Very basically there are conservation laws coming from fundamental attributes of the theories we have developed to describe natural phenomena.
Energy is an attribute of matter. Matter is built up by fundamental particles according to the physical laws governing the existence of these particles. There is a hierarchical organization where different frameworks of theory are needed in order to describe and predict the behavior of matter, but still the fundamental conservation laws hold. Elementary particles form protons and neutrons, protons and neutrons form nuclei, nuclei and electrons form atoms and molecules, atoms and molecules form solids liquids gases and plasma.
When studying elementary particles, for example, the $E=mc^2$ condition is dominant to what happens when an electron and a positron collide. The total energy of the collision is distributed among elementary particles . The result of annihilation of the original input is not reversible in any reasonable way, because in many cases one would have to collide many particles in order to build up the original electron and positron. Look at page 5 of this link . to see how complicated annihilation of the two simplest elementary particles is.
When we go one step in complexity, protons colliding on antiprotons, then again the total energy is distributed to many particles when the antiprotons annihilate , but reversibility is unthinkable experimentally. Antiprotons are created in even higher energy collisions at great effort and expense .
Going to the complexity of atom on antiatom a second level of difficulty of the creation of antiatoms enters the game, and from then on, molecules and solids etc, annihilation cannot be done in the lab, because there is no naturally found bulk antimatter and it cannot be created with our technology.
We do use the $ E=mc^2$ relation in getting energy from nuclei according to the differences in the binding energy of the nucleons in the nucleus. This is a small part of the energy of the nucleus in each case and certainly not reversible in any sensible way.