Intrinsic lipid binding activity of ATG16L1 supports efficient membrane anchoring and autophagy

Abstract Membrane targeting of autophagy‐related complexes is an important step that regulates their activities and prevents their aberrant engagement on non‐autophagic membranes. ATG16L1 is a core autophagy protein implicated at distinct phases of autophagosome biogenesis. In this study, we dissected the recruitment of ATG16L1 to the pre‐autophagosomal structure (PAS) and showed that it requires sequences within its coiled‐coil domain (CCD) dispensable for homodimerisation. Structural and mutational analyses identified conserved residues within the CCD of ATG16L1 that mediate direct binding to phosphoinositides, including phosphatidylinositol 3‐phosphate (PI3P). Mutating putative lipid binding residues abrogated the localisation of ATG16L1 to the PAS and inhibited LC3 lipidation. On the other hand, enhancing lipid binding of ATG16L1 by mutating negatively charged residues adjacent to the lipid binding motif also resulted in autophagy inhibition, suggesting that regulated recruitment of ATG16L1 to the PAS is required for its autophagic activity. Overall, our findings indicate that ATG16L1 harbours an intrinsic ability to bind lipids that plays an essential role during LC3 lipidation and autophagosome maturation.

Thank you for submitting your manuscript entitled "Intrinsic lipid binding activity of ATG16L1 is essential for membrane anchoring and autophagy" (EMBOJ-2018-100554) to The EMBO Journal. Your study has been sent to three referees for evaluation, and we have now received reports from them, which are enclosed below for your information.
As you can see, while the referees consider the findings interesting, they also raise critical points that need to be addressed before they can support publication in The EMBO Journal. In particular, the referees find that the nature of the ATG16L1 puncta seen in ATG16L1LE mutant cells should be investigated further (points 1 and 4 from referee #1 and #3, respectively). Importantly, we agree with referee #2 that the direct lipid-binding ability of the ATG16L1 CCD domain should be tested by independent biochemical assays and that a role for the WD40 domain of ATG16L1 in lipid binding needs to formally be ruled out. Referee #3 points out that the RAB33B binding activity and the lipid-binding specificity of ATG16L1 need in depth investigation. In addition, and while this was not specifically mentioned by the referees, we had to notice that functional consequences of lipid binding by ATG16L1 in cell are only described up to the level of puncta formation and PAS recruitment. This issue was also mentioned by an external advisor we consulted before sending the manuscript out for peer-review. From our side, and to further strengthen the revised manuscript, we would therefore encourage you to include additional data on the functional contribution from ATG16L1 lipid binding in cells.
Given the referees' interest and overall positive recommendations, I would like to invite you to submit a revised version of the manuscript, addressing the comments of all three reviewers. I should add that it is EMBO Journal policy to allow only a single round of revision, and acceptance of your manuscript will therefore depend on the completeness of your responses in this revised version. In addition, I realize that addressing all the referees' criticisms will require a lot of additional time and effort and be technically challenging. I would therefore understand if you were to choose not to undergo an extensive revision here and rather pursue a submission at an alternative venue, in which case please inform us about your decision at your earliest convenience.
Given the rather open-ended nature of the revision experiments -and to make sure that you focus on experiments that will help you address the most crucial referee concerns -I would like to discuss the data that you could imagine including in a revised manuscript. Could you therefore please provide me with a preliminary point-by-point response to the referee concerns in which you outline your plans for the revision? Based on this, I would be happy to discuss the exact requirements for the revised manuscript with you. This manuscript entitled "Intrinsic lipid binding activity of ATG16L1 is essential for membrane anchoring and autophagy" by Leo J Dudley et. al. reported the PI3P binding potential in Atg16L1 for the first time. They discovered that conserved residues within the CCD of ATG16L1 that mediate direct binding to PI3P. This binding potential was critical in Atg16L1 recruitment to PAS and autophagy progression. Interestingly, the point mutant that strengthen the binding provided negative effect in autophagy, suggesting proper regulation of PI3P binding is crucial for autophagy. The data are clean and convincing. Suggested model would be mostly consistent with the past literatures, and will provide fundamental insight into autophagy mechanism. One point this referee ask to deepen this study is about the nature of Atg16L1 positive puncta in ATG16L1LE mutant as follows. Major point 1. What is the nature of Atg16L1 positive puncta in ATG16L1LE mutant? Is it positive for other Atg proteins including FIP200 and WIPI2b? This is important point because if they are positive, Atg16L1 could recruit those proteins and provide novel hierarchy in Atg relationship.
