Three-step docking by WIPI2, ATG16L1 and ATG3 delivers LC3 to the phagophore

The covalent attachment of ubiquitin-like LC3 proteins prepares the autophagic membrane for cargo recruitment. We resolve key steps in LC3 lipidation by combining molecular dynamics simulations and experiments in vitro and in cellulo. We show how the E3-like ligase ATG12– ATG5-ATG16L1 in complex with the E2-like conjugase ATG3 docks LC3 onto the membrane in three steps by (1) the PI(3)P effector protein WIPI2, (2) helix α2 of ATG16L1, and (3) a membrane-interacting surface of ATG3. Phosphatidylethanolamine (PE) lipids concentrate in a region around the thioester bond between ATG3 and LC3, highlighting residues with a possible role in the catalytic transfer of LC3 to PE, including two conserved histidines. In a near-complete pathway from the initial membrane recruitment to the LC3 lipidation reaction, the three-step targeting of the ATG12–ATG5-ATG16L1 machinery establishes a high level of regulatory control.


Introduction
vitro 15 in the absence of its cognate E3, but in vivo 16 and even in a giant unilamellar vesicle (GUV) reconstitution system 17,18 , the downstream E3 complex components are essential.
The autophagic counterpart of the Ub E3 is the ATG12-ATG5-ATG16L1 complex 19 , which is structurally and evolutionarily unrelated to any of its functional equivalents in ubiquitylation. The ATG12-ATG5 unit is itself covalently bonded through an ATG10-dependent reaction 20 . ATG12-ATG5 binding allosterically activates ATG3 by increasing the exposure and reactivity of its Cys264-linked ATG8 thioester for transfer to PE 14,21 . The ATG16L1 portion of the complex is responsible for delivery and positioning on the membrane 19 . ATG16L1 is itself delivered to membranes by the β-propeller protein WIPI2 18,22 . WIPI2 (and other WIPIs) are recruited to membranes early in autophagy induction by the lipid phosphatidylinositol 3-phosphate (PI3P) 23 , which is generated by the class III PI 3-kinase complex I (PI3KC3-C1) early in autophagy initiation 24 .
The problem of how the chemistry and structural biology of a protein ubiquitylation-like system is adapted to act on a membrane substrate has been one of the major open questions in the mechanistic biochemistry of autophagy. A number of pieces of the puzzle have come together in recent years. The structural basis of the assembly of a fragment of ATG3 with the ATG12-ATG5-ATG16L1 unit was worked out for the human proteins 21 . ATG16L1 contains an amphipathic helix ɑ2, adjacent to its ATG5 binding site, that is strongly sensitive to membrane curvature 25 and essential for promotion of LC3 lipidation in liposomes and in cells 26 . It is puzzling that ATG16L1 ɑ2 is so important for catalysis, given that WIPI2 is capable of recruiting ATG16L1 to membranes through its WIPI2-interacting region (W2IR) 18,22 . The structural basis for human ATG16L1 recruitment by WIPI2 has also been worked out 18,27 . The ATG12-ATG5 and WIPI2-binding regions of ATG16L1 are separated by a coiled coil with > 100 amino acids. The resulting extended shape and its flexibility challenge experimental structure determination of the full membranebound WIPI2-ATG12-ATG5-ATG16L1-ATG3 system. Here, we approached the problem beginning with large-scale all-atom simulations of the WIPI2-ATG12-ATG5-ATG16L1-ATG3 on lipid membrane. Predictions from the simulations were verified experimentally in vitro and in cellulo. In this way, we connect structural and biochemical information into a near-complete view of the lipidation pathway.

