Structure of the endocytic adaptor complex reveals the basis for efficient membrane anchoring during clathrin-mediated endocytosis

During clathrin-mediated endocytosis, a complex and dynamic network of protein-membrane interactions cooperate to achieve membrane invagination. Throughout this process in yeast, endocytic coat adaptors, Sla2 and Ent1, must remain attached to the plasma membrane to transmit force from the actin cytoskeleton required for successful membrane invagination. Here, we present a cryo-EM structure of a 16-mer complex of the ANTH and ENTH membrane-binding domains from Sla2 and Ent1 bound to PIP2 that constitutes the anchor to the plasma membrane. Detailed in vitro and in vivo mutagenesis of the complex interfaces delineate the key interactions for complex formation and deficient cell growth phenotypes demonstrate its biological relevance. A hetero-tetrameric unit binds PIP2 molecules at the ANTH-ENTH interfaces and can form larger assemblies to contribute to membrane remodeling. Finally, a time-resolved small-angle X-ray scattering study of the interaction of these adaptor domains in vitro suggests that ANTH and ENTH domains have evolved to achieve a fast subsecond timescale assembly in the presence of PIP2 and do not require further proteins to form a stable complex. Together, these findings provide a molecular understanding of an essential piece in the molecular puzzle of clathrin-coated endocytic sites.

Please mention the molecular weight of the proteins in the manuscript. Please add that to the text. In figure 4(g), spectrum 1 is labeled as 'ANTH WT.' Should it be 'AENTH WT'? Also, there are some additional peaks which may be different numbers of PIP2 bound to the complex. Please label them clearly and mention in the text if they are nonspecific interactions. Figure 4(g) -Please show the entire m/z range so that both monomeric and the multimeric species can be visible on the spectrum. Figure 4(g) spectrum 5 -are they 6:6+PIP2 oligomers? If so, please state clearly in the text. Moreover, specify the population of PIP2 adducts. Figure 4g -Please change the labeling on each spectrum to show that it is an AENTH complex. (Follow the spectrum 2 style of labeling) I recommend the paper to be accepted once the above concerns are addressed.
Reviewer #2 (Remarks to the Author): You report a convincing cryoEM structure of a 16-mer complex of the membrane binding domains ENTH and ANTH bound to PIP2, which you say constitutes the anchor to the plasma membrane, as an important step in endocytosis. I understand that the structure and kinetic data on its assembly /disassembly is novel and sound, as far as I can judge (lacking the expertise in cryoEM).
As such, I am not sure, however, that the 16-mer complex by itselfs explains sufficiently how the achoring works. You mention force transmission, and that the complex 'facilitates the formation of larger assemblies which would contribute to membrane bending and remodeling'. But how does this work ? How does the molecular structure reported help in this ? What interactions with the membrane are relevant ? Is the bound PIP2 still anchored in the membrane ? From Fig.1 , it is unclear where and how the 1

REVIEWER COMMENTS
Reviewer #1 (Remarks to the Author): The manuscript titled "Structure of the endocytic adaptor complex reveals the basis for efficient membrane anchoring during clathrin-mediated endocytosis" is well executed and comprehensive.
The authors report the oligomeric structure and interfaces of the adaptor complex AENTH of Sla2 and Ent1 in association with PIP2 using native MS and Cryo-EM, which compliment each other. It contributes to the understanding of protein-protein and protein-lipid interactions involved in the clathrin-mediated endocytosis.