Referee #2: The present study focuses on a new feature of ATG16L1, an intrinsic lipid-binding ability. A direct binding between ATG16L1 and lipid membrane has not been reported, which makes this subject important for the autophagy field. There is no doubt that the triple mutation (I171D, K179D, and R193D) in the CCD of ATG16L1 is affecting the PAS and autophagy. But that can be expected when a mutation in a relatively rigid α-helix is introduced into a protein. A novel contribution would be represented by linking this in vivo change to a convincing experimental demonstration that the CCD domain of ATG16L1 has a direct lipid-binding ability. In this aspect, the present study falls short, and additional work is needed to address major issues.
1. The microscopy-based technique in Figure 3B and C represents the only in vitro experimental approach to the binding of full-length ATG16L1 to lipids. At least two independent biochemical methods should confirm that. Why didn't the authors use the liposome co-sedimentation assay in Fig. 3? 2. The authors show in Fig.3 A and D and Fig. 6A that the rigid helix in the CCD domain, in between residues L162 and P249, has I171, K179 and R193 on the interface contacting the lipid membrane. However, the same face of the helix also contains E178, E185 and E186. It is difficult to imagine these 3 negatively charged residues not interfering with binding of I171, K179 and R193 to negatively charged phospholipids. In other words, why do E178, E185 and E186 not engage in electrostatic repulsions from the negatively-charged membrane? The authors need to address this issue.
3. The lipid-binding experiment in Fig. 3B, C, E was carried out with full-length ATG16L1-GFP in the presence of small unilamellar vesicles (SUVs). The authors see the strongest binding with SUVs enriched with PI3P. In Fig. 6C, the authors shows a liposome co-sedimentation assay with a much smaller wild-type peptide, ATG16L1[72-307] lacking the WD40 domain. This small peptide does not bind to SUVs at all--there is no protein in the P fraction. How is it that the large GFP-tagged protein binds, but a smaller peptide does not bind to SUVs? The authors need to address this discrepancy. Of course, the techniques in Fig. 3 and Fig. 6 are different, but they should be consistent with each other.
4. This leads to my most significant point, which is that the authors completely overlooked a possible effect of the WD40 domain. The crystal structure of this domain was published recently (Bajagic et al., Prot Sci 2017), and shows that WD40 is a 7-bladed β-propeller. An earlier paper (Baskaran et al., Mol Cell 2012) shows that 7-bladed β-propeller structures (e.g. WIPI2, Atg18, etc.) can bind PI3P. The ATG16L1 WD40 domain and WIPI-2 have homology and the WD40 of ATG16L1 has two positively charged pockets. Therefore, it cannot be excluded that the WD40 of ATG16L1 has an intrinsic membrane binding ability. Can the missing WD40 in the wild-type ATG16L1[72-307] peptide be a reason for not seeing this peptide in the P fraction (Fig. 6C)? Could be the preferential binding of ATG16L1-GFP to PI3P-enriched SUVs actually be a result of WD40 binding to lipids (Fig. 3)?
5. ATG16L1 carries a WD40 domain and binds WIPI2. Thereby, it has at its disposal 1-2 membrane binding modules that prefer PI3P. Why would the protein need a third module, the CCD domain, for membrane binding? To convincingly prove the membrane binding ability of the CCD domain in ATG16L1, the authors need to carry out all microscopy-based and liposome-based experiments with ATG16L1 lacking the WD40 domain. If a membrane binding mechanism via the CCD mediates recruitment of the protein to the PAS, the authors need to show this using the triple mutant of ATG16L1 that has the disabled WD40 domain along with the WIPI2 and FIP200 binding sites in ATG5-/-cells.
Minor points: 1. The experiment in Fig. 2E and Fig. 3G needs to be done with disabled WD40, because it has been proposed that the WD40 might facilitate dimerization via a highly conserved interface (Bajagic et al., Prot Sci 2017). Fig. 3E does not reflect the quantification, a better image should be chosen.