Results
Docking step 1: WIPI2 recruits ATG12-ATG5-ATG16L1 loaded with ATG3-LC3 to phagophore As a key first step in targeting the lipidation machinery to the phagophore membrane, we concentrated on the WIPI2-mediated membrane interaction of ATG12-ATG5-ATG16L1. The central homodimer-forming coiled-coil domain (residues 78-230) of the human ATG16L1 protein is predicted 28 to form a continuous stretch of α-helical coiled coils spanning the major part of the domain (~115 amino acids from the N-terminal side), allowing reconstruction of the dimeric ATG16L1 structure by fitting geometric parameters 29 based upon Crick's equations 30 (SI Fig . 1A).
The resulting coiled-coil structure is in excellent agreement with crystal structures 31,32 of the mouse orthologue in which an overlapping region of the coiled-coil domain has been resolved (SI Fig.   1A), providing validation for our ATG16L1 model. Using AlphaFold 33,34 , we also obtained a structural model of the E2-like ATG3 conjugase loaded with LC3 (SI Fig. 1B). The predicted ATG3-LC3 complex adopts a conformation compatible both with binding to the E1-like ATG7 homodimer in the preceding step and with formation of a thioester bond between the catalytic Cys264 side chain of ATG3 and the C-terminal Gly120 of LC3 to yield the E2-substrate conjugate (SI Fig. 1B). The core of the human ATG3 structure is architecturally similar to the yeast and Arabidopsis Atg3 proteins, as has been previously reported 35 , with an intrinsically disordered region 36 forming an ~100-residue loop that contains the ATG12-binding sequence 21 (Fig. 1A) as well as a region predicted to participate in β-sheet formation in the presence of LC3. Combined with crystallographic structures 21,37 of ATG12-ATG5 in quaternary complex with a bound fragment of ATG3 and the N-terminal ATG5-binding domain of ATG16L1, we present an atomistic model of the full LC3 lipidation machinery consisting of the E3-like ATG12-ATG5-ATG16L1 complex bound to the E2-substrate conjugate, ATG3-LC3 (Fig. 1A).
To determine the configuration of the ATG12-ATG5-ATG16L1 complex recruited to phagophore membranes by the PI(3)P effector WIPI2, we first performed atomistic molecular dynamics (MD) simulations on a WIPI2-ATG16L1 co-crystal structure 18 in which WIPI2 is bound to the WIPI2-interacting region of ATG16L1 (residues 207-230). Initially placed at a minimum distance of ~2 nm above PI(3)P-containing membranes mimicking the ER lipid composition 38 , WIPI2 formed spontaneous membrane contacts in an expected orientation, with the two putative phosphoinositide binding sites [39][40][41] in its β-propeller blades 5 and 6 interacting with PI(3)P (Fig.   1A) and the N-terminal side of the bound ATG16L1 segment oriented away from the membrane.
By aligning our structural model of ATG12-ATG5-ATG16L1 described above to the membraneassociated WIPI2-ATG16L1 configuration established during simulations, a first model of the membrane-recruited LC3 lipidation machinery was thus obtained.
Upon atomistic MD simulations of ATG12-ATG5-ATG16L1 complexed with the ATG3-LC3 conjugate and anchored via WIPI2 to membranes approximating the phagophore lipid composition, all components maintained structural integrity across all five 1 µs simulation replicates (SI Fig. 2). Flexing and tilting motions of the dimeric coiled-coil structure of ATG16L1 were accompanied by considerable flexibility exhibited in the inter-domain loop regions of ATG16L1 and ATG3 (Fig. 1B), allowing the ATG3-LC3 conjugate to explore favorable binding configurations near the membrane (SI Movie 1). However, membrane binding was observed only in the case where ATG3-LC3 was already at the membrane surface upon initiation of the simulation. This finding indicates that the upward tilt of the WIPI2-attached coiled coil tends to keep the ATG3-LC3 conjugate above the membrane, even if direct interactions of ATG3-LC3 are possible in principle.
To reconcile the prevailing model of membrane recruitment of ATG12-ATG5-ATG16L1 by WIPI2 with the requirement for the membrane-interacting ATG16L1 N-terminal helix α2 26 , we hypothesized that upon initial recruitment through WIPI2, direct membrane binding by helix α2 constitutes a crucial second step in delivering ATG3-LC3 nearer to the target membrane. We focus here on the cis configuration in which the entire LC3 lipidation machinery becomes associated with the same patch of membrane 42 . However, our molecular model does not rule out the alternative possibility whereby ATG12-ATG5-ATG16L1 anchored at omegasomal membranes would bridge an intermembrane distance to facilitate LC3 conjugation to the nascent phagophore in trans 22 .

