Divergent engagements between adeno-associated viruses with their cellular receptor AAVR

Adeno-associated virus (AAV) receptor (AAVR) is an essential receptor for the entry of multiple AAV serotypes with divergent rules; however, the mechanism remains unclear. Here, we determine the structures of the AAV1-AAVR and AAV5-AAVR complexes, revealing the molecular details by which PKD1 recognizes AAV5 and PKD2 is solely engaged with AAV1. PKD2 lies on the plateau region of the AAV1 capsid. However, the AAV5-AAVR interface is strikingly different, in which PKD1 is bound at the opposite side of the spike of the AAV5 capsid than the PKD2-interacting region of AAV1. Residues in strands F/G and the CD loop of PKD1 interact directly with AAV5, whereas residues in strands B/C/E and the BC loop of PKD2 make contact with AAV1. These findings further the understanding of the distinct mechanisms by which AAVR recognizes various AAV serotypes and provide an example of a single receptor engaging multiple viral serotypes with divergent rules.

Reviewer #2: Remarks to the Author: The manuscript by Lou and colleagues is focused on mapping divergent interactions of AAVR with different AAV serotypes, specifically AAV1/2/5 through cryoEM. The study clearly demonstrates that AAV5 interacts with AAVR through PKD1 and AAV1 does so by interacting with PKD2. The striking differences underlying the mechanics of how AAVR is recognized by two distinct AAVs using different surface domains/footprints for docking are a particularly novel finding in the current study. The study is carefully executed, technically sound, comprehensive in utilizing biophysical analysis as well as biological experiments (infectivity/transduction) and the manuscript is well written. Overall, the findings are novel, provide new insight into AAV biology -particularly, the study breaks new ground on how AAVR interactions might dictate AAV cell entry and transduction. Some concerns are outlined below.
First, are the mutations affecting AAV infection in a serotype specific manner or a global fashion? i.e., are AAVR PKD1/2 mutants able to rescue the non-cognate serotype? Specifically, do PKD1 mutants rescue AAV1 infection or PKD2 mutants rescue AAV5 infection?
Second, what is the impact of AAV capsid mutations on binding to wt AAVR or their respective PKD domains? Only transduction data is shown (the term "Functional titer" in Figures 2H/3H needs to be better defined. Third, the results should be discussed against the backdrop of the glycan binding footprint on AAV1 and AAV5, both of which have been shown to bind N-linked sialic acid. Further, it is unclear how interaction of different AAVs with the same AAVR (albeit through different sites) can account for the divergent tissue tropisms observed. This should be highlighted as well.
Reviewer #3: Remarks to the Author: This study is investigating the structural basis of the interaction between AAVR and different AAV strains. The authors determined four high resolution structures and showed that AAV1 binds to the second domain of AAVR (PKD2, similar to AAV2) whereas AAV5 binds to PKD1. Sequence alignments and structural comparisons were used to interpret the results. The topic is of general interests and the findings are very inspiring. The manuscript is clearly written, and the figures are professionally prepared. Two important tasks behind the story are: 1. To show that indeed PKD1 is observed in the map of AAV5-AAVR; 2. To experimentally determine the roles of different amino acids that cause these distinct binding patterns. The authors have done good jobs addressing these issues.
Here are a few specific comments: 1. Since the cryo-EM structures are the foundations of this manuscript, would it be possible to release the maps, either through the EMDB deposition system or private links?

Response to Reviewer #1
The manuscript by Zhang et al., identifies the interactions of Adeno associated virus serotypes 1 and 5 with cellular receptor, AAVR, using cryo-EM and 3D reconstruction. The structures were resolved to 3.30 Å for AAV1-AAVR and 3.18 Å for AAV5-AAVR. AAVR is a glycoprotein, which contains a cellular membrane spanning domain and 5 polycystic kidney disease domains (PKD1-5). AAV1 specifically interacts with PKD2, whilst AAV5 interacts with PKD1. This group previously published the structure of AAV2 complexed with AAVR, also using cryo-EM and showed that AAV2 binds to PKD2.
Mutagenesis followed by transduction assays confirmed the AAV and AAVR binding interface, and showed the importance of some of these residues in viral transduction. Using surface plasmon resonance, the binding affinities of variant AAVRs were assessed against AAV1 and AAV5, with variants resulting in decreased or higher KDs. Mutated AAVR was also ectopically expressed in AAVR knockdown HEK293T cells, to observe the effect on viral transduction efficiency, with variant KD correlating with viral transduction.
Capsid variants had a spectrum of phenotypes from no effect to an increase in transduction. This study represents the second report of the ubiquitously acting AAVR complexed with AAV capsids. However, while there is no major concerns with the context of this manuscript, it does not rise to the level for publication in Nature Communications following the publication of the AAV2-AAVR complex structure in Nature Microbiology (Zhang et al, Nature Microbiology 4, pages675-682 (2019)). This current manuscript reads like a continuation of this previous publication.
Response: Thank you for your comment to improve our manuscript. This work in not only a complementary to our previously work in Nature Microbiology, but also provide an example of a single receptor engaging multiple viral serotypes with divergent rules. These information will not only be interested to AAV research field, but also advance the understanding of virusreceptor recognition in the virology field.

