Model-Based Process Development for Virus-Like Particles

Computer-guided optimization of the purification of a virus-like particle

Figure: Comparison of measured (dotted lines) and simulated UV chromatograms (solid lines) with 30 CV salt gradient for complex VLP feedstock on the AEX membrane capsule. Feedstock was divided into 16 contaminant species (grey lines) and the product (red lines).

Ebola, malaria and influenza are highly pathogenic viruses and organisms that challenge global health system. Virus-like particles (VLPs) are promising candidates in the fight against these diseases. VLPs are protein assemblies without any genetic material, which mimic the structure of viruses or present viral surface proteins. They are produced by recombinant systems such as bacteria, yeast, plant or insect cells. To remove host-cell impurities after production, a suitable downstream process must be found.

There is no platform process available for purifying VLPs. The search for a fast and robust purification process is a challenge due to complex biological feedstocks that contain components of unknown size and concentration. Currently, no platform process exists for purifying VLPs. In the given case study, a purification process for a Sf9 insect cell-derived VLP developed via experiment should be optimized through a model-based approach. Not only the experimental approach, but also the modeling of mass transfer is challenged by the presence of uncertainties and the unknown feed composition.

Process modeling based on UV absorption instead of molar concentration. The purification process was modeled by means of a lumped-rate model in combination with the stoichiometric displacement model. UV peaks were then utilized to calibrate a model using an inverse peak fitting method. Using this method, it is possible to identify an arbitrary number of unknown parameters at the same time – in particular the unknown feed composition. Afterwards, salt concentrations of a step elution process were varied to evaluate the effect on purity and yield and to identify an optimal process set-up.

Chromatography modeling is a valuable design tool for the process development of VLPs. In-silico process modeling and simulation were performed to control and predict the purification process of VLPs. In a later optimization procedure, optimal process parameters were identified that enabled the separation of the VLP from the majority of impurities with minimal experimental effort.

See full paper.

Related Projects