The representative image in
Referee #3: ATG16L1 is an essential factor for autophagosome formation and together with the ATG12-ATG5 conjugate promotes lipidation of ATG8 family proteins such as LC3. To fulfill its role in autophagy, ATG16L1 must be recruited to preautophagosomal structures (PAS) at an appropriate time; however, its precise targeting mechanism is not fully understood. In this manuscript, Dudley and coworkers showed that a previously less-characterized region (AA185-205) of the coiled-coil domain of ATG16L1 is required for its recruitment to the PAS and that the corresponding region has the ability to bind phosphoinositides such as PI3P. Furthermore, they found that ATG16L1 mutants with decreased or increased PI3P binding activity failed to promote LC3 lipidation. Finally, the authors proposed that a proper lipid binding ability of ATG16L1 is essential for its localization at the PAS as well as LC3 lipidation during starvation-induced autophagy. Overall, the manuscript is well-written and well-organized, and their findings are potentially interesting to the general readers of EMBO J. To strengthen the authors' conclusions, however, followings points need to be addressed prior to publication.
Major points: 1. The authors claimed in the text that sequences within the CCD dispensable for known proteinprotein interactions are required for the recruitment of ATG16L1 to the PAS. However, this statement is not accurate. Previous reports have already shown that the C-terminal portion of the CCD binds to RAB33B and that this region is required for starvation-induced autophagy and Salmonella-induced autophagy (MBoC, PMID: 18448665; JCB, PMID: 24100292). Thus, the authors should investigate the RAB33B binding activity of WT and LD/LE mutants of ATG16L1 to exclude the possibility that the LD/LE mutations also affect RAB33B binding. 3. In Figure 6C, the authors showed the weak interaction between recombinant ATG16L1(AA72-307) and liposomes. However, this reviewer cannot see any significant binding activity, because no difference in the pellet fraction of ATG16L1(AA72-307) was observed in the presence and absence of liposomes. Thus, the most straightforward explanation is that purified ATG16L1(AA72-307) is unable to bind to lipids, and the ATG5-ATG12/ATG16L1 complex rather than ATG16L1 alone mediates lipid binding (see EMBO J., PMID:23064152). Although the liposome binding activity was clearly observed by microscopy-based protein-liposome binding assays ( Figure 3B and 3E), the authors cannot exclude the possibility that the ATG16L1 beads used are contaminated by a small amount of endogenous ATG5 during the purification process, because they expressed recombinant ATG16L1 in 293T cells, which endogenously express ATG5. Thus, the authors should use recombinant ATG16L1 from bacteria or ATG5-KO cells to exclude this possibility. Alternatively, the ATG16L1-delta2 mutant lacking an ATG5-binding site could be used for microscopy-based protein-liposome binding assays. Figure 7D, it is not clear why ATG16L1-LE shows punctate structure even under fed conditions. Is this an autophagy-related structure? IF analysis using phagophore and autophagosome markers should be performed. Figure 7E, the authors should detect p62 in addition to LC3 to confirm the impact of the LE mutant on autophagy. Furthermore, the expression level of the WT and LE mutant is not comparable (middle panel in Figure 7E). The results should be compared with same/similar expression levels.

In
Other minor points: 6. For multiple comparisons, t-test should not be used. Appropriate statistical analyses are necessary.

For general readers, addition of molecular size markers to each blot is highly recommended in all figures.
1st Revision -authors' response 10th December 2018 (EMBOJ-2018-100554) This is indeed an important point. Purified full length ATG16L1 is prone to sedimentation in the absence of liposomes in our system. To overcome this limitation, we used the truncation fragment (residues 72-307) in the liposome cosedimentation assay which has been previously shown to enhance the solubility of ATG16L1. We added a note to clarify this point on pg.14 of the revised manuscript. In order to confirm the lipid binding activity of wild type ATG16L1 using an independent biochemical assay, we tested the binding of full length ATG16L1 to commercially available beads conjugated to lipids. The newly added Fig EV1C confirms the ability of full length ATG16L1 to bind PI3P, relevant to its localisation to pre-autophagosome structures, as well as to other lipids (discussed below, referee #3 point #2). Fig.3 A and D and Fig. 6A that the rigid helix in the CCD domain, in between residues L162 and P249, has I171, K179 and R193 on the interface contacting the lipid membrane. However, the same face of the helix also contains E178, E185 and E186. It is difficult to imagine these 3 negatively charged residues not interfering with binding of I171, K179 and R193 to negatively charged phospholipids. In other words, why do E178, E185 and E186 not engage in electrostatic repulsions from the negatively-charged membrane? The authors need to address this issue.

The referee's suggestion that other acidic residues facing positions 'f' and 'g' on the lipid-interacting interface (including E178, E185 and E186) will affect lipid binding activity of the CCD in a similar manner as mutating residues D164 and E165
is consistent with our model. We have modified the revised manuscript to further emphasise this point on pg.8 of the results section and pg.11 of the discussion. Fig. 3B, C, E was carried out with full-length ATG16L1-GFP in the presence of small unilamellar vesicles (SUVs). The authors see the strongest binding with SUVs enriched with PI3P. In Fig. 6C, the authors shows a liposome co-sedimentation assay with a much smaller wild-type peptide, ATG16L1[72-307] lacking the WD40 domain. This small peptide does not bind to SUVs at all--there is no protein in the P fraction. How is it that the large GFPtagged protein binds, but a smaller peptide does not bind to SUVs? The authors need to address this discrepancy. Of course, the techniques in Fig. 3 and Fig. 6 are different, but they should be consistent with each other.