Docking step 2: Helix α2 of ATG16L1 pulls ATG3-LC3 to membrane
As a possible second step in membrane targeting, we focused on helix α2 of the ATG16L1 Nterminal domain, which has been shown to bind membranes 26 with a preference for positive membrane curvature 25 . For an ATG12-ATG5-ATG16L1 complex attached to the membrane via WIPI2, this mode of membrane interaction is made possible by the flexibility of the ~30-residue inter-domain loop between the N-terminus of ATG16L1 and its coiled coil. We demonstrated this ability of ATG16L1 to engage with the membrane simultaneously, at one end, through recruitment by WIPI2 and, at the other end, via helix α2 by gently pulling the ATG16L1 helix α2 toward the membrane in steered MD simulations and then relaxing the complex in extended MD simulations ( Fig. 2A).
Building upon our previous MD simulations of ATG12-ATG5-ATG16L1 binding to curved membranes 25 , with ATG16L1 interacting either at the membrane surface or with an embedded hydrophobic face of the amphipathic helix α2, we obtained models of membrane-bound ATG12-ATG5-ATG16L1 loaded with ATG3-LC3 (Fig. 2B). In atomistic MD simulations, the flexible ATG3 loop then allowed ATG3-LC3 to reach the membrane spontaneously whilst maintaining its interactions with the α2-anchored ATG12-ATG5-ATG16L1 complex (SI Movie 2). In this configuration, we found the N-terminal helix of the ATG3 conjugase to embed into the membrane ( Fig. 2C and SI Movie 2) without any biasing force. Once inserted, the ATG3 amphipathic helix remained embedded in the membrane through the course of the simulation, establishing stable membrane contacts. This finding is consistent with the previously reported role of the ATG3 N-terminus as an essential membrane-targeting element 43 . Geometrically, the two ATG3-LC3 conjugates flexibly connected to the ATG12-ATG5-ATG16L1 complex can simultaneously engage with the membrane for parallel lipidation reactions.

Docking step 3: Catalytic domain of ATG3 forms stable membrane interaction interface upon membrane insertion of ATG3 N-terminal helix
For a decisive third and final targeting step, we explored how the catalytic domain established membrane contact. The ATG3 conjugase has been reported to show basal activity in vitro for catalyzing LC3 conjugation in the absence of ATG12-ATG5-ATG16L1 21 . Having observed stable membrane association of the ATG3-LC3 conjugate held near phagophore-mimetic membranes by ATG12-ATG5-ATG16L1, we sought to further collect lipid contact data on ATG3-LC3 by initiating a set of longer (2 µs) replicates of smaller simulation systems containing the isolated conjugate placed directly above membranes. In eight of the twenty trajectories thus obtained, spontaneous membrane insertion of the ATG3 N-terminal helix occurred within the first 1 µs.
Comparison of ATG3-LC3 lipid contacts (post-insertion of the ATG3 helix) reveals a consistent membrane interaction interface in presence or absence of ATG12-ATG5-ATG16L1 (Fig. 3A). We found that the ATG3 protein dominated the interactions of the conjugate with PE-containing membranes.

ATG3-LC3 presents active site towards the membrane in configuration conducive to lipidation reaction
With ATG3-LC3 at the membrane, we explored the structural foundation of the actual lipidation reaction. The folded core of ATG3 comprises a six-stranded β-sheet (strands β1-β6) surrounded by α-helices 44 . Among regions of ATG3-LC3 that formed frequent membrane interactions in our simulations were short sequences of residues within inter-secondary structure loops of the ATG3 core, namely (i) catalytic domain residues 208-211 and 242-243 of the β3/β4 and β4/β5 loops, respectively; (ii) residues 262-265 encompassing the thioester-forming Cys264 between β6 and the succeeding α-helix; and to a lesser extent (iii) residues 61-64 within the β1/β2 loop. The catalytic site, which contains Cys264 of ATG3 covalently bonded to the C-terminus of LC3, was situated centrally on the membrane interaction interface identified above and exposed towards membrane lipids (Fig. 3A). Furthermore, the ATG3-LC3 conjugate formed distinct interactions with different types of lipids present in the membrane, with PE localizing particularly near the catalytic center (Fig. 3B). Our data thus suggest a preferred orientation of ATG3-LC3 on the membrane that is compatible with catalyzing LC3 conjugation to the phagophore.