Comments:
1) Please mention the molecular weight of the proteins in the manuscript. Please add that to the text.
Response: We have added a table (Table 1) indicating the experimental and theoretical masses for ANTH, ENTH, PIP 2 lipids bound and complexes.
The following text was added on page 4, line 100: "The ANTH and ENTH domains from Sla2 and Ent1 (33.2 and 18.9 kDa, respectively, see Table 1)…" 2) In figure 4(g), spectrum 1 is labeled as 'ANTH WT.' Should it be 'AENTH WT'? Also, there are some additional peaks which may be different numbers of PIP2 bound to the complex. Please label them clearly and mention in the text if they are nonspecific interactions.
Response: The labels on Figure 4(g) have been corrected and the number of PIP 2 molecules described in Table 1. Furthermore, Supplementary Fig 7 shows in detail that the different peaks correspond to different PIP 2 molecules bound to the A 8 E 8 complex.
3) Figure 4(g) -Please show the entire m/z range so that both monomeric and the multimeric species can be visible on the spectrum. Response: We have added a supplementary figure (Supplementary Fig.7) showing the whole m/z range spectrum collected for the ANTH WT + ENTH WT sample. The presence of free ANTH and ENTH is observed at lower m/z values while the complex appears at around 10000 m/z 2 ( Supplementary Fig. 7a). As suggested by the reviewer, those additional peaks corresponding to different numbers of PIP 2 (22, 23 and 24 are clearly distinguishable) bound to the complex (highest intensity +41 charge state of the 8:8 ANTH/ENTH complex) are now shown in Supplementary Fig. 7c. A close up of the spectrum at lower m/z values where ANTH and ENTH are bound to 1, 2 and 3 PIP 2 molecules is shown in Supplementary Fig. 7d. For the ANTH domain, species bound to 1, 2 and 3 Na + molecules (coming from the PIP 2 sample) were also detected (see Supplementary Fig. 7e and Materials and Methods).
The main text of the manuscript has been modified and the following added: -Page 7, line 171: "Native MS allows the identification of charge state distributions corresponding to A 8 E 8 complexes at higher m/z, with 22-24 PIP 2 molecules bound ( Supplementary Fig. 7)." -Page 10, line 250: "In addition to the 20 PIP 2 lipids described in the entire 16-mer complex, up to 24 PIP 2 molecules were detected by native MS indicating the presence of four additional bound lipid molecules that could not be resolved in our structure (Supplementary Fig. 7c).". Figure 4(g) spectrum 5 -are they 6:6+PIP2 oligomers? If so, please state clearly in the text.
Response: As the reviewer points out, the ANTH R3A/I4A/D37R/H38A mutant spectrum corresponds to a 12-mer (A 6 E 6 ) complex. The ANTH R3A/I4A/D37R/H38A masses and charge states, as well as the number of PIP 2 bound are content in Table 1. We have relabeled the spectrum in Fig. 4g.
The following text has been added to the main text of the manuscript: -Page 9, line 224: "Interestingly, native MS for the R3A/I4A/D37R/H38A mutant showed a shift in the signal of the complexes obtained to lower m/z, corresponding to 12-mers ( Fig. 4g and Table 1), in agreement with the DLS data. Assemblies of 12-mers have been previously reported as lower abundance species 19,36 and are formed by 6 ANTH and 6 ENTH molecules (termed A 6 E 6 ). However, mutation of the ANTH-ANTH interface did not cause growth defect phenotype in vivo (Supplementary Fig. 9c)." 3 -Page 12, line 307: "In case of ANTH R3A/I4A/D37R/H38A, the typical 16-mer AENTH complex was not observed for this mutant, but a 12mer assembly ( Fig. 4g and Table 1) previously described corresponding to 6 ANTH and 6 ENTH molecules (termed A 6 E 6 ) 19,36 , was sufficient to introduce membrane reshaping in vitro." Figure 4g -Please change the labeling on each spectrum to show that it is an AENTH complex. Follow the spectrum 2 style of labeling.

5)
Response: Figure 4g has been relabeled. 6) I recommend the paper to be accepted once the above concerns are addressed.

