Gene therapies are transforming modern medicine, with adeno-associated virus (AAV) being the most used vector for delivery. High manufacturing costs however continue to limit drug discovery, clinical development, and patient access. Plasmid DNA (pDNA) has been identified as a critical bottleneck in AAV development and small-scale production. In this study, we compared conventional manual plasmid purification kits with a scalable, automated purification system across Maxi, Mega, and Giga scales. Time savings increased with purification scale, e.g., when purifying 4 plasmid samples at Giga scale (10 mg column plasmid capacity), 8 h of active time was saved using the automated system and total time was cut from almost 10 h to 5 h, although with generally lower yields (3–67% lower) and higher endotoxin levels. Both purification methods produced comparable AAV9 titres using standard PEI-based transfection of HEK293F cells. Surprisingly the AAV-MAX production system, plasmids purified by the automated method generated 14-fold higher titres than manually purified plasmids (3.2 ×10 ¹¹ vg/ml vs. 2.3 ×10 ¹⁰ vg/ml, respectively). Further analyses, including single and triple plasmid transfections and dynamic light scattering, revealed that the elution buffer used in the automated system formed larger pDNA transfection complexes in the AAV-MAX system and resulted in increased transfection efficiency. These findings provide insights that extend beyond AAV manufacturing, highlighting opportunities to optimise the transient production of other biologics and further improvement of plasmid preparation methods.

In vivo gene therapy is limited by inefficient delivery to target tissues and off-target accumulation in critical organs. We describe a whole body-scale retargeting platform for adeno-associated viruses (AAVs) that enables cell-selective DNA delivery through grafting of Affibody affinity proteins onto the capsid. Systematic evaluation identified the VR4 and VR8 loops of virion protein 2 (VP2) as preferred grafting sites, preserving vector integrity, potency, and manufacturability. Six distinct Affibody molecules conferred receptor-dependent infectivity across clinically relevant human receptors, demonstrating platform modularity. A HER2-targeting prodrug strategy enabled selective killing of gastric cancer cells, and dual Affibody integration generated bivalent or bispecific AAVs with enhanced transduction of refractory cell lines, including metastatic adenocarcinoma. In mice, Affibody-AAVs were redirected away from the liver, with 5-fold and 75-fold increases in brain and pancreas uptake relative to wild-type AAV2. Whole-proteome mining across 36 human organs identified organ-enriched receptors as candidates for future cell-selective gene therapies.

Adeno-associated viruses (AAVs) have transformed the treatment of genetic disease, but their use is constrained by limited organ specificity of the natural AAV serotypes, leading to dose-related toxicity. AAVs with defined cell targeting through integration of affinity scaffold proteins have therefore been previously developed. From a structural point of view, there are two questions central to such capsids: how an integrated targeting protein is spatially arranged on the AAV capsid, and whether displaying the protein compromises capsid integrity. Here, we study both aspects for an AAV2 capsid engineered to display Affibody affinity proteins (Z). By Cryo-EM studies of Z-AAV2, we determined the capsid structure to 1.67 Å resolution, the highest-resolution AAV2-derived structure reported to date. Through subunit-level 3D variability analysis, determined the local spatial positioning of the Affibody on VP2, providing the first ever structural data that place an affinity scaffold on an AAV. Thermal denaturation showed that Affibody integration does not destabilise the capsid. Integration of Affibody molecules using a flexible linker raised the capsid melting temperature by 2-4 degrees Celsius.

Adeno-associated viruses (AAVs) are the backbone of in vivo gene therapy, but patient access is hindered by high production costs. Adherent HEK293T cells are commonly used for AAV production in both development scale and large-scale, despite suspension HEK293F cells being the preferred format for industrial upscaling. As comparative studies of AAV production in adherent versus suspension HEK293 are scarce, we systematically evaluated the production of 10 different AAV serotypes in adherent HEK293T cells and suspension HEK293F cells. All serotypes produced at higher levels in adherent than suspension (4- to 19-fold increase in viral genome titres; 3- to 633-fold in capsid titres). Notably, AAV2 exhibited a 316-fold difference in secretion between adherent and suspension cells. Investigating the capsid protein sequences of 9 naturally occurring AAV serotypes, we identify 12 divergent regions, named Serotype Divergent Regions (SDRs), which could contribute to the observed differences in secretion. We created AAV hybrids by exchanging SDR-regions of a low-secreting (AAV2) and a highly secreting (AAV8) serotype, which demonstrated increased capsid secretion. Comparison of cell surface heparan sulphate proteoglycan expression levels, the class of receptor targeted by AAV2, revealed considerably lower expression levels on adherent HEK293T than suspension HEK293F.We postulate that capsid-glycan interactions can be engineered to increase AAV2 secretion and production.