Hit to lead (H2L) optimization is a key step for drug and agrochemical discovery. A critical challenge for H2L optimization is the low efficiency due to the lack of predictive method with high accuracy. We described a new computational method called Computational Substitution Optimization (CSO) that has allowed us to rapidly identify compounds with cytochrome bc1 complex inhibitory activity in the nanomolar and subnanomolar range. The comprehensively optimized candidate has proved to be a slow binding inhibitor of bc1 complex, ~73-fold more potent (Ki = 4.1 nM) than the best commercial fungicide azoxystrobin (AZ; Ki = 297.6 nM) and shows excellent in vivo fungicidal activity against downy mildew and powdery mildew disease. The excellent correlation between experimental and calculated binding free-energy shifts together with further crystallographic analysis confirmed the prediction accuracy of CSO method. To the best of our knowledge, CSO is a new computational approach to substitution-scanning mutagenesis of ligand and could be used as a general strategy of H2L optimisation in drug and agrochemical design.

It has been difficult to directly measure the spontaneous hydrolysis rate of urea and, thus, 1,1,3,3-tetramethylurea (Me4U) was used as a model to determine the “experimental” rate constant for urea hydrolysis. The use of Me4U was based on an assumption that the rate of urea hydrolysis should be 2.8 times that of Me4U hydrolysis because the rate of acetamide hydrolysis is 2.8 times that of N,N-dimethyl-acetamide hydrolysis. The present first-principles electronic-structure calculations on the competing non-enzymatic hydrolysis pathways have demonstrated that the dominant pathway is the neutral hydrolysis via the CN addition for both urea (when pH<~11.6) and Me4U (regardless of pH), unlike the non-enzymatic hydrolysis of amides where alkaline hydrolysis is dominant. Based on the computational data, the substituent shift of free energy barrier calculated for the neutral hydrolysis is remarkably different from that for the alkaline hydrolysis, and the rate constant for the urea hydrolysis should be ~1.3×109-fold lower than that (4.2×10−12 s−1) measured for the Me4U hydrolysis. As a result, the rate enhancement and catalytic proficiency of urease should be 1.2×1025 and 3×1027 M−1, respectively, suggesting that urease surpasses proteases and all other enzymes in its power to enhance the rate of reaction. All of the computational results are consistent with available experimental data for Me4U, suggesting that the computational prediction for urea is reliable.