Abstract
Early-stage drug discovery relies on the availability of stable protein for reliable biophysical characterization of ligand binding. However, many Plasmodium falciparum proteins are challenging to produce in heterologous systems, which limits their experimental utility. To address this, we tested whether ProteinMPNN-guided sequence design could generate stabilized surrogate constructs that retain wild-type-like structure and binding thermodynamics. Designs were generated with constraints to maintain conserved and binding-site residues for three therapeutically relevant targets: PfBDP1-BRD, PfBDP4-BRD, and PfK13-KREP. The resulting constructs showed markedly increased thermal stability. Using PfBDP1-BRD as a benchmark, isothermal titration calorimetry confirmed that the stabilized variants retained wild-type-like binding thermodynamics with a known ligand. Extending this approach to other targets, a PfK13-KREP construct led to an apo structure with a binding pocket closely matching the wild type. For PfBDP4-BRD, virtual screening against a previously reported wild-type crystal structure identified putative binders, while a stabilized surrogate for this otherwise unstable target enabled their experimental validation and the determination of a 1.25 Å co-crystal structure with a newly identified inhibitor. Our findings demonstrate that computationally stabilized surrogates are practical and effective tools for robust biophysics and structure-enabled drug discovery against otherwise challenging malaria proteins.