Back

Structural dynamics insights into principles underlying the fitness of new broadly potent AAVs

Johnson, M. E.; Ozturk, B. E.; Tugwell, T. H.; Lambros, M.; Janowitz, H. N.; Flohr, K.; Sedorovitz, M.; Campello, L.; Hartung, J.; Hogle, B.; Gillespie, M.; Schriever, H.; Aweidah, H.; Koester, J.; Clausen, I.; Seeber, S.; Revelant, F.; Schreurs, R.; Koechl, F.; Sieving, P. A.; Sahel, J.-A.; Stauffer, W. R.; Peixoto, R. T.; Fauser, S.; Lin, R.; Conway, J. F.; da Silva, S.; Krol, J.; Betegon, M.; Byrne, L. C.

2026-03-25 bioengineering
10.64898/2026.03.24.713814 bioRxiv
Show abstract

Adeno-associated virus (AAV) is a leading platform for gene therapy, but current clinical-stage vectors require high doses associated with adverse events. Engineering of AAVs has produced more efficient vectors, although the mechanism underlying these improvements often remains poorly understood, limiting further development and raising potential safety concerns. Here, we leveraged a new workflow for AAV engineering with single-cell resolution, called scAAVengr-Hunt, to create best-in-class AAVs for gene delivery. ATX002, the top-performing vector, demonstrates broad potency across species, including nonhuman primate, mouse, and human, as well as across retina and brain. To understand the mechanism underlying this broad potency, we performed molecular dynamics simulations comparing AAV variants spanning a range of fitness levels. Structural dynamics analysis revealed a bifunctional molecular mechanism that confers potency through increased affinity of the capsid to the AAV receptor and regulation of heparan sulfate binding. This work provides critical insights relating structural mechanism to the fitness of engineered AAVs and establishes rich new avenues for AAV engineering through the integration of sequence-level analysis with computational biophysics.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.