Reviewer #2 (Remarks to the Author):
You report a convincing cryoEM structure of a 16-mer complex of the membrane binding domains ENTH and ANTH bound to PIP2, which you say constitutes the anchor to the plasma membrane, as an important step in endocytosis. I understand that the structure and kinetic data on its assembly /disassembly is novel and sound, as far as I can judge (lacking the expertise in cryoEM).
1) "As such, I am not sure, however, that the 16-mer complex by itselfs explains sufficiently how the achoring works".
Response: We would like to point out that we are proposing that the anchoring structure is a tetramer that can further associate into bigger assemblies. We have presented the first highresolution structure of the AENTH complex bound to lipids. The structure was solved for a 16mer assembly composed of four AENTH tetramers. We have shown how mutations on the tetrameric interfaces disrupt the complex formation (no tetramers detected, therefore no bigger assemblies) in vitro and have associated growth defect phenotypes in vivo. We have now performed additional experiments on GUVs and LUVs to show the crucial role of the reported tetrameric AENTH structure in membrane remodeling. We now thoroughly revised the relevant parts of the manuscript: 2) "You mention force transmission, and that the complex 'facilitates the formation of larger assemblies which would contribute to membrane bending and remodeling. " Response: We have modified the text to better stress the role of the AENTH tetramer in membrane anchoring (required for force transmission) and improved the section for the formation of larger assemblies. Regarding its role in membrane remodeling we have performed two new experiments.

New experiments:
Following the reviewer's comment, new experiments performed on GUVs and LUVs ( Fig. 6 and 7) show how ANTH and ENTH mutants that are not able to form the AENTH tetramer fail to contribute to membrane remodeling using in-vitro model membranes.
2) "But how does this work ? How does the molecular structure reported help in this?" Response: We have repurposed a main figure (Fig. 5) to depict the involved protein-lipid interactions and the role of the tetramer in membrane anchoring.
3) What interactions with the membrane are relevant ?  (Fig. 6 and 7) have been added to the manuscript. When ANTH and ENTH domains are combined in the presence of GUVs containing PIP 2 the AENTH complex reshapes GUVs into tubular structures, as previously reported by Skruzny et al., (2015). We produced ANTH-GFP or ENTH-GFP, mixed them with GUVs with and without PIP 2 and imaged them using confocal fluorescence microscopy. ANTH-GFP or ENTH-GFP by themselves did not cause remodeling effect of GUVs, while when both proteins were together hairy structures protruding from the vesicles could be visualized. Importantly, ANTH and ENTH tetrameric mutants had different membrane deformation capabilities. Those mutants shown to be unable to form the AENTH complex have failed to form the hairy structures observed for the wild-type domains.
Similar results were obtained for LUVs containing PIP 2 observed by negative staining EM.
ANTH does not seem to deform LUVs notably, in agreement with literature (Ford et al. 2001).
ENTH causes LUVs to become aggregated. Both domains together then lead to clear LUVs tubulation. As for GUVs, those mutants that cannot assemble a functional AENTH complex were unable to tubulate LUVs. The conclusion from these experiments using membrane models is that only ANTH and ENTH proteins which can assemble into tetrameric (and therefore bigger assemblies) are able to remodel membranes. Response: We acknowledge that SAXS data was presented with a limited explanation and analysis in the previous version of our manuscript. A new version of Supplementary Fig. 1 including SAXS curves at different concentrations has been created addressing this point raised by the reviewer. (see Supplementary Fig. 1a). So indeed, the curves fitted to the experimental SAXS data are derived from mixtures in equilibrium, not from the theoretical scattering curve of the 16-mer (see below). 7.2. "Does this correspond exactly to the A8E8 structure reconstructed from cryo-EM, or is this some emperical fit? " Response: No, it does not correspond to the 16-mer A 8 E 8 structure scattering curve. It is a fit of a mixture of oligomers based on the combination of the experimental data available: monomers, 16-mer and 32-mer complexes existing in solution as have been previously described by Heidemann et al. 2020. 7.3) "Does the Cryo-EM structure also fit the distance distribution function of SAXS?" Response: No, the 16-mer (A 8 E 8 ) complex alone does not fit the data because the system consists of a mixture in solution (see previous response). The 16-mer overall dimensions agree with the distance distribution function, however the shape of the ab-initio models generated from this model does differ from the one of the cryo-EM structure. The reason for this is the heterogeneity present in solution, which is "filtered out" during particle selection and alignment during cryo-EM data processing to achieve higher resolution of a subset of the particles embedded in the ice. 8) "Overall I find the description of the SAXS experiments insufficient. How did structural models enter in the SVD and NMF approach in detail. Please also include the relevant references here".
Response: As suggested by the reviewer, we have now extended the analysis of the time resolved SAXS data (see Fig. 8 and Supplementary Fig. 11) and re-written the Kinetics of the AENTH complex assembly and Material and Methods sections.

