. 2b). Notably, similar phenomena were also observed in our previous studies27,31.

Fig. 2: The molecular mechanism of FIP200 and ATG16L1 interaction.

a Ribbon diagram showing the overall structure of the dimeric FIP200 Claw/ATG16L1 FIR complex. In this drawing, two FIP200 Claw molecules are colored in orange and marine, while two bound ATG16L1 FIR motifs are colored in magenta and hot pink, respectively. b Ribbon representation showing the structural comparison of apo-form FIP200 Claw dimer (green, PDB ID: 6DCE) with the FIP200 Claw/ATG16L1 FIR complex (blue/magenta). In this drawing, the two dimeric structures are overlaid by aligning selected one FIP200 Claw monomer in these two structures. c Ribbon-stick model showing the detailed interactions between FIP200 Claw and ATG16L1 FIR. The hydrogen bonds and salt bridges involved in the interaction are shown as dotted lines. d The combined surface charge representation and the ribbon-stick model showing the charge-charge interactions between FIP200 Claw and ATG16L1 FIR in the solved complex structure.

The binding interface between ATG16L1 FIR and FIP200 Claw

In the ATG16L1 FIR/FIP200 Claw complex structure, each ATG16L1 FIR molecule packs extensively with a highly electropositive and hydrophobic concave groove that is situated adjacent to the β4/β5 connecting region of the monomeric FIP200 Claw, burying a total interface area of ~505 Å2 (Fig. 2a). Further careful analyses of the molecular interface in the ATG16L1 FIR/FIP200 Claw complex structure revealed that the hydrophobic side chain of ATG16L1 I240 deeply inserts into a hydrophobic pocket formed by the hydrophobic side chains of C1565, A1567, F1574, and F1582 residues together with the aliphatic side chain group of R1573 residue of FIP200 Claw (Fig. 2c and Supplementary Fig. 2d), and concurrently, the hydrophobic side chains of ATG16L1 I243 and V244 residues form hydrophobic contacts with the aromatic side chain of Y1564 and the aliphatic side chain group of K1581 from FIP200 (Fig. 2c). Furthermore, the negatively charged D238, D239 and E241 residues of ATG16L1 FIR form specific hydrogen bonding and charge-charge interactions with the positively charged K1569, R1573, and K1568 residues of FIP200 Claw (Fig. 2c, d). In addition, the backbone groups of E241, I243, and V244 residues of ATG16L1 FIR couple with the backbone groups of Q1566 and Y1564 residues of FIP200 to form five specific backbone hydrogen bonds (Fig. 2c). In keeping with their critical structural roles, all these key binding interface residues of ATG16L1 and FIP200 are highly conserved in different eukaryotic species (Supplementary Figs. 1 and 3). Consistent with our structural results, further FP-based assays showed that point mutations of key interface residues of FIP200 Claw, such as the K1568A, K1569A, R1573E, and F1574Q mutations, all significantly decrease the specific interaction of FIP200 Claw with ATG16L1 FIR (Table 1 and Supplementary Fig. 4). Reciprocally, point mutations of key interface residues from ATG16L1 FIR, including the D238R, D239R, I240Q, E241R, and I243Q mutations of ATG16L1 FIR, all largely attenuate or essentially disrupt the interaction between FIP200 Claw and ATG16L1 FIR (Supplementary Fig. 5). Therefore, all those mutagenesis-based biochemical assays confirmed the specific interaction between ATG16L1 FIR and FIP200 Claw observed in the solved ATG16L1 FIR/FIP200 Claw complex structure.

Biochemical characterizations of the interactions of ATG16L1 with ATG8 family proteins

Since the reported FIR motifs have been demonstrated to directly engage with mammalian ATG8 family proteins27,31, we suspected that ATG16L1 FIR may also recognize mammalian ATG8 orthologs. Indeed, our ITC analyses revealed that ATG16L1 FIR can directly interact with six mammalian ATG8 members, and preferentially binds to GABARAPL1 and LC3C with a Kd value of ~4.59 μM and ~6.27 μM, respectively (Fig. 3a, f and Supplementary Fig. 6). Then, we chose GABARAPL1 as a representative to further characterize the binding mechanism of ATG16L1 FIR with ATG8 family proteins. Our SEC results showed that the FIR-containing ATG16L1(78–247) and ATG16L1(235–247) fragments can readily interact with GABARAPL1 (Supplementary Fig. 7a, b). In contrast, the ATG16L1(78–235) fragment, which lacks the FIR motif of ATG16L1, is completely unable to interact with GABARAPL1 (Supplementary Fig. 7c). Concurrently, we also utilized NMR to validate the interaction between GABARAPL1 and ATG16L1. Titrations of 15N-labeled GABARAPL1 with un-labeled ATG16L1(235–247) or ATG16L1(78–247) proteins showed that many peaks in the 1H-15N HSQC spectra of GABARAPL1 undergo significant chemical shift changes or peak-broadenings, confirming that the FIR motif-containing ATG16L1 fragments can directly bind to GABARAPL1 (Supplementary Fig. 8). Taken together, all these data clearly demonstrated that ATG16L1 FIR can function as an unconventional AIM motif to directly recognize mammalian ATG8 orthologs.

