Chromatographic purification of biological macromolecules by their capture on hydrophilic surfaces with the aid of non-ionic polymers

Abstract

Viral vaccines are considered to be amongst the most successful achievements in health science. Also, the use of viral vectors for gene therapies has shown the promise to become the next medical revolution for combating a wide variety of currently untreatable diseases. The wide range of viruses and their production methods make it extremely difficult to standardize viral vaccine manufacturing. Purification processes require several steps and are typically tailored to each particular virus species and how it is produced, making process development time-consuming and potentially delaying time to market. Even for the same product, purification processes might differ almost in their totality between a small laboratory and a commercial manufacturing facility. Due to scalability, costs, efficiency, and capacity constraints, industrial purification methods are mostly limited to chromatography and filtration operations. This work presents the development of steric exclusion chromatography (SXC) as a new platform purification method for large biomolecules such as virus particles. In SXC, an unpurified sample is mixed with a non-ionic polymer — in this case polyethylene glycol (PEG) — and fed onto a device made of a porous hydrophilic stationary phase. The target product is captured without a direct chemical interaction by a thermodynamic effect caused by the presence of the PEG. Smaller impurities such as media components and proteins are unaffected by the PEG and washed away. The bigger the target product, the lesser size and/or concentration of PEG is needed for its capture. Finally, the purified product is recovered by flushing the device with a solution not containing PEG. SXC was used for the purification of 14 different cell-based virus strains and serotypes — influenza virus, yellow fever virus, adeno-associated virus (AAV), and Modified Vaccinia Ankara (MVA) — with a wide variety of sizes (20–250 nm) and from several production processes. Likewise, extracellular vesicles (EVs) of 160–230 nm from Madin-Darby canine kidney (MDCK), baby hamster kidney (BHK), and human embryonic kidney (HEK) cells were purified. A number of stationary phases were tested, including hydroxylated monoliths (1–2 μm pore size), cellulose membranes (1–1.2 μm pore size), and 3D-printed cellulose monoliths (400–500 μm pore size). Devices packed with regenerated cellulose membranes of 1.0 μm pore size were the most efficient in terms of product yield as were concentrations of 8–10% PEG-6000 for sample loading. Four different strains of influenza virus produced in suspension MDCK cells in batch systems showed product recoveries >98%. The highest measured productivity for influenza virus A/Puerto Rico/8/34 H1N1 in terms of the hemagglutinin protein antigen was around 69 000 μg per square meter per hour (4600 monovalent doses per square meter per hour). In the case of yellow fever virus, two strains used for commercial vaccine manufacture were produced separately in adherent Vero cells. Virtually full yield of infectious titer was observed and residual DNA and protein levels were below regulatory requirements. As many as 6 × 109 plaque forming units (equivalent to more than 100 000 vaccine doses) were purified from around one liter of cell culture with a productivity of more than 5 million doses per square meter per hour. For adeno-associated virus (AAV), several wild-type and recombinant variants were produced in adherent HEK cells by triple transfection and purified from both cell lysates and cell supernatants; no product losses were detected during SXC and the purified AAV (up to 2 × 1014 viral genomes per liter) successfully induced either gene expression or gene knockdown in transduced cells both in vitro and ex vivo. Exploratory results with Modified Vaccinia Ankara (MVA) virus produced in avian cells showed virtually full yield with a TCID50 titer of 3.7 × 109 virions. The virus, however, seemed susceptible to aggregation upon addition of PEG as evidenced by particle size distribution analysis. Adding sucrose or sorbitol (8% of either) to the PEG-conditioned virus seemed to lower the amount of aggregates observed compared to the PEG-conditioned sample without stabilizers. Regardless, the SXC-purified MVA virus showed a distinct monomer peak of around 220 nm without visible aggregation. It was observed that EVs from the host cells were often co-purified with the target virus particles. This was attributed to the very similar characteristics between both. Preliminary results for the capture of EVs present in cell supernatants showed particle recoveries of around 40% and concentrations close to 8 × 1010 particles per mL. Further studies should continue to evaluate SXC for the preparative purification of EVs. Clearance of protein and DNA with SXC were typically >85% and >75%, respectively, depending on the virus and the experimental setup (e.g., placing a DNA digestion step before SXC). In all cases, it was advantageous to have a nuclease treatment before SXC to achieve lower amounts of residual DNA. SXC with 3D-printed cellulose monoliths with channel diameters of 400 μm and 500 μm was inefficient in terms of product yield (around 40% for influenza A virus) compared to the 1.0 μm regenerated cellulose membranes, however, their use is interesting for future work, e.g., as an alternative to expanded-bed chromatography. The ability to load and recover the product at physiological pH and conductivity as well as the conformation stabilizing properties of PEG are relevant advantages during the purification of labile biopharmaceuticals. The high product recoveries achieved so far with SXC make it possible to allow for subsequent polishing operations for improving purity without risking unacceptably low process yields. The narrow operational range of SXC permits the purification of viruses with a high probability of success (e.g., testing 8% PEG-6000 as a starting point) and the low cost of the membranes allows single-use operation (which avoids expensive and time-consuming cleaning and sanitization steps). Scale-up of SXC is simple, as it requires only a linear increase in membrane surface, and the use of devices of up to 20 square meters would enable industrial-scale virus purification. As a capture step, SXC seems to be comparable or better than most chromatography methods available in terms of product yield, ease of use, and scalability. However, estimating capacities is challenging since there is no direct chemical bond involved. Recovery was highly dependent on certain quality attributes of the starting material, such as residual cell debris and/or aggregated product. The results shown here are the basis for further optimization and application of this technology and they indicate that membrane-based SXC has the potential for becoming a platform technology for both viral vaccine and gene therapy applications.