Mutational analysis and functional validation
The MD simulations identified a surface of ATG3 that was consistently in contact with the membrane in the context of the larger ATG12-ATG5-ATG16L1-ATG3-LC3B-WIPI2 complex.
Activity was monitored by the conversion of LC3B-I to LC3B-II. As expected, essentially complete conversion was seen for wild-type, while the mutation C264A of the catalytic cysteine as a negative controlled completely eliminated activity (Fig. 4B, C). The mutation H262A also completely abolished activity, suggesting a direct role in catalysis beyond its membrane interactions alone. This is discussed further below. Activity was nearly abolished in K208D and sharply reduced in T244A, with smaller but significant reductions seen in K62D/K64D and Y209A. Apparent reductions were seen in Y210A, R265A, and H266A, but did not rise to statistical significance. The observation that most of these mutations had at least some effect on catalysis in the SUV system confirms the predicted membrane interaction surface identified by the MD simulations.
SUV assays are relatively permissive for LC3 lipidation by ATG3, likely due to the favorable contribution of ATG3 helix α1 to curved membrane binding 15 . Giant unilamellar vesicle (GUV) assays have a more stringent dependence on regulatory cofactors 17,42 , so we repeated a subset of the above LC3B lipidation assays in this setting (Fig. 4D, E). Qualitatively, the results were similar, with C264A completely inactive, and others showing reduced activity. The rank order of the mutational effects subtly diverges from the SUV system, which is not surprising, given the differing physical presentation of the membrane to the conjugation complex. The main conclusion is that all of the mutations have at least some impact on activity in the GUV system, again consistent with the predictions of the simulations.
To determine if the predicted membrane function had the same function in living cells as in the reconstituted system, we generated an ATG3 KO HeLa cell line. ATG3 KO was verified by Western blotting (Fig. 5A). Starvation-induced autophagic flux was monitored with the HaloTag-LC3B system based on the appearance of a free HaloTag-band 17 . As expected, expression of the wild-type construct rescued autophagic flux in the KO cells, while no flux was observed in the C264A rescue (Fig. 5B, C). In a contrast to the partial effects seen for most mutants in vitro, an almost complete loss of flux was seen in most of the mutants in the ATG3 KO cells. This may reflect more stringency in the cellular system. H266A, which has a marginal (not statistically significant) reduction in activity in SUVs and a larger reduction in GUVs, has essentially no loss of activity in cells. The more variable effects of H266A in different assays may reflect a higher degree of context-dependence of the function of His266, as discussed below. The main conclusion from the ATG3 KO experiments is that the membrane interaction surface identified in the MD simulations accurately predicted loss of function in the cellular setting.