Fig. 3: Biochemical and structural characterizations of the interactions of ATG16L1 FIR with mammalian ATG8 family proteins.

a ITC-based measurement of the binding affinity of GABARAPL1 with Trx-tagged ATG16L1 FIR. b Ribbon diagram showing the overall structure of the GABARAPL1/ATG16L1 FIR complex. In this drawing, the GABARAPL1 molecule is colored in green, while ATG16L1 FIR in magenta. c The ribbon-stick model showing the detailed interactions between GABARAPL1 and ATG16L1 FIR. The hydrogen bonds and salt bridges involved in the interaction are shown as dotted lines. d The combined surface charge representation and the ribbon-stick model showing the charge-charge interactions between GABARAPL1 and ATG16L1 FIR. e The combined surface representation and the ribbon-stick model showing the hydrophobic interactions between GABARAPL1 and ATG16L1 FIR. In this drawing, ATG16L1 FIR is displayed in the ribbon-stick model, and GABARAPL1 is shown in surface representation, colored by different amino acid types. Specifically, the hydrophobic amino acid residues in the surface model of GABARAPL1 are drawn in yellow; the positively charged residues are drawn in blue; the negatively charged residues are drawn in red, and the uncharged polar residues are drawn in gray.

The structure of ATG16L1 FIR in complex with GABARAPL1

To further uncover the detailed molecular mechanism underlying the selective recognitions of ATG8 family proteins by ATG16L1 FIR, we determined the high-resolution crystal structure of GABARAPLI in complex with ATG16L1 FIR (Supplementary Table 1). In the determined GABARAPL1/ATG16L1 FIR complex structure, GABARAPL1 adopts a typical ATG8 family protein architecture assembled by a ubiquitin-like structural core and two preceding N-terminal α-helices (Fig. 3b). In the complex structure, the clearly defined ATG16L1 FIR motif contains 10 highly conserved residues spanning from Q236 to D245 of ATG16L1 (Supplementary Fig. 9a), and adopts an extended configuration to occupy the canonical AIM-binding groove of GABARAPL1, which is mainly formed by the β1-strand, β2-strand, α2-helix and α3-helix of GABARAPL1 (Fig. 3b, c). Further structural analyses of the binding interface of the GABARAPL1/ATG16L1 FIR complex revealed that the interaction between GABARAPL1 and ATG16L1 FIR is mainly mediated by both polar and hydrophobic interactions (Fig. 3c). In particular, the negatively charged side chains of ATG16L1 D239, E241 interact with the positively charged side chains of K48, K46 and R67 of GABARAPL1 to form three charge-charge interactions, and the backbone amide and oxygen groups of ATG16L1 E241 together with the backbone amide group of ATG16L1 I243 form three backbone hydrogen bonds with the GABARAPL1 K48 and L50 residues (Fig. 3c, d). In addition, the GABARAPL1/ATG16L1 FIR interaction is further strengthened by four specific hydrogen bonds, two of which are formed between the backbone oxygen groups of D238 and I240 residues of ATG16L1 and the positively charged side chain of GABARAPL1 K46, while the other two are mediated by the backbone oxygen group of ATG16L1 I243 and the positively charged side chain of GABARAPL1 R28 (Fig. 3c). In parallel, the hydrophobic side chains of ATG16L1 I240 and V242 residues pack against a hydrophobic groove of GABARAPL1 formed by the hydrophobic side chains of I21, P30, L50, F104 and the aliphatic portion of the side chain of GABARAPL1 K48 residue (Fig. 3c, e). Moreover, the hydrophobic side chain of ATG16L1 I243 occupies a hydrophobic pocket assembled by the side chains of Y49, V51, P52, L55, F60, and L63 residues of GABARAPL1 (Fig. 3c, e). Consistently, all these key binding interface residues of ATG16L1 FIR and GABARAPL1 are highly conserved during evolution (Supplementary Figs. 1 and 10). Using ITC-based assays, we further verified the specific interactions between interface residues of GABARAPL1 and ATG16L1 FIR observed in the complex structure. Individual point mutations of key residues involved in the binding interface of GABARAPL1/ATG16L1 FIR complex either from GABARAPL1 or ATG16L1 FIR, such as the I21Q, K48E, L50Q and R67E mutations of GABARAPL1 (Table 2 and Supplementary Fig. 11), or the D239R, I240Q, E241R, V242E and I243Q mutations of ATG16L1 FIR (Table 2 and Supplementary Fig. 12), all dramatically decrease or essentially disrupt the association of GABARAPL1 with ATG16L1 FIR. Meanwhile, based on the previous backbone assignment of GABARAPL1 (BMRB Entry 17412)41, we assigned the NMR peaks in the 1H-15N HSQC spectrum of the 15N-labeled GABARAPL1 and quantified the relevant chemical shift changes in our NMR titration experiments (Supplementary Figs. 8a and 13a). Further mapping the chemical shift differences onto our crystal structure of the GABARAPL1/ATG16L1 FIR complex revealed that the significant backbone amide chemical shift changes are only rich in the ATG16L1 FIR-binding regions of GABARAPL1 (Supplementary Fig. 13b), confirming that ATG16L1 FIR binds to the canonical AIM-binding sites of GABARAPL1 in solution. Notably, previous studies showed that the core unconventional AIM motif of NAP1, NDP52, or TAX1BP1 consists of an acidic Asp followed by four consecutive hydrophobic residues, all of which participate in hydrophobic interactions with the relevant ATG8 family proteins26,31,42. Interestingly, unlike that of the unconventional AIM motifs of NAP1, NDP52, and TAX1BP1, the third residue of the core unconventional AIM motif of ATG16L1 is an acidic Glu residue (Supplementary Fig. 9b), which is directly involved in the interaction with GABARAPL1 (Fig. 3c). Therefore, ATG16L1 FIR represents a distinct type of unconventional AIM motif.