The lipid-binding experiment in
To address the referee's comment, we used the imaging-based technique to test liposome recruitment to recombinant ATG16L1 fragment (ATG16L172-307 WT ) immobilised on beads. The newly added Fig EV3B shows that recombinant ATG16L172-307 WT can positively recruit liposomes similar to what is observed using full length ATG16L1-GFP. This suggests that the discrepancy between the liposome binding techniques (imaging-based and co-sedimentation assays) is likely due to differences in the experimental approach and sensitivity of the readout. We further discuss this on pg.8 of the revised manuscript.
4. This leads to my most significant point, which is that the authors completely overlooked a possible effect of the WD40 domain. The crystal structure of this domain was published recently (Bajagic et al., Prot Sci 2017), and shows that WD40 is a 7-bladed β-propeller. An earlier paper (Baskaran et al., Mol Cell 2012) shows that 7-bladed β-propeller structures (e.g. WIPI2, Atg18, etc.) can bind PI3P. The ATG16L1 WD40 domain and WIPI-2 have homology and the WD40 of ATG16L1 has two positively charged pockets. Therefore, it cannot be excluded that the WD40 of ATG16L1 has an intrinsic membrane binding ability. Can the missing WD40 in the wild-type ATG16L1[72-307] peptide be a reason for not seeing this peptide in the P fraction (Fig. 6C)? Could be the preferential binding of ATG16L1-GFP to PI3P-enriched SUVs actually be a result of WD40 binding to lipids (Fig. 3)?
The referee raises an interesting point. The role of the WD40 domain in the canonical autophagy pathway has been recently published in the EMBO J (Fletcher et al., 2018) showing that it is dispensable for LC3 lipidation during starvation induced autophagy but required for LC3 lipidation on single membranes during LC3-associated phagocytosis. It is also important to note that lipid binding deficient mutant of ATG16L1 (ATG16L1 LD ) used in Fig 3E harbours an intact WD40 domain but is deficient in recruiting liposomes. Therefore, we predict that the contribution of the WD40 domain to binding lipids in this system is minimal. We directly addressed this point in the revised manuscript using the imagingbased liposome binding assay and show that a truncation mutant of ATG16L1 lacking the WD40 domain is able to bind liposomes in a similar manner to wild type protein (newly added Fig EV1A and B).
5. ATG16L1 carries a WD40 domain and binds WIPI2. Thereby, it has at its disposal 1-2 membrane binding modules that prefer PI3P. Why would the protein need a third module, the CCD domain, for membrane binding? To convincingly prove the membrane binding ability of the CCD domain in ATG16L1, the authors need to carry out all microscopy-based and liposome-based experiments with ATG16L1 lacking the WD40 domain. If a membrane binding mechanism via the CCD mediates recruitment of the protein to the PAS, the authors need to show this using the triple mutant of ATG16L1 that has the disabled WD40 domain along with the WIPI2 and FIP200 binding sites in ATG5-/-cells.
The requirement for multiple modules to target ATG16L1 to membranes is an interesting and open question. It is also interesting that sequences that mediate ATG16L1 binding to WIPI2 and FIP200 and the WD40 domain are absent in yeast Atg16 suggesting that additional conserved mechanisms are likely to exist to target ATG16L1 to the PAS (highlighted in the revised manuscript pg.3). Our results in Fig 4E show that the lipid binding deficient mutant ATG16L1 LD is diffused when expressed in cells lacking WIPI2 supporting the conclusion that the interaction of ATG16L1 to lipids and proteins are required for PAS recruitment. As above, ATG16L1 mutant lacking the WD40 domain was able to bind liposomes when tested using the imaging-based assay (newly added Fig EV1A and B). The possibility of the WD40 domain harbouring a lipid binding capacity is however an interesting postulation and is not excluded from our data. Since the WD40 domain is dispensable for canonical autophagy (Fletcher et al., 2018), its postulated lipid binding may have independent functions that are beyond the scope of our manuscript.
Minor points: 1. The experiment in Fig. 2E and Fig. 3G needs to be done with disabled WD40, because it has been proposed that the WD40 might facilitate dimerization via a highly conserved interface (Bajagic et al., Prot Sci 2017).
We tested the dimerisation activity of ATG16L1 lacking WD40 domain in the newly added Fig 2F and show that, unlike an N-terminal truncation mutant lacking CCD sequences (ATG16L1 ∆3 ), ATG16L1 ∆WD40 retained its ability to dimerise. Importantly, the contribution of the WD40 in ATG16L1 dimerisation is not excluded from our study and the conclusion that lipid binding mutant does not affect the dimerisation of ATG16L1 remains unaltered. Fig. 3E does not reflect the quantification, a better image should be chosen. Fig 3E has been replaced in the revised manuscript.