Conserved His262 of HPC motif facilitates ATG3-catalyzed LC3 lipidation
Previous studies have shown that the transfer of LC3 from ATG3 to lipid substrates is sensitive to pH and takes place more efficiently under slightly basic conditions in vitro, most likely through an effect on the ATG3 conjugase activity 47,48 . Whilst the protonation state of the target PE amine group is expected to show little variation within the pH range of interest, we note the presence of two histidine residues, His262 and His266, in close proximity to the catalytic Cys264. Both histidines are fully conserved across ATG3 homologues and, with their characteristic pKa just below physiological pH, serve as possible acidity sensors for the ATG3-catalyzed reaction.
In atomistic MD simulations of the ATG3-LC3 conjugate with the His262 and His266 side chains both in their unprotonated state (which is predicted to be the dominant species at pH ≥ 7), His266 remained oriented towards the protein interior with a minimum distance of ~1 nm to the nearest lipid (Fig. 6A). By contrast, frequent lipid interactions formed by His262 are suggestive of a direct role in the LC3 conjugation reaction. Whereas the nucleophile of the reaction, the PE amine group, did spontaneously approach the backbone carbonyl carbon of Gly120 (LC3) to be attacked (reaching a minimum distance of ~0.4 nm), such interactions were infrequent. Meanwhile, the unprotonated nitrogen of the His262 imidazole was observed to interact with the positively charged primary amine of PE headgroups within bonding distance (< 0.2 nm) to the amine proton ( Fig. 6A). Furthermore, our simulations capture a configuration in which the His262:PE interaction coincided with that between PE and Gly120 (Fig. 6B).
His262 and Cys264 of human ATG3 form part of the HPC motif that is conserved across orthologues of ATG3 as well as ATG10, an E2-like autophagic enzyme that catalyzes ATG12 conjugation to ATG5 6 . Combined with previous 35,49 and present evidence of a critical role of His262 for ATG3 conjugase activity, our simulation results are suggestive of a plausible reaction mechanism in which the His262 imidazole ring would deprotonate the PE amine group for nucleophilic attack on the Gly120 carbonyl of LC3 (Fig. 6C). As part of such a proposed mechanism, the unique backbone conformational restraints conferred by the cyclic side chain of Pro263 in the HPC motif would be crucial for orienting His262 and Cys264 side chains in relative positions conducive to catalysis, explaining their full conservation (SI Fig. 3). The protein backbone conformation conferred by Pro263 also holds the backbone amide of Cys264 within bonding distance of the carbonyl oxygen of Gly120 (Fig. 6C), which would stabilize the oxygen anion intermediate formed during the reaction. Energetically favorable breakage of the thioester bond will then yield the LC3-PE conjugate, a stable amide product. Alternatively, ATG3 has been reported to catalyze conjugation of ATG8 family proteins to phosphatidylserine (PS) lipids in the non-canonical pathway of autophagy 50 . In accordance with this, our simulations of ATG3-LC3 conjugate also capture an analogous membrane-interacting configuration likely poised for reaction with a PS molecule (SI Fig. 4).
His266, the second of the two conserved histidine residues described above, has been implicated in the pH-dependent conjugase activity of ATG3 in a recent study 35 . To assess the effect of altering the protonation state of His266, we performed additional MD simulations of the ATG3-LC3 conjugate, in which the His266 imidazole ring was doubly protonated. Strikingly, the extra proton destabilized the local protein structure (Fig. 6D). A reorientation brought the His266 side chain into direct membrane contact within the first hundreds of nanoseconds in seven out of ten simulation replicates (Fig. 6E, F). These results are consistent with His266 fulfilling a pH-sensitive structural role, as previously proposed for its counterpart in ATG3 orthologues (His236 in the yeast protein and His260 in Arabidopsis) 48 , and provide an explanation for the alternative conformations in this region between available crystal structures obtained at different pH 44,51 .

