Synchrotron-based high-resolution photoemission, x-ray absorption near-edge spectroscopy, and first-principles density-functional calculations are used to examine the interaction of SO2 with pure and modified surfaces of magnesium oxide. On a MgO(100) single crystal, SO2 reacts with O centers to form SO3 and SO4 species. The bonding interactions with the Mg cations are weak and do not lead to cleavage of S-O bonds. An identical result is found after adsorbing SO2 on pure stoichiometric powders of MgO and other oxides (TiO2, Cr2O3, Fe2O3, NiO, CuO, ZnO, V2O5, CeO2, BaO). In these systems, the occupied cations bands are too stable for effective bonding interactions with the LUMO of SO2. To activate an oxide for S-O bond cleavage, one has to create occupied metal states above the valence band of the oxide. DF calculations predict that in the presence of these "extra" electronic states the adsorption energy of SO2 should increase, and there should be a significant oxide --> SO2(LUMO) charge transfer that facilitates the cleavage of the S-O bonds. In this article, we explore three different approaches (formation of O vacancies, promotion with alkali metals, and doping with transition metals) that lead to the activation of SO2 and S-O bond breaking on MgO and oxides in general. Basic principles for a rational design of catalysts with a high efficiency for the destruction of SO2 are presented.