Referee #3:
ATG16L1 is an essential factor for autophagosome formation and together with the ATG12-ATG5 conjugate promotes lipidation of ATG8 family proteins such as LC3. To fulfill its role in autophagy, ATG16L1 must be recruited to preautophagosomal structures (PAS) at an appropriate time; however, its precise targeting mechanism is not fully understood. In this manuscript, Dudley and coworkers showed that a previously less-characterized region (AA185-205) of the coiled-coil domain of ATG16L1 is required for its recruitment to the PAS and that the corresponding region has the ability to bind phosphoinositides such as PI3P. Furthermore, they found that ATG16L1 mutants with decreased or increased PI3P binding activity failed to promote LC3 lipidation. Finally, the authors proposed that a proper lipid binding ability of ATG16L1 is essential for its localization at the PAS as well as LC3 lipidation during starvation-induced autophagy. Overall, the manuscript is well-written and well-organized, and their findings are potentially interesting to the general readers of EMBO J. To strengthen the authors' conclusions, however, followings points need to be addressed prior to publication.
We thank the referee for their positive recommendation.
1. The authors claimed in the text that sequences within the CCD dispensable for known protein-protein interactions are required for the recruitment of ATG16L1 to the PAS. However, this statement is not accurate. Previous reports have already shown that the C-terminal portion of the CCD binds to RAB33B and that this region is required for starvationinduced autophagy and Salmonella-induced autophagy (MBoC, PMID: 18448665; JCB, PMID: 24100292). Thus, the authors should investigate the RAB33B binding activity of WT and LD/LE mutants of ATG16L1 to exclude the possibility that the LD/LE mutations also affect RAB33B binding.
As suggested by the referee, residues within the middle region of ATG16L1 (W194 and M195) have been shown to be required for autophagy when mutated in conjunction with additional sequences (combined mutations in residues 239-242 and deletion of the WD40 domain) (Fujita et al., 2013). It is important to note that these residues do not overlap with those required for lipid binding identified in our study (I171, K179 and R193). Furthermore, the middle region of ATG16L1 was also shown to bind Rab33B. However, genetic knockdown of Rab33B did not affect LC3 lipidation while overexpression of a constitutively active mutant of Rab33B or OATL, a RabGAP for Rab33B, abrogated autophagosome formation at a later maturation step (Itoh et al., 2008; Itoh et al., 2011). Collectively, these data suggest that published activities of the CCD and lipid binding identified in our study independently mediate the role of ATG16L1 during autophagy. We have modified the manuscript text to specify that the CCD lipid binding region is dispensable for protein interactions tested in this study (e.g. with FIP200 and WIPI2, changes highlighted throughout the text). Fig EV2A). On the contrary, the lipid-enhanced mutant ATG16L1 LE retained its ability to bind Rab33B (newly added Fig EV3A). We further tested whether genetic inhibition of Rab33B implicated autophagy in a similar manner as mutating lipid binding residues in ATG16L1 or inhibiting PI3P synthesis by employing CRISPR/Cas9-mediated deletion of Rab33B. Our findings, included as newly added Fig EV2B-C, show that the absence of Rab33B did not affect LC3 lipidation or ATG16L1 puncta formation in a similar manner to mutating lipid binding residues in ATG16L1 or depleting PI3P by 3'MA treatment. Altogether, these findings suggest that binding of ATG16L1 to Rab33B and lipids result in functionally distinct consequences in cells and may require proper localisation of ATG16L1 or occur in a mutually exclusive manner. The revised manuscript was modified to include a discussion of the relevance of ATG16L1 binding to Rab33B and lipids (pg.10-11).