Main text
Please run a spell check to correct typos (e.g. AA1) Response: Sorry for the typo errors. We carefully checked the manuscript and correct the mistakes.

Results
Line 120 -PKD2 also spans the 2/5 fold wall of the capsid Response: We correct this point as suggestion.
Line 135, 198 -these are non-covalent interactions, or weak hydrophobic interaction Response: We correct these two points as suggestion.
Line 141-143 -The AAV5 structure citation does not define secondary structure in this manner, cite the supplemental definition.

Response:
We correct this point as suggestion.
Line 162-163 -Why is AAV2 used as control? Could be typo? Need AAV5. If this is AAV5, comparison with AAVR-AAV2 is not conclusive since AAV2 binds PKD2.
Response: Sorry for this typo error. It is AAV5. We remove the comparison with AAV2 in this revision.
Line 233 -Intermolecular bonds needs to be change to non-covalent interactions Response: We change this point as suggestion.

Response:
We correct K446 to K464 in Figure 3g.
Response: Sorry for the mistake. We correct it to N383A.

Line 265 -Mutant and AAV1 needs to be swapped in sentence
Response: We correct this sentence.
Line 269 -not sure they have evidence for conformational alteration Response: We remove this speculation and clarify that "The mechanism for this discrepancy needs further investigation".
Comparison of the receptor interface on the AAV capsids-should keep nomenclature consistent, text is repeated from previous AAV2 paper mentioned above. This section is not really clear, and is most apparent in line 289-311 Response: As suggested by reviewers, we provide the comparison with SIA binding site with re-writing.
Line 282-283 -appears in previous AAV2-AAVR paper with a one word changed.

Please rephrase
Response: We rephrase this sentence. Response: As suggested by reviewers, we provide the information for the comparison with SIA binding sites in AAV1 and AAV5 in this revision. The SIA binding pocket on the AAV1 capsid has overlap with PKD2 binding region.
The major SIA binding pocket (site A) on the AAV5 capsid, which play essential role for AAV5 transduction, is distant from PKD1 binding region.
Line 326 -P414 instead of P432 for supplementary figure 9 Response: This error is corrected.
Response: AAV2 is the correct word here. We compare the different impacts of glycosylation of N395 on AAV2 and AAV5 transduction here.
Line 397 -need a product number for PEIMAX

Response: PEI MAX (Polyscienxes, 24765).
Lines 465-472 -The references, 27, 28 and 29 should be updated to the programs actually used Response: We update the correct references.
Lines 475 -Why use AAV2 crystal structure instead of AAV1 -see RCSB pdb No. 3NG9.

Response:
We change the initial model to the structure of AAV1 (PDB code: 5EGC) and cite the correct citation here.
Line 548 -Why overexpression (just expression) Response: Line548 "AAVR mutant overexpression and mCherry reporter assays", is rephrased to "Ectopic expression of AAVR mutant and mCherry reporter assays" Line 554 -1000 ng instead of 100ng Response: line 554-555 "A total of 100 ng or 1 μg plasmid was used for each well of the 96-well plate or 12-well plate" is rephrased to "A total of 100 ng plasmid was used for each well of the 96-well plate, or a total of 1000ng plasmid is used for each well of the 12-well plate."  AAV5 is the same. We provide these data in Supplementary Figure 9.

Supplemental
We also performed virus overlay assays to verify the impact of AAV capsid mutations on binding to wt AAVR. These data are shown in Supplementary   Figure 8. The changes of the binding of AAVR and AAV capsid mutations are similar with the changes of virus transduction.
In our case, functional titer refers to the number of viral particles that can infect the cell in every μl. We use functional titer to evaluate the infectivity of mutant capsids. We define this in the figure legends.
Third, the results should be discussed against the backdrop of the glycan binding footprint on AAV1 and AAV5, both of which have been shown to bind N-linked sialic acid. Further, it is unclear how interaction of different AAVs with the same AAVR (albeit through different sites) can account for the divergent tissue tropisms observed. This should be highlighted as well.

Response:
As suggested by reviewers, we provide the information for the comparison with SIA binding sites in AAV1 and AAV5 in this revision. The SIA binding pocket on the AAV1 capsid has overlap with PKD2 binding region.
The SIA binding pocket (site A) on the AAV5 capsid, which play essential role for AAV5 transduction, is distant from PKD1 binding region. We also provide more descriptions in Discussion section.