This PhD thesis contributes to the field of biopharmaceutical process development and production. In the pharmaceutical industry biopharmaceuticals, mostly protein-based drugs, are gaining more and more importance compared to classical, small molecule drugs. They often bind more specifically to the target or receptor. At the same time they trigger a complex mode of action. This often leads to higher efficacy, while at the same time side effects are reduced. The complex mode of action also opens up the possibility of treating new diseases or those for which there is currently no therapy available. In the past, the biggest growth in the biopharmaceutical market has taken place in the field of cancer immunotherapy. Against the background of an aging society, the challenge of fighting cancer will continue to grow. In addition, the demand for therapies for the treatment of autoimmune diseases, diabetes, and Alzheimer's disease will continue to rise as well. All biopharmaceutical products have high development and production costs in common. The increasing demand for new products and higher production capacities stands in contrast to the ever scarcer healthcare budgets. ... mehrWith the approval of the first biosimilars, generic versions of the originator drug, the economic pressure on the biopharmaceutical sector increases further. While for the first generation of biopharmaceutical blockbusters was hardly a link between production costs and sales prices, the development of biosimilars also increases the need for faster process development and more economical processes.
The development of biopharmaceuticals can be roughly divided into the expression of the product in living cells, the upstream process development (USP) and the purification of the produced molecule, the downstream process development (DSP). In the past, great success has been achieved in the development of a more efficient USP. In particular, higher cell culture titers led to higher product concentrations. In order to keep pace with these developments, new, and cost-effective alternatives for the purification process are needed. Today, the DSP of biopharmaceutical products is mainly based on chromatographic purification steps. These have a high selectivity, which leads to a high purity after the unit operation. Unfortunately, these advantages are associated with the disadvantage of high resin costs, and a slow volumetric throughput. Consequently, chromatographic steps contribute most to the cost of DSP. However, the development of a purification process is mostly based on chromatographic steps, as platform processes are available. For monoclonal antibodies (mAb) this purification platform is based on Protein A affinity chromatography as a capture step. This step is generic for all mAbs since the binding to the adsorber is attributed to an interaction with the Fc-region present in all mAbs. Therefore, depending on the used mAb, only minor adjustments to the process are required. This makes process development easier, faster, and more efficient. Consequently, this also leads to a loss of flexibility. New biopharmaceuticals, such as alternative mAbs, virus-like particles (VLP), or nucleic acids will not fit into existing platforms. As a result, alternatives to the convenient but expensive methods have to be developed. In academia, preparative protein precipitation has already shown that it can be an easy-to-use and cost-effective alternative. In addition to salt-, or ethanol-induced precipitation, polyethylene glycol (PEG) has proven its suitability as effective precipitating agent for the purification of proteins. PEG is comparatively cheap, non-flammable, non-toxic, and already approved as an excipient in many drugs. However, several obstacles still have to be overcome in order to make industrial application possible.
In order to increase the acceptance of preparative protein precipitation in the biopharmaceutical industry, deeper knowledge of the mechanism, new development strategies and an improved monitoring of the process are necessary. Knowledge of the process is of great importance for a successful process development, and also for a successful approval by the relevant authorities. Although much research has been done in the field of protein stability, the involved interactions are not fully understood. Knowledge of the mechanism can not only help to understand protein precipitation, but also to improve process development. Today, DSP is often performed expert based, with the background of a platform process. To replace existing platform processes, flexible, fast, and material-saving process development strategies for preparative protein precipitation are missing. Furthermore, when developing a precipitation step, the purification step is often optimized as a stand-alone operation, while a concept for integrating the unit operation into the overall production process, including USP, and other polishing steps, is missing. In a preparative precipitation step, the phase transition is intentionally induced. This leads to a turbidity in the solution that makes the process step difficult to monitor. In addition, precipitants are added to the solution to cause this phase transition. In the following purification step, the complete degradation of these components must be ensured and monitored. The fact that these components are often not UV/Vis active makes common techniques inapplicable.
The overall aim of this work was to define and solve problems in the area of process development, process integration, and process monitoring. The role of hydrophobic interactions on the phase state of proteins should be investigated. With the knowledge gained about the precipitation mechanism, a mathematical description that enables model-based process development should be found. For the process development of a precipitation step for complex mAb solutions, a development platform should be generated. The influence of the previous and subsequent production steps should be taken into account. Finally, an in-line monitoring strategy should be developed that is capable of detecting the product and impurities, including the added precipitants.
In the third chapter the influence of the hydrophobic effect on the retention behavior during hydrophobic interaction chromatography (HIC), the crystallization behavior, and the thermal stability of the model protein glucose isomerase were investigated. For this purpose, a high-throughput screening platform in microbatch scale was used, which made it possible to produce a protein phase diagram with variation of the salt concentration and simultaneous change of the pH value. The HIC retention behavior could successfully be correlated to the crystal size and shape of glucose isomerase. The thermal stability was used to obtain information on the protein stability. Using a combination of HIC retention behavior and thermal stability, it was possible to successfully explain and estimate the phase behavior for glucose isomerase for the different investigated salt types, salt concentrations, and pH values. With the help of HIC, a deeper understanding of the influence of hydropbobic interactions during protein-protein interaction was obtained.