The relationships of FIP200, GABARAPL1, and WIPI2 in binding to ATG16L1

Based on our aforementioned structural analyses, ATG16L1 FIR adopts essentially the same key residues to recognize FIP200 Claw and GABARAPL1, such as the D239, I240, E241, and I243 residues (Figs. 2c and 3c and Supplementary Fig. 1). Thus, FIP200 and ATG8 family proteins should be mutually exclusive in binding to ATG16L1 FIR. As expected, further SEC coupled with SDS-PAGE assays confirmed that FIP200 Claw and GABARAPL1 are competitive in binding to ATG16L1 (Fig. 4a, b). Given that ATG16L1 FIR is C-terminally adjacent to the WBS1 of ATG16L1 (Fig. 1a), we also tested the relationship between WIPI2 and FIP200 or GABARAPL1 in binding to ATG16L1. Intriguingly, we revealed that FIP200 Claw but not GABARAPL1 can form a stable ternary complex with WIPI2 and the ATG16L1(207–247) fragment that contains both WBS1 and FIR (Supplementary Fig. 14a), suggesting that WIPI2 and FIP200 can simultaneously bind to ATG16L1, consistent with a previous study11. In contrast, WIPI2 and ATG8 family proteins, such as GABARAPL1, are competitive in binding to ATG16L1(207–247) (Supplementary Fig. 14b), likely due to the potential steric hindrance.

Fig. 4: Biochemical characterizations of the interactions of ATG16L1 variants with FIP200 Claw and GABARAPL1.

