Bayesian optimization is a stochastic, global black‐box optimization algorithm. By combining Machine Learning with decision‐making, the algorithm can optimally utilize information gained during experimentation to plan further experiments—while balancing exploration and exploitation. Although Design of Experiments has traditionally been the preferred method for optimizing bioprocesses, AI‐driven tools have recently drawn increasing attention to Bayesian optimization within bioprocess engineering. This review presents the principles and methodologies of Bayesian optimization and focuses on its application to various stages of bioprocess engineering in upstream and downstream processing.

Refolding is known as the bottleneck in inclusion body (IB) downstream processing in the pharmaceutical industry: high dilutions leading to large operating volumes, slow refolding kinetics and low refolding yields are only a few of the problems that impede industrial application. Solubilization prior to refolding is often carried out empirically and the effects of the solubilizate on the subsequent refolding step are rarely investigated. The results obtained in this study, however, indicate that the quality of the IB solubilizate has a severe effect on subsequent refolding. As the solubilizate contains chaotropic reagents in high molarities, it is commonly analyzed with sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). SDS-PAGE, however, suffers from a long analysis time, making at-line analytical implementation difficult. In this study, we established an at-line reversed phase liquid chromatography method to investigate the time-dependent quality of the solubilizate. To verify the necessity of at-line solubilization monitoring, we varied the essential solubilization conditions for horseradish peroxidase IBs. The solubilization time was found to have a major influence on subsequent refolding, underlining the high need for an at-line analysis of solubilization. Furthermore, we used the developed reversed phase liquid chromatography method for an in-process control (IPC). In conclusion, the presented reversed phase liquid chromatography method allows a proper control of IB solubilization applicable for tailored refolding.

Inclusion bodies (IBs) are protein aggregates formed as a result of overexpression of recombinant protein in E. coli. The formation of IBs is a valuable strategy of recombinant protein production despite the need for additional processing steps, i.e., isolation, solubilization and refolding. Industrial process development of protein refolding is a labor-intensive task based largely on empirical approaches rather than knowledge-driven strategies. A prerequisite for knowledge-driven process development is a reliable monitoring strategy. This work explores the potential of intrinsic tryptophan and tyrosine fluorescence for real-time and in situ monitoring of protein refolding. In contrast to commonly established process analytical technology (PAT), this technique showed high sensitivity with reproducible measurements for protein concentrations down to 0.01 g L $$^{-1}$$- 1 . The change of protein conformation during refolding is reflected as a shift in the position of the maxima of the tryptophan and tyrosine fluorescence spectra as well as change in the signal intensity. The shift in the peak position, expressed as average emission wavelength of a spectrum, was correlated to the amount of folding intermediates whereas the intensity integral correlates to the extent of aggregation. These correlations were implemented as an observation function into a mechanistic model. The versatility and transferability of the technique were demonstrated on the refolding of three different proteins with varying structural complexity. The technique was also successfully applied to detect the effect of additives and process mode on the refolding process efficiency. Thus, the methodology presented poses a generic and reliable PAT tool enabling real-time process monitoring of protein refolding. Graphical Real-time intrinsic fluorescence monitoring in protein refolding

Therapeutic proteins are often produced intracellularly in the bacterium Escherichia coli in aggregated form, known as inclusion bodies, which must be solubilized and refolded to recover the biologically active product. However, the industrially established refolding method of batch dilution is still carried out at low protein concentrations, leading to high water consumption, large equipment footprints, and poor sustainability. An underexplored way to increase refolding yields and thereby enable higher product concentrations after refolding is the optimization of cell lysis and inclusion body washing. In this study, we developed and evaluated a "recursive high pressure homogenization" approach, wherein wash steps and additional homogenization cycles were combined to enhance the final purity of inclusion bodies before solubilization. The degree of cell lysis, nucleic acid release and final inclusion body purity were compared to conventional "linear washing", where all homogenization cycles are completed prior to the washing of inclusion bodies. In total, seven process variations were compared and the resulting batches of inclusion bodies were solubilized and refolded to investigate the effect of early downstream processing on the product concentration in the final refolds. For the most efficient process variant, the recursive high pressure homogenization protocol led to the highest measured product concentration of 855 mg/L after refolding. Thus, the proposed recursive high pressure homogenization approach led to an estimated 18 % reduction of CO₂ footprint caused by urea, and increased the product yield per biomass from 5.17 g/kg to 7.84 g/kg compared to its linear wash counterpart - without introducing non-standard equipment or chemicals.

Ever since the potential of inclusion bodies (IBs) has been recognized, substantial advances have been made towards understanding IB processes and enabling efficient and controlled development strategies. Still, the influence of the chosen upstream processing (USP) strategy on the properties of inclusion bodies (IBs) and their refolding performance remains poorly understood. This work aims to target this challenge by investigating the influence of two chosen USP parameters, namely the specific substrate uptake rate and the temperature during induction, on IB titer, IB properties, namely IB purity, size and secondary protein structure of the IBs, as well as refolding yield of single-chain variable fragment M (scFvM) IBs. Contrary to findings in the literature, USP conditions neither had a statistically significant effect on the aforementioned IB properties nor on the refolding yield, but could clearly alter the IB titer. Our results provide detailed analytical insights on the independence of IB properties from USP conditions for this protein, while increasing the volumetric IB productivity proved feasible through variations in USP parameters. Therefore, titer maximization appears to be the sole optimization strategy for scFvM IBs and these findings may also apply to other target proteins with similar structural properties.

The pivotal role of proteins in pharmaceuticals is challenged by stability issues, making the study of inclusion bodies—a source of insoluble protein aggregates—increasingly relevant. This review outlines the critical procedures in inclusion body processing, focusing on ’mild solubilization concepts’ and refolding methodologies. Attention is afforded to the emerging role of ionic liquids with unique and tunable physicochemical properties in optimizing protein unfolding and refolding processes. The review critically assesses the existing literature at the intersection of inclusion bodies and ionic liquids, identifying recent advancements, potential applications, and avenues for future research. This comprehensive analysis aims to elucidate the complexities in efficient protein processing from inclusion bodies.