Based on these findings, in chapter four the mechanism of hydrophobic interaction during the precipitation process was used to derive a mechanistic model for preparative protein precipiation. The new model is based on an existing isotherm for HIC. It was assumed that both, adsorption to a hydrophobic ligand, and protein precipitation are mainly caused by a reduction of the hydrophobic area on the protein surface. Furthermore, it was assumed that this was due to a change in the water structure. Consequently, water was introduced as an additional component into the precipitation process, and an equilibrium between well-ordered and bulk-like ordered water was assumed. A high-throughput platform was used to generate precipitation curves for proteins of different size, surface charge, or hydrophobicity. The isotherm could describe the amount of precipitated protein as a function of the initial protein and PEG concentrations. The model was able to accurately describe the precipitation curve over the entire investigated range. Together with the ability to describe the precipitation behavior of small proteins, it is superior to existing models, such as the widely used Cohn-equation. Since the isotherm has a mechanistic foundation it is also possible to describe data beyond the calibration range. This was demonstrated with validation data that were not included into the model building.
While the previous studies mainly dealt with with pure proteins solutions, the investigations presented in chapter five concentrated on the influence of complex protein solutions and the cell culture medium on the precipitation behavior of proteins. Therefore, the precipitation behavior of two different mAbs was investigated. Using high-throughput experiments, precipitation curves of pure mAb solutions were compared to solutions, which were spiked with impurities prior to precipitation. In solutions containing contaminants, a destabilization of mAbs and the impurities was observed. This led to a reduction of the purity, but at the same time to an increase of the yield, respectively precipitation at lower PEG concentrations. In order to investigate the influence of the different components present in the complex solution, the contaminant solution was re-buffered before spiking. Thereby, small molecules were depleted while macromolecules, such as host cell proteins (HCP), or DNA, were still present. Using this experimental setup, destabilization could be attributed to mAb-macromolecular interactions. On the other hand, a stabilization of the mAbs, and also of the impurities, could be assigned to small molecules, present in the cell culture medium. This led to higher purities, but at the same time more PEG was necessary to achieve acceptable yields. The results showed how important it is to consider the process development needs of the downstream precipitation step already during the USP.
In chapter six, a process development strategy, and the corresponding development platform for an integrated precipitation and ion exchange (IEX) step was developed. By means of selective precipitation and selective resolubilisation, process-related impurities such as HCP and DNA were reduced. The subsequent cation exchange chromatography (CEX) step was used to remove product-related impurities such as mAb aggregates. In addition, CEX was used as product concentration step. By combining high-throughput process development, empirical, and mechanistic modeling, a fast, flexible, and material-saving development platform was designed. In a case study, a process for the purification of a complex mAb feedstock was developed. It was shown that the developed process is a suitable alternative to conventional Protein A affinity chromatography. With the presented process development strategy it is possible to develop a traditional batch process. In addition, the calibrated mechanistic chromatography model was used for in silico optimization of periodic counter-current chromatography (PCCC). Thus, the feasibility of developing a continuous process, using the developed platform, was demonstrated.
While the previous chapters dealt with process understanding, and new development strategies, the last part of the work was dedicated to new monitoring strategies to meet the requirements of a protein precipitation based process. In chapter seven, Fourier-transform infrared (FTIR) spectroscopy was implemented as in-line process analytical technology (PAT) for monitoring chromatographic processes. The potential of FTIR spectroscopy was demonstrated in three case studies. The first showed the possibility to distinguish two different proteins by differences in their secondary structure. Using a partial least squares (PLS) regression model, it was possible to distinguish between mAb and lysozyme during CEX elution. In a second case study, a process that separates different types of PEGylated lysozyme with CEX was monitored by FTIR spectroscopy. This also showed the possibility to detect the non UV/Vis active molecule PEG. In the context of a PEG-induced precipitation process this is of great interest, as PEG is added to the process and therefore a later depletion must be guaranteed and monitored. In the last case study, the typical process-related impurities and also the non UV/Vis active component Triton X-100 were selectively quantified from the FTIR 3D field.
In summary, this dissertation provides solutions for the development of new, flexible, and cost-effective alternatives to frequently used chromatographic based methods. The focus was set on the establishment of precipitation processes with respect to the current needs of DSP. By the investigation of the role of hydrophobic interactions, new insights into the precipitation step were gained. With the help of this, a mechanistic precipitation model could be developed. This showed the possibility to simplify the process development and to reduce material consumption at the same time. The generated process development platform enables a simple process development with regard to the previous and subsequent process steps. The developed process showed comparable results in yield, purity and aggregate reduction compared to the industrial standard, Protein A affinity chromatography. With the development of an in-line sensor based on FTIR spectroscopy strategies were developed to monitor both the product and additives such as PEG, in the subsequent purification steps.