a Size-exclusion chromatography coupled with SDS-PAGE analysis of the Trx-ATG16L1(78–247)/GST-GABARAPL1 complex incubated with increasing molar ratio of FIP200 Claw proteins. b The SDS-PAGE combined with Coomassie-blue staining analyses of the protein components of the indicated “fraction 1” and “fraction 2” fractions collected from the analytical gel filtration chromatography experiment at different molar ratios of FIP200 Claw in (a). c, d Size-exclusion chromatography coupled with SDS-PAGE analysis of Trx-tagged ATG16L1 FIR wild-type with GABARAPL1 or FIP200 Claw. In this panel, “Sum” stands for the theoretical sum of Trx-tagged ATG16L1 FIR wild-type and GABARAPL1 or FIP200 Claw profiles, while “Mixture” stands for the Trx-tagged ATG16L1 FIR wild-type and GABARAPL1 or FIP200 Claw mixture sample. e, f Size-exclusion chromatography coupled with SDS-PAGE analysis of Trx-tagged ATG16L1 FIR D239R/I240F mutant with GABARAPL1 or FIP200 Claw. In this panel, “Sum” stands for the theoretical sum of Trx-tagged ATG16L1 FIR D239R/I240F mutant and GABARAPL1 or FIP200 Claw profiles, while “Mixture” stands for the Trx-tagged ATG16L1 FIR D239R/I240F mutant and GABARAPL1 or FIP200 Claw mixture sample. g, h Size-exclusion chromatography coupled with SDS-PAGE analysis of Trx-tagged ATG16L1 FIR I240Q/I243Q mutant with GABARAPL1 or FIP200 Claw. In this panel, “Sum” stands for the theoretical sum of Trx-tagged ATG16L1 FIR I240Q/I243Q mutant and GABARAPL1 or FIP200 Claw profiles, while “Mixture” stands for the Trx-tagged ATG16L1 FIR I240Q/I243Q mutant and GABARAPL1 or FIP200 Claw mixture sample.

The rational design of selective ATG16L1 mutant for binding to GABARAPL1 but not FIP200

Based on our previous study27, the consensus FIR motif (Ψ-Θ-Χ1-Χ2-Φ, where Ψ represents a phosphorylated Ser/Thr residue or an acidic Asp, Glu, Θ represents a bulk hydrophobic Ile, Leu, Met or aromatic Phe, Tyr, Trp residue, Φ represents a hydrophobic Leu, Ile or Val residue, and Χ1/Χ2 represents any residue), bears a striking resemblance to the sequence pattern of the core AIM sequence. Actually, the similarity between FIR and AIM severely interferes with our assessment of the individual contribution made by FIP200 and ATG8 family proteins when binding to ATG16L1 FIR. Therefore, it is necessary to develop selective ATG16L1 mutants that can exclusively interact with FIP200 or ATG8 family proteins. Based on our previous biochemical and structural characterizations of relevant FIR and AIM motifs27,43, we realized that the first residue of the consensus core sequence of AIM, which is corresponding to the Ψ residue of FIR, can tolerate basic residues to some extent, such as Arg residue, while FIR cannot. In addition, the Θ residue of AIM prefers an aromatic Phe, Tyr, or Trp residue rather than a hydrophobic Ile, Leu, or Met residue. Eventually, we managed to devise a selective ATG16L1 D239R/I240F (DRIF) double mutant that can well recognize most ATG8 family proteins but not FIP200 (Fig. 4c–f and Supplementary Fig. 15). In addition, we also obtained a ATG16L1 I240Q/I243Q (IQIQ) double mutant, which essentially binds neither ATG8 family proteins nor FIP200 (Fig. 4c, d, g, h and Supplementary Fig. 15). Importantly, in agreement with our biochemical data (Fig. 4c–h and Supplementary Fig. 15), further overexpression-based co-immunoprecipitation assays showed that the DRIF mutant can selectively bind to GABARAPL1 but not FIP200 in cells (Fig. 5a, b). Concomitantly, the IQIQ mutation of ATG16L1 completely abolishes the specific interactions of ATG16L1 with FIP200 and GABARAPL1 in cells (Fig. 5a, b). Notably, in order to avoid potential ATG8ylation of ATG16L1 in cells44, the C-terminal Gly-Lys of GABARAPL1 was removed in these co-immunoprecipitation assays. Unfortunately, albeit with numerous attempts, we failed to obtain an ATG16L1 mutant, which can only bind to FIP200 but not GABARAPL1 with a comparable binding ability as the wild-type ATG16L1.

Fig. 5: Cell-based assays of the interactions of ATG16L1 with GABARAPL1 and FIP200 in autophagy.

