Enzymes are a class of protein molecule that catalyze chemical reactions. And by catalyze, I mean 'make them go faster'. Enzymes are pretty important to you and me. We rely on them to make otherwise ridiculously slow reactions occur at a pace that is suitable for life as we know it. In other words, we can't live without 'em. Of course, neither can mean pathogenic bacteria and viruses, so one of our main aims over the past few decades has been to identify molecules (i.e. drugs) that can stop bacterial/viral enzymes without interfering with our own. Helping our efforts in this respect is that enzymes are usually highly specific, which is to say that they often target a specific molecule (the substrate) whilst leaving other ones alone. So if we want to find new drug, maybe all we have to do is find a molecule that has a similar gemometry and chemical properties to the substrate, so that it binds to the enzyme, but is then non-reactive. There are, in fact, many drugs that work this way - a handy example is Fluorouracil, which is a second generation anti- cancer drug. Fluorouracil looks and acts like Uracil (which we use in RNA) except that it has a fluorine where there should be a hydrogen. When a critical RNA-building enzyme comes around, one of the things it has to do is remove that hydrogen, which it cannot do if instead it finds a fluorine. Consequently, the enzyme gets stuck. Cancer cells are worse off when this enzyme is inhibited because they are constantly needing to make RNA for cell division.
But there are some drawbacks to this approach. Firstly, molecules that imitate the substrate don't always bind to the enzyme all that tightly. Secondly, it is often fairly easy for bugs to evolve a version of the enzyme that still works on the real substrate, but doesn't interact strongly with the drug. A very cool way of getting around these drawbacks is to find small molecules that imitate not the substrate but the transition state of the reaction. The transition state is the highest energy structure along the path from substrate to product - that is, if a bond is being made or broken during the reaction, the transition state occurs when that bond is (more-or-less) half made or broken. The idea behind this approach is simple and elegant: All enzymes work by lowering the energy of the transition state and to do this they must bind it very tightly. Thus, not only will a 'transition state mimic' tightly bind the enzyme, it also interfereswith a step that must occur in the chemical reaction. As a consequence, any evolutionary steps that lower the tendency of an enzyme to bind the transition state mimic will also necesarily reduce the catalytic efficiency (because the ability to bind the real transition state will be similarly reduced).