ABS 030: High-Throughput Guided Discovery of Binders and Inhibitors Against TEV Protease

Emily Prins ¹, Sevednima Ajayebi ¹, Carl Denard ¹ ²

¹ Department of Chemical Engineering, University of Florida
² UF Health Cancer Center

The Van Wickle Journal (2026) Volume 2, ABS030

Introduction: Proteases are essential to cellular regulation but can also contribute to diseases including cancer, neurodegeneration, and infection when dysregulated. Protein-based inhibitors such as nanobodies offer a promising route for controlling protease function due to their stability and ability to access allosteric sites. However, traditional discovery methods typically identify binders first and only later determine which function as inhibitors.

To address this limitation, we previously developed High-throughput Activity Reprogramming of Proteases (HARP), a yeast surface display system that directly reports protease inhibition through dual epitope-tagged substrates and enriches inhibitory populations through multiple rounds of fluorescence-activated cell sorting (FACS). Using HARP, we identified nanobodies that inhibit the model protease tobacco etch virus protease (TEVp) and validated their inhibitory potency using FRET assays.

In parallel, we generated a nanobody library displayed on the yeast surface and used biotinylated TEVp and FACS to isolate nanobodies based solely on binding affinity. Our results demonstrate that many TEVp binders do not exhibit inhibitory activity, highlighting that binding affinity alone is insufficient to predict functional inhibition.

This work establishes HARP as a powerful strategy for identifying true inhibitory nanobodies. In future studies, we will leverage the resulting high-throughput dataset to train ensemble machine learning models capable of predicting inhibitory nanobodies beyond the starting library and exploring the broader nanobody sequence–function landscape.

Methods: A yeast surface display nanobody library was generated and screened against tobacco etch virus protease (TEVp). For inhibitory screening, High-throughput Activity Reprogramming of Proteases (HARP) was used to directly detect protease inhibition through dual epitope-tagged reporter substrates displayed on yeast cells. Multiple rounds of fluorescence-activated cell sorting (FACS) were performed to enrich inhibitory nanobody populations. In parallel, a separate binding-based selection was conducted using biotinylated TEVp to isolate nanobodies based solely on target affinity. Enriched populations were analyzed through sequencing to identify dominant clones. Candidate inhibitory nanobodies were further validated using fluorescence resonance energy transfer (FRET)-based protease activity assays to quantify inhibition. Comparative analysis between binding-selected and inhibition-selected populations was performed to evaluate the relationship between binding affinity and functional inhibition.

Results: Using HARP, we successfully identified nanobodies capable of inhibiting TEVp activity and validated their inhibitory potency through FRET assays. In contrast, binding-based selection using biotinylated TEVp enriched numerous high-affinity binders that lacked measurable inhibitory activity. Sequencing of enriched inhibitory populations revealed convergence toward distinct nanobody clones following iterative FACS selection. These findings demonstrate that binding affinity alone is insufficient to predict functional inhibition and highlight the effectiveness of HARP in selectively enriching true inhibitory nanobodies.

Discussion: This work demonstrates that functional screening approaches such as HARP provide significant advantages over traditional binding-based discovery methods for identifying inhibitory nanobodies. Our findings highlight the disconnect between target binding and functional inhibition, emphasizing the importance of activity-based selection strategies in protease engineering and therapeutic discovery. The resulting high-throughput sequence–function dataset provides a foundation for future machine learning applications aimed at predicting inhibitory nanobodies from sequence information alone. More broadly, this platform may accelerate the discovery of selective protein modulators for proteases and other therapeutically relevant targets.