100. Evaluation of Ion-Exchange Membranes for the Purification of AAV8 from Culture Media

Abstract

Introduction: AAV is the vector of choice for over 6% of all gene therapy clinical trials, with 111 clinical trials open worldwide as of July 2015. However, production of clinical material often relies on inefficient (e. g., size exclusion chromatography, cesium chloride centrifugation) or expensive and highly serotype-specific (e.g., affinity) downstream processes. Therefore, we investigated scalable and potentially serotype-agnostic alternative primary capture and purification methods. Here, we evaluate three ion-exchange membranes with different ligands for their ability to capture and purify AAV8 from culture media, and show highly efficient capture and purification using a Sartobind salt-tolerant interaction chromatography (STIC) with a primary amine (PA) ligand membrane.Methods: AAV8 particles were produced in adherent 293 cells using either serum-free DMEM or DMEM supplemented with 1% FBS. AAV was harvested at day 6 post-transfection, at which point approximately 90% of AAV8 particles were in the culture media. Media was sterile filtered and, in equal volumes using an AKTA Explorer, run over one of three ion-exchange membranes: Sartobind Q or STIC PA anion-exchange membranes, and a Sartobind S cation-exchange membrane. Load, flow-through, wash, and elution fractions were analyzed by qPCR or AAV8 capsid ELISA to determine purification efficiency and BCA was used to quantify total protein in each fraction.Results: Recovery with Sartobind Q and S membranes was very poor, with 12% and 15% recovery, respectively, in the elution fractions and the vast majority of AAV present in the flow-through fractions. The STIC PA membrane performed considerably better, with 99.9% of particles captured by the membrane and approximately 67% of particles in the eluate when the media was adjusted to pH 9 prior to loading and the elution was performed with a gradient of 2M NaCl. However, adjusting the pH to 7 and using a gradient of 3M NaCl led to recoveries of greater than 90% as measured by AAV8 particle ELISA. In addition, the eluate showed an 80% decrease in total protein compared to the load. Results were comparable using both serum-free DMEM and DMEM supplemented with 1% FBS, which showed a dynamic binding capacity of approximately 1×1014 capsids per mL of membrane at 6% breakthrough.Conclusions: While Sartobind Q and S membranes inefficiently captured AAV8, the Sartobind STIC PA membrane captured AAV8 highly efficiently and eluted AAV in a pH and salt-dependent manner with concomitant reduction in total protein. While it remains to be determined whether this modality is suitable for serotypes other than AAV8, the membrane format would allow for processing of an equivalent batch of AAV in less than 1% of the time of a size exclusion column, without the requirement for column packing, and with more efficient recovery.