Functional Selection of Protein Biosensors Using High-throughput Screening in Core-Shell Microparticles

Introduction: Directed evolution has sparked a new era of discovery and innovation in functional protein development. This innovative method capitalizes on the principles of natural selection by cycling through mutant library creation and selecting for novel functionalities, thus allowing scientists to replicate and expedite evolutionary processes. However, a major limitation in the field has been the low throughput in isolating and screening variants generated in a directed evolution experiment and the ability to measure complex phenotypic functions to perform efficient selection. This constrains our ability to efficiently explore the fitness landscape of protein libraries and discover functionally superior molecules.
In this work, we use core-shell microparticles termed ‘PicoShells’ [1] for screening extensive library of a calcium sensing protein – GCaMP, using a function-first selection approach to select protein based on their response in specific buffer compositions. PicoShells, picolitre-scale porous hydrogel shells, facilitate the compartmentalization, culture, and phenotypic screening and sorting of single clones via fluorescence activated cell sorting (FACS). Their semi-permeable structure promotes the exchange of nutrients and analytes while simultaneously retaining large proteins and DNA, thereby maintaining a clear link between phenotype and genotype for each isolated particle. These hydrogel microparticles seamlessly integrate into any cell screening or directed evolution process, allowing for the handling of millions of clones and the execution of complex multi-step assays, such as solution transfers and exposing encapsulated variant proteins to specific sequential assay conditions, making them perfect for sorting based on intricate functions using standard FACS instruments.

Materials and

Methods: Core-shell microparticles were fabricated microfluidically using a droplet generator and polymer precursors consisting of 4-arm polyethylene glycol and dextran (Fig. 1a). 1,4-dithiothreitol was used as the crosslinker for chemical-based gelation of the hydrogel upon droplet formation. Cells were encapsulated into the core-shell microparticles during particle fabrication, ensuring a single cell is encapsulated per particle by dilution of sample. We employed site saturation mutagenesis to create 1024 sequence variants of the jGCaMP7 biosensor's N-linker region [2]. These variants were transformed into E. coli and encapsulated within PicoShells during the microfluidic particle fabrication process. The proteins were then expressed in culture and the variants analyzed using standard FACS instruments. A three-step function-based measurement and sorting process was implemented, screening over 1.5 million colonies and selecting based on total fluorescence activity, dynamic response, and photoreversibility of the biosensor. The enriched variants were subsequently purified and functionally characterized to validate the efficacy of the function-based screening approach.

Results, Conclusions, and Discussions: Results
PicoShells can be used to encapsulate single cells and grow clonal populations by culture over several hours (Fig. 1b). We show that we can combine PicoShells with FACS to sort millions of clones expressing biosensor constructs based on multiple fluorescence properties in the presence of different concentrations of analyte (Fig. 2). Further, the PicoShells can be placed in different buffer environments allowing us to measure response of the protein variants encapsulated to varying levels of calcium and perform function-based selection based on the fluorescence response (Fig. 3). Leveraging this ability to measure function by alternating between different environmental conditions, we performed multi-function selection of clones that exhibited the highest total fluorescence, dynamic response and photoreversibility for calcium sensing (Fig. 4). We isolated 50 clones in the enriched pool and our preliminary results indicate greater than 5-fold improvement in dynamic response (dF/F) (Fig. 4c). Unique motifs of amino acids emerged as important in achieving optimum fluorescence sensor properties, and the large data sets generated can be used to develop a learning-based understanding of protein sequence to function characteristics. Leveraging standard FACS, we achieve an unprecedented throughput of ~106 clones and sorting based on unique functionality, massively reducing the timeline for library screening experiments.

PicoShells are a powerful tool for high-throughput screening and analysis of complex functions that are challenging to measure using conventional tools. This technology has the potential to advance high throughput function-based screening, accelerating directed evolution research paving the way for novel discoveries and advancements in protein engineering, biomanufacturing and cell engineering.

Acknowledgements (Optional):