Model-based and rational design of plant purification processes

  • Model basiertes und rationales Design von Reinigungsprozessen aus Pflanzen

Bernau, Catherine Rose; Buyel, Johannes Felix (Thesis advisor); Blank, Lars M. (Thesis advisor)

Aachen : RWTH Aachen University (2021, 2022)
Dissertation / PhD Thesis

Dissertation, RWTH Aachen University, 2021


Plants are advantageous for the production of biopharmaceuticals because upstream production is economical and scalable, post-translational modification is possible, and plants do not support the growth of human pathogens. However, downstream processing can be even more labor-intensive compared to microbial or cell culture systems due to low product concentrations and high impurities by host cell proteins. Prediction of chromatographic separation by modeling and knowledge about the elution behavior of frequent impurities such as the host cell proteins can reduce the number of experiments for process development and thus reduce costs and development time. An important part of such models is the sorption isotherm such as the steric mass action (SMA) model, which describes the multicomponent protein-salt equilibria in ion-exchange systems. Protein specific parameters of the isotherm can be correlated with the respective protein properties to predict parameters of further proteins. For the determination of protein specific SMA parameters, pure proteins are required. Therefore, a purification process for nine host cell proteins from tobacco (Nicotiana tabacum and N. benthamiana) was developed and host cell proteins were purified in preparative and pilot scale, resulting in purities of at least 89% except for the Proteasome and glutathione reductase (<68%) and yields up to 178x10-5 g g-1 (RuBisCO). Furthermore, three Halohydrindehalogenases being HheA2. HheB2 and HheC were selected for SMA parameter determination due to their high structural resemblance but low sequence identity, expressed in Escherichia coli and purified resulting in purities up to of 79% and yields up to 751x10-5 g g-1 (HheA2). In addition, seven commercial proteins were selected. Next to pure proteins, multiple factors possibly influencing the precision of models predicting the behavior of plant host cell proteins during chromatographic separations were examined. Out of the precision influencing factors, the resin batch showed a high reproducibility, whereas plant batch, cultivation and species showed a high variance and would therefore reduce the predictive power of a future prediction model. The influence of the pH on ν and Keq is not yet clear, but due to the significant higher scattering of the determined values compared to repeated measurements with the same protein at the same pH, but an effect seems to be present. In contrast, an indirect pH effect of RuBisCO by high molecular mass aggregates probably through beginning denaturation on the corresponding shielding factor was present. Moreover, the shielding factor was extraordinary high which probably originates from Size exclusions effects due to the size of RuBisCO which is not represented in the used SMA model. The analysis of all here examined proteins resulted that the influence of surface charge distribution on the ν and Keq was identified as important parameter to set-up a model for SMA parameter prediction. Overall as proof of concept, a gradient elution of a mixture of four proteins was carried out and successfully matched with predictions of single elution for each protein using the determined respective SMA parameters. Hence, the prediction of individual protein elution can be transferred to the prediction of a multicomponent elution and is a possible approach to predict separation of complex protein mixtures as purification tasks.