Reviewer #3 (Remarks to the Author):
The authors present a structure of the membrane binding domains of two yeast proteins involved in clathrin-mediated endocytosis, Sla2 and Ent1, obtained using single particle cryoEM. At 3.9 Ang resolution, this map represents a substantial advance in detail over a previously published map of the Sla2 ANTH and Ent1 ENTH domains aligned in a helical array on lipid tubules.
Intriguingly the authors find that these domains form a 16-mer composed of heterotetramers derived from both proteins that provides a new interpretation of the previous cryoEM map and reveals PIP2 binding sites in more detail. The authors also present functional studies that probe the interfaces between subunits. They show that the subunits can assemble rapidly that assembly and disassembly depend on concentration. This is a comprehensive study that uses a variety of experimental methods to characterise this complex and probe the functional significance of the new structural information. The results have implications for our understanding of the role of these assemblies in clathrin coat formation and provide a new interpretation likely to extend current thinking.
I would strongly support publication of this work subject to consideration of the concerns and suggestions listed below.

Structural model
1) The wwPDB validation report gives a relatively low score (below 50%) for Ramachandran and side chain outliers in the structure. Could the authors comment on whether these scores can be improved?
Response: We have conducted extensive model building and refinement in COOT and ISOLDE and have attempted to improve Ramachandran and side chain outliers, but were only able to do so at the expense of an increased clash-score. Although the scores are below 50% the majority of the model can reliably be built with the problematic regions being those with relatively lower resolution at the periphery of the complex, which proved challenging to build.
2) It took a while to discover that the numbering in the PDB file was 5 residues greater than the numbering used in the text. Could this be reconciled or explained in the text and PDB entry?
Response: We apologise for the confusion caused and have now corrected the numbering in the PDB file to match the labels/numbering in the main text.
3) Also, it would be very helpful if the chains could be labelled such that the ENTH and ANTH domains can be distinguished easily when examining the structure.
Response: We have added appropriate labels for all chains. ENTH chains are now labelled as "ENTH domain of epsin Ent1" and ANTH chains are labelled as "ANTH domain of Sla2".
Manuscript 1) "Page 9 lines 216-220 The authors appear to state that they have found the true physiologically relevant form of the ANTH/ENTH interaction. While this seems to be a reasonable new interpretation of the earlier helical map, the wording here seems too strong for the evidence provided. Please rephrase or provide additional justification". Currently the only interpretation given is that there was a rapid increase in complex formation, which could arguably have been discovered using a much less sophisticated technique. If structural modelling was attempted but was unsuccessful or not appropriate in this application it would be helpful if the authors could comment on this".

Response:
We have now extended the analysis of the time resolved SAXS data (see Fig. 8 and Supplementary Fig. 11) and re-written the Kinetics of the AENTH complex assembly and tetramers forming in solution that later on assemble into larger oligomeric states (see Fig. 8).
However, we are aware that this highly speculative statement will need to be supported in future time-resolved techniques about AENTH assemblies.

5) "
Page 23 lines 544-545 -What was the reason for incubating the sample for an hour? The SAXS experiments indicated that complex formation was much more rapid than this. Could the structure have altered during that time to form a stable but less functionally relevant assembly?" Response: We agree with the reviewer that the equilibrium state (one hour incubation) differs from the fast assembly reported after 1200 ms. This has been now reflected in the manuscript