A computational modeling/protein engineering approach was applied to probe H234, C457, T509, Y510, and W587 within Saccharomyces cerevisiae oxidosqualene-lanosterol cyclase (ERG7), which spatially affects the C-10 cation of lanosterol formation. Substitution of Trp587 to aromatic residues supported the "aromatic hypothesis" that the π-electron-rich pocket is important for the stabilization of electron-deficient cationic intermediates. The Cys457 to Gly and Thr509 to Gly mutations disrupted the pre-existing H-bond to the protonating Asp456 and the intrinsic His234:Tyr510 H-bond network, respectively, and generated achilleol A as the major product. An H234W/Y510W double mutation altered the ERG7 function to achilleol A synthase activity and generated achilleol A as the sole product. These results support the concept that a few-ring triterpene synthase can be derived from polycyclic cyclases by reverse evolution, and exemplify the power of computational modeling coupled with protein engineering both to study the enzyme's structure-function-mechanism relationships and to evolve new enzymatic activity.