Discussion
Building upon an increasing collection of structural and biochemical data on the components and interactions that form the autophagic LC3 lipidation machinery, we set out to complete the molecular puzzle of how the E3-like ATG12-ATG5-ATG16L1 complex and the E2-like conjugase ATG3 deliver LC3 to phagophore membranes. Results from atomistic MD simulations point toward a multistage mechanism progressively localizing the ATG3-LC3 conjugate nearer to the target membrane and orienting the reactive center of LC3 conjugation toward lipid substrates.
This process requires the sequential action of three previously identified membrane sensors within the assembly: (i) WIPI2 as the PI(3)P effector protein that drives membrane recruitment of ATG12-ATG5-ATG16L1 22,52 , (ii) the curvature-sensitive ATG16L1 helix α2 within the (ATG12-)ATG5-binding domain 25,26 , and (iii) the N-terminal amphipathic helix and membrane docking face of ATG3 43 .
As an emerging theme in cellular processes, with analogies to the multi-step process of docking in vesicle fusion 53 , the stepwise mechanism for the membrane targeting of LC3 provides additional layers of regulatory potential to the autophagic pathway. On the protein side, phosphorylation and other post-translational modifications will affect the stability, accessibility, and affinity of the distinct interaction elements. On the membrane side, variations in lipid composition and PI phosphorylation will modulate membrane recruitment. The phagophore lipid composition in particular modulates the recruitment of WIPI2 as anchor for ATG16L1 in docking step 1 as well as the membrane insertion of the ATG16L1 α2 helix and the ATG3 N-terminal helix in steps 2 and 3, respectively. Growing evidence points to a second WIPI2-interacting site within the ATG16L1 coiled-coil domain 27  Simulations and experiments identify distinct roles for two fully conserved histidine residues in the vicinity of the catalytic cysteine of ATG3. We found neutral His266 to stabilize a catalytically competent structure of the active site, consistent with retained lipidation activity of the H266A mutant. By contrast, protonation of His266 disrupted the active site in our MD simulations, consistent with a role of His266 as pH sensor 35 . Whilst uncharged His266 serves to stabilize the catalytic loop conformation, our data point to active participation of His262 in the initiation of the LC3 lipidation reaction. In particular, we found the unprotonated His262 imidazole nitrogen to be positioned as proton acceptor from PE. Consistent with a possible catalytic role, the H262A mutation abolished function. His262 is the starting residue in the highly conserved HPC motif 45 of ATG3, which is shared with the ATG10 conjugase family. However, in ATG10 the nearby His266 of ATG3 is changed to a threonine, which may reflect the distinct substrate specificity of the two enzymes (SI Fig. 3).
The critical biological role of the ATG12-ATG5-ATG16L1 complex in mammalian autophagy 19 , and before that, the role of the corresponding Atg12-Atg5-Atg16 complex in yeast 16 , has long been appreciated. Yet the precise role of this complex in LC3 lipidation has been hard to define. The role of the extensive structural elements linking the N-terminal helix of ATG3 on the one hand, and the established WIPI2-dependent membrane docking site on the other, have proven difficult to characterize as the membrane-associated system is too large for NMR, yet too dynamic for X-ray crystallography or single-particle cryo-EM. Under the "computational microscope" of molecular dynamics simulations, the role of the connecting elements in mediating a stepwise docking process has now been unveiled. As a core element in the molecular machinery of selective autophagy, this new and far more detailed insight into the membrane docking steps of LC3 will undoubtedly facilitate the therapeutic targeting of autophagy in Parkinson's disease and other neurodegenerative diseases.

Structural models of protein complexes
Atomistic models of the ATG3-ATG12-ATG5-ATG16L1 and WIPI2d-ATG16L1 complexes were based on crystal structures with PDB IDs 4NAW 21 and 7MU2 18 , respectively. The ATG16L1 N-terminal domain in the former complex was replaced by a more complete structure (PDB ID: 4TQ0 56 ). As introduced previously 25 , an alternative conformation of the same ATG16L1 region was generated in PyMOL 57 by rotation of helix α2 relative to helix α1 at the Gln30/Ala31 hinge.
A model for the dimeric central ATG16L1 domain was completed through (i) homology modeling of residues 141-225 using SWISS-MODEL 58 based on crystal structures of the mouse protein (PDB IDs: 6ZAY 32 and 6SUR 31 ) and (ii) parameter fitting for residues 78-193 with CCBuilder 2.0 29 upon coiled-coil prediction 28 by NPS@ 59 . Unstructured inter-domain loops were added using the DEMO server 60 to yield an ATG16L1 dimer encompassing residues 1-247. AlphaFold v2.2 34 was used to model ATG3-LC3B in complex with the ATG7 homodimer. The Cys264 side chain of ATG3 was connected to the LC3B C-terminus by a thioester bond, parameterized using CHARMM-GUI 61,62 . The ATG5 Lys130 side chain was similarly connected to the ATG12 Cterminus, via an isopeptide bond. The ATG16L1 WD40 domain (dispensable for canonical autophagy 63 ) was excluded from the model, as were the unstructured ATG12 residues 1-52 and WIPI2d residues 1-11 and 362-425. Exposed N-or C-terminal groups at the end(s) of each incomplete structure or truncated construct were neutralized. Protonation states of amino acid side chains were assigned according to pKa prediction by PROPKA 64 . His183 and His255 at the putative PI(3)P binding sites of WIPI2d were protonated. Six models of the ATG3-LC3B conjugate were generated, with the imidazole of ATG3 His262 uncharged (protonated at the δor ϵ-nitrogen in alternative models) and that of His266 uncharged (protonated at δ-or ϵ-nitrogen in alternative models) or cationic (doubly protonated).

