We present a density functional theory (DFT) hybrid quantum mechanical/molecular mechanical (QM-MM) implementation developed for simulations of reactions in complex environments. It is particularly suited to study enzyme active sites or solutes in condensed phases. The method combines a QM description of the solute with a MM treatment of the environment. The QM fragment is treated using DFT as implemented in the computationally efficient program SIESTA, while the environment is treated using the Wang et al. Amber force field parametrization. We applied our new QM-MM scheme to study the conversion of chorismate to prephenate by computing the reaction energy profile in vacuo, aqueous solution and in the active site of the B. subtilis chorismate mutase enzyme. We have performed calculations for two different choices of the QM subsystem in the enzyme simulations: including only the substrate moiety and the substrate plus the charged side chains glu78 and arg90, respectively. In both cases, our results are in good agreement with experiment. The catalytic activity achieved by chorismate mutase relative to the uncatalyzed reaction in solution is due to both a minor destabilization of the substrate molecule by compression and a major electrostatic stabilization of the transition state, which reduce the activation energy of the reaction.