Protein phase behavior characterization is essential during different stages in biopharmaceutical process development. For example, induced crystallization can be of interest for purification processes or 3-D structure generation, while stable product formulations are desired for long-term storage. Protein phase behavior is the result of protein interactions, which can involve other protein molecules, the solvent, and cosolutes. Protein properties, such as surface charge or secondary structure, determine the strength and type of these interactions. In turn, protein properties are influenced by formulation conditions such as pH or salt type. Interplay between formulation conditions and protein properties makes characterization of protein phase behavior a multidimensional challenge.<br/><br/>Protein phase diagrams are used to map out condition effects on protein phase behavior. These diagrams can also be used as a foundation for understanding, manipulation, and prediction of phase behavior. For phase diagram construction, high-throughput screening experiments are used to cover a large range of conditions. Due to its time-consuming and trial-and-error nature, it is desired to move towards knowledge-based and automated formulation screening. This can be achieved by automation and miniaturization of high-throughput experiments, but also by generating a better understanding of the relation between protein properties and corresponding protein phase behavior. These innovations led to large biological datasets, which resulted in a need for advanced data processing and evaluation methods. Insights obtained from these datasets can aid protein phase behavior understanding, but also aid model development using in silico biophysical protein properties extracted from 3-D structures.<br/><br/>This work shows case studies to present an overview of the current state of our high-throughput phase diagram preparation and subsequent data analysis workflow. The workflow has been set up to maximize speed, eliminate subjective and labor-intensive analysis, and to move towards phase behavior prediction. This includes the use of a liquid handling station, automatic phase behavior monitoring, microfluidic absorption measurements, image recognition and subsequent image-based feature extraction algorithms, multidimensional empirical protein analytics, advanced data visualization approaches, and an in-house developed high-throughput 3-D protein structure refinement pipeline.

Automated, high-throughput, and computational methods for construction and analysis of protein phase diagrams

2019 Spring National Meeting

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.