Structure-Based Affinity Tuning for Protease-Activatable Antibody Systems

Abstract

Activatable antibodies have emerged as a strategy to improve tumor selectivity by suppressing antigen binding in normal tissues and restoring activity only in response to tumor-associated stimuli. In protease-activatable antibody platforms, a masking domain blocks the antigen-binding site and is released after protease cleavage. Anti-idiotypic single-domain antibodies （dAbs） are promising masking modules because they recognize the complementarity-determining regions of a parental antibody through a defined antibody–antibody interface. However, excessive mask affinity can prevent efficient recovery of antigen binding even after cleavage. To address this limitation, this study applied structure-based affinity tuning to an anti-trastuzumab dAb mask fused to the anti-HER2 4D5 single-chain variable fragment. Structural analysis of the dAb–trastuzumab interface identified Y61 and W104 as key residues involved in masking interaction. Alanine substitution revealed that the wild-type construct maintained strong masking but showed poor recovery after TEV protease cleavage, whereas Y61A and W104A enabled efficient protease-dependent unmasking. The double mutant Y61A/W104A showed insufficient masking, indicating that excessive weakening of the interface disrupts mask function. Further tuning at W104 identified W104F as the optimal variant, providing strong masking before cleavage and efficient recovery of HER2 binding after cleavage, with an approximately 22-fold masked-to-unmasked dynamic range based on EC₅₀ values. This optimized masking strategy was further evaluated using a tumor-relevant MMP-9 cleavage system. The dAb（W104F）-MMP-9-4D5scFv construct showed an approximately 125-fold reduction in HER2 binding before MMP-9 cleavage and recovered binding to within approximately 3-fold of parental 4D5scFv after cleavage. In HER2-positive BT-474 cells, the W104F construct exhibited clear MMP-9-dependent cellular HER2 binding, whereas the wild-type construct remained inactive. These findings demonstrate that tuning the affinity of an anti-idiotypic dAb mask can optimize the functional balance between masked and activated states. This structure-based strategy provides a rational framework for developing protease-activatable antibody platforms with improved tumor selectivity and may be extendable to other antibody–mask systems.