Separation of empty and full adeno-associated viral capsids by MCSGP: From model-based design to implementation

Adeno-associated viruses (AAVs) are acquiring growing importance in the gene therapy scenario. However, an efficient separation of particles enveloping the desired genetic cargo, known as full AAVs, from empty or misloaded particles is yet to be achieved. In this work, we systematically investigated the empty/full separation for AAV2 by anion exchange chromatography. First, the role of the gradient slope and salt concentration in the wash and equilibration stages were elucidated in single-column operations. Then, a model-based optimization algorithm was developed to establish the most effective combinations of these parameters to maximize yield and productivity at a given purity specification (Pspec). We demonstrated that a tradeoff is in place. In fact, the yield could be maximized to 11.3 % (for Pspec=70.0 %) when increasing the gradient duration to 30 column volumes (CV), which led to a decrease in the productivity to 3.28E13 viral particles per liter of resin per hour (VP/(L_resin h)). On the other side, with a 20 CV gradient, the productivity improved to 3.85E13 VP/(L_resin h), but the yield correspondingly decreased to 10.2 % This tradeoff could be alleviated by adopting the multicolumn countercurrent solvent gradient purification (MCSGP). After having highlighted the operability region of MCSGP, defined through model simulations, an experimental run allowed us to obtain simultaneously high yield (57.8 %) and productivity up to 8.65E13 VP/(L_resin h) when targeting the same Pspec=70.0 %, greatly improving the batch performance. At the same time, the process mass intensity could be cut by a factor 3, thus demonstrating the potential of this twin-column process in the empty/full AAV separation.