Mechanistic Insights into Chemoselective Reduction of Engineered Cysteine Residues in Antibodies Using Computational Modeling and Rationally Designed Phosphines.

Site-specific antibody-drug conjugates (ADCs) offer improved stability, efficacy, and toxicity relative to conventional ADCs. However, the chemoselective reduction of engineered cysteine residues employing phosphine-based reductants remains incompletely understood and difficult to generalize. Here, we integrate experimental screening with computational modeling to elucidate the mechanism of selective reduction and to derive principles for de novo design of chemoselective phosphine reductants. We propose a two-factor computational framework guided by phosphine accessibility to each disulfide site on the antibody surface and intrinsic reactivity, captured by the TS1 activation barrier, to explain selective versus nonselective behavior. This framework successfully rationalizes the activity of phosphine reductants toward engineered cysteine residues across two distinct engineered cysteine sites on the IgG1 backbone (S375C in the Fc and S168C in the Fab). Guided by these structure-function insights, we developed new reagents with improved chemoselectivity. Finally, by exploiting site- and reagent-dependent differences in accessibility and reactivity, we demonstrated the feasibility of a sequential strategy to prepare homogeneous dual-payload ADCs without complex linkers or additional antibody engineering. These findings provide broadly applicable guidance for phosphine selection and design to achieve chemoselective, site-specific conjugation.