Note: We aimed to test whether ATG16L1 binding to lipids and Rab33B occurred in a mutually exclusive manner. Unfortunately, we could not detect an interaction between recombinant Rab33B and ATG16L1 in our system therefore we could not perform in vitro competition assays. Future work will involve further characterisation of ATG16L1 binding to its partners which falls beyond the scope of this manuscript.
2. Lipid binding specificity of ATG16L1 and its mutant should be invested in more detail. Other phospholipids such as PI(3,4)P2, PI(3,4,5)P3, PS, and PA should also be investigated in Figures 3 and 6.
In our system, we observe that ATG16L1 cannot bind liposome preparations lacking PI3P, PI4P or PI(4,5)P2 but containing PS, PE and PC. This suggests that phospholipids, such as PS, PE and PC, are not sufficient to bind ATG16L1 when tested using the liposome co-sedimentation and imaging-based assays. We chose to focus on binding to phosphoinositides as they are associated with membrane compartments where ATG16L1 has been shown to localise and emphasised on PI3P binding given its relevance in the PAS recruitment of ATG proteins.
Using an alternative assay to test lipid binding of ATG16L1 (as suggested by referee #2, point #1), we observed consistent ability of ATG16L1 to bind PI3P and PI(3,4)P2 but not to PE or PS (in agreement with our liposome-based assays, newly added Fig EV1C). This assay also revealed a potential ability of ATG16L1 to bind PA, although the data were not statistically significant. It is possible that the binding of ATG16L1 to other phospholipids, such as PA, may implicate its activities independently of PAS recruitment. Such possibilities are further discussed in the revised manuscript (pg.10). Figure 6C, the authors showed the weak interaction between recombinant ATG16L1(AA72-307) and liposomes. However, this reviewer cannot see any significant binding activity, because no difference in the pellet fraction of ATG16L1(AA72-307) was observed in the presence and absence of liposomes. Thus, the most straightforward explanation is that purified ATG16L1(AA72-307) is unable to bind to lipids, and the ATG5-ATG12/ATG16L1 complex rather than ATG16L1 alone mediates lipid binding (see EMBO J., PMID:23064152). Although the liposome binding activity was clearly observed by microscopy-based protein-liposome binding assays ( Figure 3B and 3E), the authors cannot exclude the possibility that the ATG16L1 beads used are contaminated by a small amount of endogenous ATG5 during the purification process, because they expressed recombinant ATG16L1 in 293T cells, which endogenously express ATG5. Thus, the authors should use recombinant ATG16L1 from bacteria or ATG5-KO cells to exclude this possibility. Alternatively, the ATG16L1-delta2 mutant lacking an ATG5-binding site could be used for microscopy-based protein-liposome binding assays. Fig EV1A and B). We also show that the recombinant ATG16L172-307 fragment can bind to liposomes in the imaging-based assay (newly added Fig EV3B). Together, our data suggest that the differences between liposome binding in the co-sedimentation and imaging-based assays are likely due to differences in the experimental systems and readout rather than ATG5 binding (further discussed on pg.8 of the revised manuscript). Figure 7D, it is not clear why ATG16L1-LE shows punctate structure even under fed conditions. Is this an autophagyrelated structure? IF analysis using phagophore and autophagosome markers should be performed. Figure 7E, the authors should detect p62 in addition to LC3 to confirm the impact of the LE mutant on autophagy. Furthermore, the expression level of the WT and LE mutant is not comparable (middle panel in Figure 7E). The results should be compared with same/similar expression levels.

In
We agree with the referee that mutant ATG16L1 LE appears to express at lower levels in ATG16L1-deficient background. We addressed this issue by improving our virus titre in order to obtain comparable levels between ATG16L1 WT and ATG16L1 LE -expressing cells. The updated Fig 7E shows that LC3 lipidation and p62 degradation are strongly inhibited in the newly generated ATG16L1 LE -expressing cells further supporting our conclusion that enhanced lipid binding of this mutant suppresses autophagy.
Other minor points: 6. For multiple comparisons, t-test should not be used. Appropriate statistical analyses are necessary.
We have double checked the use of t-tests in our statistical analyses to ensure that only two data sets are being compared. This has been clarified in the figure legends as well as the materials and methods section of the revised manuscript.