a Co-immunoprecipitation assays showing that point mutations of key interface residues of GABARAPL1 observed in the GABARAPL1/ATG16L1 FIR complex structure essentially disrupt their specific interaction in cells. “IB” means immunoblotting. b Co-immunoprecipitation assays showing that point mutations of both devised ATG16L1 mutant and key interface residues of FIP200 observed in the FIP200 Claw/ATG16L1 FIR complex structure decrease or essentially disrupt the specific interaction between FIP200 and ATG16L1 in cells. c Western blot-based measurements of the LC3B lipidation and p62 degradation levels in ATG16L1-knockout HeLa cells (16KO) as well as that rescued with mEGFP-tagged WT ATG16L1 (16KO + WT), ATG16L1 D239R/I240F mutant (16KO + DRIF), or ATG16L1 I240Q/I243Q mutant (16KO + IQIQ) treated for 4 h using D10 + PS normal medium (F), D10 + PS normal medium treated with bafilomycin A1 at 400 nM (F + B), amino acid starvation medium (S), or amino acid starvation medium treated with bafilomycin A1 at 400 nM (S + B). d The levels of p62 and β-actin in (c) were quantified in ImageJ and normalized to that of 16KO cells treated with S + B. The data is presented as means ± SEM from four independent experiments. e The levels of LC3B-II and β-actin in (c) were quantified in ImageJ and normalized to the 16KO + WT cells treated with S + B. The data is presented as means ± SEM from four independent experiments. f Western blot-based measurements of the LC3B lipidation in ATG16L1-knockout HeLa cells (16KO) as well as that rescued with mEGFP-tagged WT ATG16L1 (16KO + WT), ATG16L1 D239R/I240F mutant (16KO + DRIF), ATG16L1 I240Q/I243Q mutant (16KO + IQIQ), ATG16L1 K490A mutant (16KO + K490A), ATG16L1 K490A/D239R/I240F mutant (16KO + K490A/DRIF), or ATG16L1 K490A/I240Q/I243Q mutant (16KO + K490A/IQIQ) treated for 1 h using D10 + PS normal medium (F), D10 + PS normal medium treated with monensin at 100 μM (M), or D10 + PS normal medium treated with monensin at 100 μM and bafilomycin A1 at 400 nM (M + B). g The levels of LC3B-II and β-actin in (f) were quantified and normalized to that of 16KO + DRIF cells treated with monensin at 100 μM (M). The data are presented as means ± SEM from three independent experiments. Statistical analyses were all performed in GraphPad Prism 9 by two-way analysis of variance (ANOVA) followed by Bonferroni multiple comparisons test, and P value style is P = 0.1234 [not significant (ns)], *P = 0.0332, **P = 0.0021, ***P = 0.0002, and ****P < 0.0001. h A proposed model depicting the functions of the FIP200/ATG16L1 and ATG8s/ATG16L1 interactions in canonical autophagy.

The essential role of ATG16L1 FIR for the effective autophagic flux in canonical autophagy

To further unravel the functional relevance of ATG16L1 FIR in autophagy, we back-transfected the ATG16L1-knockout HeLa cell line, which was generated in our previous study12, with relevant ATG16L1 plasmids to stably express the wild-type ATG16L1 or relevant ATG16L1 mutants, such as the DRIF mutant of ATG16L1 that only loses the FIP200-binding ability, or the ATG16L1 IQIQ mutant that simultaneously loses the abilities for interacting with FIP200 and ATG8 family proteins. In line with our previous study12, the autophagic flux in the ATG16L1-knockout cells was effectively rescued by the wild-type ATG16L1 (Fig. 5c–e). Meanwhile, starvation-induced lipidation of LC3B is unperturbed by the ATG16L1 DRIF or IQIQ mutation (Fig. 5c, e). However, both ATG16L1 DRIF and ATG16L1 IQIQ mutations significantly impair the autophagic degradation of p62 (Fig. 5c, d), underscoring an indispensable role of ATG16L1 FIR in starvation-induced autophagy. Notably, the DRIF mutation of ATG16L1 induces a much more severe impediment of the p62 degradation than the ATG16L1 IQIQ mutation (Fig. 5c, d), implying that the interaction of ATG16L1 FIR with FIP200 facilitates the autophagy flux, whereas the interactions of ATG16L1 FIR with ATG8 family proteins attenuate the autophagy process. Taken together, these cell-based functional data clearly demonstrated that the interaction between ATG16L1 FIR and FIP200 is essential for the effective autophagic flux in canonical autophagy.

Discussion

In this work, we uncovered that in addition to interacting with FIP200 Claw, ATG16L1 FIR can also serve as an unconventional AIM to recognize ATG8 family proteins. Furthermore, our biochemical and structural analyses revealed that ATG16L1 FIR can selectively bind to six mammalian ATG8 family proteins, but preferentially bind to GABARAPL1 and LC3C (Fig. 3a, f and Supplementary Fig. 6). Of note, detailed sequence alignment analysis elucidated that several key interface residues for interacting with ATG16L1 are quite different among six mammalian ATG8 family proteins (Supplemen