Engineered enzyme selectivity is frequently judged by improved substrate preference or product distributions, yet the kinetic origin of those changes is often unresolved. Conformational dynamics offers a mechanistic explanation when mutations redistribute catalytically competent states, change interconversion rates, or alter access to competing reaction channels. This critical review distinguishes the measurable endpoints of substrate discrimination from assignments to binding, chemical conversion, product release, and protein-state exchange. Evidence from evolved hydrolases, isomerases, aminotransferases, designer catalysts, and computationally engineered biocatalysts is examined according to the strength of kinetic, structural, spectroscopic, and prospective design tests. Several studies link altered ensembles to functional improvement, but equivalent endpoint activity can also conceal different rate-controlling steps. Computational sequence-to-function models can broaden experimental search without independently demonstrating a conformational mechanism. A conditional design synthesis therefore couples explicit selectivity objectives with matched-substrate kinetics, transient measurements, orthogonal dynamics probes, and held-out sequence or substrate evaluation. The proposed interpretation prioritizes falsifiable alternatives and recognizes stability, expression, assay environment, and model sampling as possible confounders. Dynamics becomes a useful design variable when its predicted state change, measured kinetic consequence, and substrate-dependent outcome are jointly supported; absent that convergence, selectivity improvement remains an important engineering result without a uniquely identified cause.