For general readers, addition of molecular size markers to each blot is highly recommended in all figures.
We thank the referee for this useful suggestion and have included molecular weight markers to all western blots in the revised manuscript. Thank you for submitting a revised version of your manuscript and please accept my apologies for the delay in getting back to you with our decision due to the recent seasonal holidays. Your study has now been seen by the original referees, whose comments are shown below.
As you will see, while referee #1 finds that all criticisms have been sufficiently addressed, referee #2 stresses that your structural model in Figure 3D does not reflect the findings of this work. In particular, this referee requests you to clearly state whether or not you are reporting the first direct electrostatic binding between solvent-exposed residues of the dimeric CCD of AT16L1 and PI3P or PI(3,4)P2. If not, s/he invites you to discuss in depth any existing example in the literature. Finally, referee #3 recommend you to tone down the title of the manuscript, as the possibility that unidentified binding partners other than lipids are involved in the ATG16L1 recruitment to PAS cannot be excluded at this stage. The authors addressed adequately to my concern, and now the paper is acceptable.
Referee #2: The revised manuscript provides much better elaboration on binding of the ATG16L1 CCD to lipids. It shows that the ATG16L1 CCD does not bind to phospholipids, such as PI, PS, PE, PC, and PA. The CCD in its dimeric form directly binds only to PI3P and PI(3,4)P2, and the binding residues are solvent-exposed at the positions "f" and "g". I have one major point that I think the authors need to address prior to publication. I think that their structural model in Figure 3D  In the publication by Klocek et al. (2009), melittin is a random coil peptide that folds into an α-helix upon binding to lipid bilayer, which involves four steps: i) electrostatic attraction of cationic peptide to an anionic membrane surface, ii) hydrophobic insertion into the lipid membrane, iii) a conformational change from random coil to α-helix, and iv) peptide aggregation in the lipid phase. Thus, the random coil melittin is completely unrelated to the well-folded and dimeric CCD of ATG16L1.
Pluhackova et al. (2015) is a study on the adsorption of the model SNARE (E and K) peptides to the phospholipid/cholesterol membrane, where initial binding of the positively charged N terminus of the peptide to the phospholipid headgroups is followed by insertion of the peptide at the interface between the hydrophobic core and polar headgroup region of the membrane. The main interacting force is hydrophobic and involves residues at the positions "a" and "d". Moreover, K and E peptides dissociate from a heterodimeric coiled-coil, bind as single helix, not a dimer, and use mainly hydrophobic residues for interaction with the membrane. In contrast, the residues at the positions "a" and "d" in the CCD of ATG16L1 are hidden in the dimer interface, and are not available for interaction with lipids. The ATG16L1 CCD utilizes solvent-exposed residues at the positions "f" and "g", as clearly suggested by the authors.
Woo and Lee (2016) studied insertions of E and K peptides into lipid bilayers and how the longer hydrocarbons of Lys residues, as compared to Glu residues, form stronger hydrophobic interactions with lipid tails. In contrast, the CCD of ATG16L1 does not interact with the head groups of PI, PE, PS, or PC and thus, its Lys residues cannot reach the lipid tails in the hydrophobic portion of the lipid bilayer.
Direct electrostatic binding between solvent-exposed residues of the dimeric CCD of AT16L1 and PI3P or PI(3,4)P2 proposed by this work is surprising. If this mechanism has been reported for some other dimeric coiled-coil peptide, the authors need to present a correct example in the Discussion section. If no analogous mechanism has been reported, the authors need to clearly acknowledge this, and not mislead or  Figure 3D is highly recommended.
The finding on the interaction between the dimeric CCD of ATG16L1 and phosphoinositide phosphates will be a major part of the publication. Therefore, a correct structural model is important for good understanding by the general readers of EMBO J.
Referee #3: In the revised manuscript, the authors performed additional experiments and addressed most of the concerns raised by this reviewer. I have only one minor suggestion. The title "Intrinsic lipid binding activity of ATG16L1 is essential..." is too strong, because the authors cannot exclude the possibility that unidentified binding partners other than lipids are involved in the ATG16L1 recruitment to PAS as described in the Discussion. Thus, I strongly recommend the authors to weaken the expression of the title. Referee #1: The authors addressed adequately to my concern, and now the paper is acceptable.
Referee #2: The revised manuscript provides much better elaboration on binding of the ATG16L1 CCD to lipids. It shows that the ATG16L1 CCD does not bind to phospholipids, such as PI, PS, PE, PC, and PA. The CCD in its dimeric form directly binds only to PI3P and PI(3,4)P2, and the binding residues are solvent-exposed at the positions "f" and "g". I have one major point that I think the authors need to address prior to publication. I think that their structural model in Figure 3D  In the publication by Klocek et al. (2009), melittin is a random coil peptide that folds into an α-helix upon binding to lipid bilayer, which involves four steps: i) electrostatic attraction of cationic peptide to an anionic membrane surface, ii) hydrophobic insertion into the lipid membrane, iii) a conformational change from random coil to α-helix, and iv) peptide aggregation in the lipid phase. Thus, the random coil melittin is completely unrelated to the well-folded and dimeric CCD of ATG16L1.
Pluhackova et al. (2015) is a study on the adsorption of the model SNARE (E and K) peptides to the phospholipid/cholesterol membrane, where initial binding of the positively charged N terminus of the peptide to the phospholipid headgroups is followed by insertion of the peptide at the interface between the hydrophobic core and polar headgroup region of the membrane. The main interacting force is hydrophobic and involves residues at the positions "a" and "d". Moreover, K and E peptides dissociate from a heterodimeric coiled-coil, bind as single helix, not a dimer, and use mainly hydrophobic residues for interaction with the membrane. In contrast, the residues at the positions "a" and "d" in the CCD of ATG16L1 are hidden in the dimer interface, and are not available for interaction with lipids. The ATG16L1 CCD utilizes solvent-exposed residues at the positions "f" and "g", as clearly suggested by the authors.
Woo and Lee (2016) studied insertions of E and K peptides into lipid bilayers and how the longer hydrocarbons of Lys residues, as compared to Glu residues, form stronger hydrophobic interactions with lipid tails. In contrast, the CCD of ATG16L1 does not interact with the head groups of PI, PE, PS, or PC and thus, its Lys residues cannot reach the lipid tails in the hydrophobic portion of the lipid bilayer.
Direct electrostatic binding between solvent-exposed residues of the dimeric CCD of AT16L1 and PI3P or PI(3,4)P2 proposed by this work is surprising. If this mechanism has been reported for some other dimeric coiled-coil peptide, the authors need to present a correct example in the Discussion section. If no analogous mechanism has been reported, the authors need to clearly acknowledge this, and not mislead or  Figure 3D is highly recommended.
The finding on the interaction between the dimeric CCD of ATG16L1 and phosphoinositide phosphates will be a major part of the publication. Therefore, a correct structural model is important for good understanding by the general readers of EMBO J.
We have modified the manuscript to clarify our conclusions from the structural prediction of ATG16L1-lipid interaction. We rephrased the text to clarify that residues within the CCD are likely to mediate the interaction with phosphorylated phosphoinositide head groups that are embedded within the lipid bilayer rather than mediate the binding to the lipid bilayer directly. Referee #3:

We have cited published work that show interactions between CCD's and phospholipids but
In the revised manuscript, the authors performed additional experiments and addressed most of the concerns raised by this reviewer. I have only one minor suggestion. The title "Intrinsic lipid binding activity of ATG16L1 is essential..." is too strong, because the authors cannot exclude the possibility that unidentified binding partners other than lipids are involved in the ATG16L1 recruitment to PAS as described in the Discussion. Thus, I strongly recommend the authors to weaken the expression of the title.
We have modified the title to the following: "Intrinsic lipid binding activity of ATG16L1 supports efficient membrane anchoring and autophagy".
For animal studies, include a statement about randomization even if no randomization was used.
4.a. Were any steps taken to minimize the effects of subjective bias during group allocation or/and when assessing results (e.g. blinding of the investigator)? If yes please describe. Do the data meet the assumptions of the tests (e.g., normal distribution)? Describe any methods used to assess it.
Is there an estimate of variation within each group of data?
Is the variance similar between the groups that are being statistically compared?

YOU MUST COMPLETE ALL CELLS WITH A PINK BACKGROUND ê
Sample size was chosen empirically.

N/A
No measurements were excluded from this study.

N/A N/A
No blinding was done.

N/A
Student's T-test was used to compared no more than two data sets.

Yes
Yes Yes Each figure caption should contain the following information, for each panel where they are relevant:

Captions
The data shown in figures should satisfy the following conditions: Source Data should be included to report the data underlying graphs. Please follow the guidelines set out in the author ship guidelines on Data Presentation.
a description of the sample collection allowing the reader to understand whether the samples represent technical or biological replicates (including how many animals, litters, cultures, etc.).
Please fill out these boxes ê (Do not worry if you cannot see all your text once you press return) a specification of the experimental system investigated (eg cell line, species name). the assay(s) and method(s) used to carry out the reported observations and measurements an explicit mention of the biological and chemical entity(ies) that are being measured. an explicit mention of the biological and chemical entity(ies) that are altered/varied/perturbed in a controlled manner.
the exact sample size (n) for each experimental group/condition, given as a number, not a range; 1. Data the data were obtained and processed according to the field's best practice and are presented to reflect the results of the experiments in an accurate and unbiased manner. figure panels include only data points, measurements or observations that can be compared to each other in a scientifically meaningful way. graphs include clearly labeled error bars for independent experiments and sample sizes. Unless justified, error bars should not be shown for technical replicates. if n< 5, the individual data points from each experiment should be plotted and any statistical test employed should be justified

B-Statistics and general methods
a statement of how many times the experiment shown was independently replicated in the laboratory.
Any descriptions too long for the figure legend should be included in the methods section and/or with the source data.
In the pink boxes below, please ensure that the answers to the following questions are reported in the manuscript itself. Every question should be answered. If the question is not relevant to your research, please write NA (non applicable). We encourage you to include a specific subsection in the methods section for statistics, reagents, animal models and human subjects. This checklist is used to ensure good reporting standards and to improve the reproducibility of published results. These guidelines are consistent with the Principles and Guidelines for Reporting Preclinical Research issued by the NIH in 2014. Please follow the journal's authorship guidelines in preparing your manuscript.