Molecular dynamics simulations
Molecular dynamics simulations were performed with GROMACS 2020 65 using the CHARMM36m force field 66 . Following the same protocol as previously described 25 , all membranes consisted of 60% DOPC, 20% DOPE, 5% DOPS, 10% POPI 38 , and 5% PI(3)P based on the ER lipid composition 12 and were prepared initially in a coarse-grained representation using the insane method 67 . Curved membranes were constructed using LipidWrapper 68 by fitting the amplitude of the membrane buckle as a sine function of its x-coordinate. Each coarse-grained membrane system was solvated with 150 mM of aqueous NaCl, equilibrated for 200 ns, and converted into an atomistic representation using the CG2AT2 69 tool. Atomistic models of protein complexes were placed above membranes after CG2AT2 conversion, followed by re-solvation and 10 ns of further equilibration. Simulation replicates were independently prepared and equilibrated.
During equilibration, harmonic positional restraints with a force constant of 1000 kJ mol -1 were applied to non-hydrogen protein atoms or backbone beads. The xy dimensions of buckled membrane systems were fixed in simulations. System temperature and pressure were maintained at 310 K and 1 bar, respectively, using the velocity-rescaling thermostat 70 and a semi-isotropic Parrinello-Rahman barostat 71 during the production phase. The integration time step was 2 fs.
Long-range electrostatic interactions were treated using the smooth particle mesh Ewald method 72,73 with a real-space cut-off of 1 nm, a Fourier spacing of 0.12 nm, and charge interpolation through fourth-order B-splines. The LINCS algorithm was used to constrain covalent bonds involving hydrogen atoms 74 . Simulation trajectories were analyzed through MDAnalysis 2.0 75,76 in Python 3.6.

Protein expression and purification
ATG3 constructs used for in-vitro lipidation assays, GUV assays were all expressed in E. coli Purification of ATG12-ATG5-ATG16L1, ATG7, and LC3 used for liposome lipidation assays, GUV assays were performed as previously described 42 . Purification of WIPI2d was performed as previously described 18 .

GUV assay
GUVs were prepared by hydrogel-assisted swelling as described previously 17  Penicillin-streptomycin in a 5% CO2 incubator at 37°C.

HaloTag-LC3B processing assay
HaoTag-based assay was performed as previously described 78 . Cells were seeded at 100K Cells/well in 12-well plate one day before. Next day, cells were incubated in complete DMEM medium with 50 nM JF549 (Promega, GA1120) for 1 hr, and then washed twice with 1x PBS buffer followed by incubation with EBSS buffer (Thermo fisher, 24010043) to induce autophagy by starvation. After 4 hr starvation, cells were harvested with Trypsin (Thermo fisher, 25300120).
For each sample, 20 µg clarified lysates were loaded onto NuPAGE 4-12% Bis-Tris Gel (Thermo Fisher, NP0322BOX). For in-gel fluorescence imaging, the gel was immediately visualized with ChemiDoc MP imaging system (Bio-Rad) after SDS PAGE. Band intensities were acquired by exciting samples at 546 nm and 647 nm. The protocol of HaloTag-LC3B processing assay to assess autophagy is available at dx.doi.org/10.17504/protocols.io.3byl4qexzvo5/v1. were incubated with GUVs (64.8% DOPC: 20% DOPE: 5% DOPS: 10% DOPI (3)P: 0.2% Atto647 DOPE) at room temperature. Images taken at 30 min are shown. Scale bars, 5 µm. (E) Quantification of relative intensities of mCherry-LC3B on GUV membranes (means ± SDs are shown; N = 30).  imidazole ring in the uncharged and the doubly protonated state, respectively, in simulations of alternative models of the ATG3-LC3 conjugate. (F) Snapshot of membrane-associated ATG3-LC3 in which the side chain imidazole of His266 (indicated with a red dashed circle) was doubly protonated with a charge of +1. At t = 1 μs of the 2 μs simulation replicate shown, destabilization of the local protein structure has brought the His266 side chain into membrane contact.