Vaccination plays a pivotal role in modern preventive healthcare and contributes to a global decline in infectious diseases. Efficient production of vaccines is essential to meet the growing demand which results from factors like a growing global population and increased international travel. Protein subunit vaccines are a vaccine modality that contains parts of the infectious pathogen as the active ingredients. These subunits are recognized by the immune system, which is trained to respond more effectively and reduce symptoms upon actual infection. Production of these vaccines is divided into upstream processing (USP), which involves fermentation using expression hosts, downstream processing (DSP) where the protein subunit is purified, and finally formulation where the vaccines are prepared for distribution. During the DSP, multiple chromatography modes are often used to reach the required purity. Selection of the optimal chromatographic resin types, as well as operating conditions can be expensive and time consuming. Model-based process development has the potential to speed up this selection by using computational methods to predict protein behavior. Especially in early phase development, models allow in silico screening of resins and conditions in tandem to classical experiments, reducing required material. These computational models can be divided into knowledge-driven, datadriven, or a combination thereof.

BACKGROUND: Selecting an optimal chromatography resin during biopharmaceutical downstream process development is a great challenge. This is especially the case for recombinant subunit vaccines, where product properties vary greatly and recovery often involves cell lysis, which yields a complex mixture of different host cell materials. Host cell protein (HCP) impurities may remain similar for platform processes, but their critical impact on separation efficiency is relative to specific product properties. Therefore, every process needs to be designed per product. Prior knowledge on the elution behavior of HCPs would support the identification of critical compounds. However, determining chromatographic behavior of HCPs experimentally is a time-consuming approach. RESULTS: In this work, we leverage quantitative structure–property relationship (QSPR) models calibrated with retention data of 13 commercial proteins, collected at pH 7, 8, 9 and 10 to predict the anion-exchange retention of Escherichia coli HCPs. These models use features calculated from the molecular structure to describe protein behavior, like chromatographic retention. A multilinear regression model containing two features (isoelectric point and sum of negative surface electrostatics) was able to predict the retention times of 288 HCPs accurately (error ≤ 5%). Moreover, we identified the key attributes missing in the training dataset, which is important to increase model performance in the future. CONCLUSION: This work showcases how chromatographic data obtained using commercial proteins can be translated to a clarified E. coli lysate to accelerate chromatography resin selection for new products.

Protein-based biopharmaceuticals require high purity before final formulation to ensure product safety, making process development time consuming. Implementation of computational approaches at the initial stages of process development offers a significant reduction in development efforts. By preselecting process conditions, experimental screening can be limited to only a subset. One such computational selection approach is the application of Quantitative Structure Property Relationship (QSPR) models that describe the properties exploited during purification. This work presents a novel open-source Python tool capable of extracting a range of features from protein 3D models on a local computer allowing total transparency of the calculations. As open-source tool, it also impacts initial investments in constructing a QSPR workflow for protein property prediction for third parties, making it widely applicable within the field of bioprocess development. The focus of current calculated molecular features is projection onto the protein surface by constructing surface grid representations. Linear regression models were trained with the calculated features to predict chromatographic retention times/volumes. Model validation shows a high accuracy for anion and cation exchange chromatography data (cross-validated R² of 0.87 and 0.95). Hence, these models demonstrate the potential of the use of QSPR to accelerate process design.

Mechanistic models mostly focus on the target protein and some selected process- or product-related impurities. For a better process understanding, however, it is advantageous to describe also reoccurring host cell protein impurities. Within the purification of biopharmaceuticals, the binding of host cell proteins to a chromatographic resin is far from being described comprehensively. For a broader coverage of the binding characteristics, large-scale proteomic data and systems level knowledge on protein interactions are key. However, a method for determining binding parameters of the entire host cell proteome to selected chromatography resins is still lacking. In this work, we have developed a method to determine binding parameters of all detected individual host cell proteins in an Escherichia coli harvest sample from large-scale proteomics experiments. The developed method was demonstrated to model abundant and problematic proteins, which are crucial impurities to be removed. For these 15 proteins covering varying concentration ranges, the model predicts the independently measured retention time during the validation gradient well. Finally, we optimized the anion exchange chromatography capture step in silico using the determined isotherm parameters of the persistent host cell protein contaminants. From these results, strategies can be developed to separate abundant and problematic impurities from the target antigen.

Purification of recombinantly produced biopharmaceuticals involves removal of host cell material, such as host cell proteins (HCPs). For lysates of the common expression host Escherichia coli (E. coli) over 1500 unique proteins can be identified. Currently, understanding the behavior of individual HCPs for purification operations, such as preparative chromatography, is limited. Therefore, we aim to elucidate the elution behavior of individual HCPs from E. coli strain BLR(DE3) during chromatography. Understanding this complex mixture and knowing the chromatographic behavior of each individual HCP improves the ability for rational purification process design. Specifically, linear gradient experiments were performed using ion exchange (IEX) and hydrophobic interaction chromatography, coupled with mass spectrometry-based proteomics to map the retention of individual HCPs. We combined knowledge of protein location, function, and interaction available in literature to identify trends in elution behavior. Additionally, quantitative structure–property relationship models were trained relating the protein 3D structure to elution behavior during IEX. For the complete data set a model with a cross-validated R2 of 0.55 was constructed, that could be improved to a R2 of 0.70 by considering only monomeric proteins. Ultimately this study is a significant step toward greater process understanding.

Optimizing a biopharmaceutical chromatographic purification process is currently the greatest challenge during process development. A lack of process understanding calls for extensive experimental efforts in pursuit of an optimal process. In silico techniques, such as mechanistic or data driven modeling, enhance the understanding, allowing more cost-effective and time efficient process optimization. This work presents a modeling strategy integrating quantitative structure property relationship (QSPR) models and chromatographic mechanistic models (MM) to optimize a cation exchange (CEX) capture step, limiting experiments. In QSPR, structural characteristics obtained from the protein structure are used to describe physicochemical behavior. This QSPR information can be applied in MM to predict the chromatogram and optimize the entire process. To validate this approach, retention profiles of six proteins were determined experimentally from mixtures, at different pH (3.5, 4.3, 5.0, and 7.0). Four proteins at different pH's were used to train QSPR models predicting the retention volumes and characteristic charge, subsequently the equilibrium constant was determined. For an unseen protein knowing only the protein structure, the retention peak difference between the modeled and experimental peaks was 0.2% relative to the gradient length (60 column volume). Next, the CEX capture step was optimized, demonstrating a consistent result in both the experimental and QSPR-based methods. The impact of model parameter confidence on the final optimization revealed two viable process conditions, one of which is similar to the optimization achieved using experimentally obtained parameters. The multiscale modeling approach reduces the required experimental effort by identification of initial process conditions, which can be optimized.