This thesis includes the work of two separate projects, studies on pyrimidine degrading enzymes and studies on in vitro evolved enzymes. The common denominator of both projects was the use of structural information to explain functional effects, observed in the studied biocatalysts.

In humans, and other eukaryotic organisms, the nucleobases uracil and thymine are catabolized by the reductive pyrimidine degradation pathway. This pathway is one of the factors that control the pyrimidine nucleotide concentrations in a cell. Furthermore, it is the main clearance route for pyrimidine analogues, often used as cancer drugs, like 5-fluorouracil and other fluoropyrimidines. Deficiencies in any of the enzymes, involved in this pathway, can lead to a wide range of neurological disorders, and possibly fatal fluoropyrimidine toxicity in cancer patients. Two out of the three involved enzymes, dihydropyrimidine dehydrogenase (DPD) and β-ureidopropionase (βUP), were studied in the first project of this thesis. This resulted in the first crystal structure of a human β-ureidopropionase variant, which could be used to explain functional characteristics of the enzyme. Structural analyses on novel DPD variants, found in patients suffering from DPD deficiency, could explain the decrease in catalytic activity of these enzyme variants. This strategy, of using structural information to predict functional effects from sequential mutations, has the potential to be used as a cheap and fast first assessment of possible deficiencies in this pathway.

Enzymes are, however, not only involved in many diseases, but also used for industrial applications. The substitution of classical organic synthetic reactions with enzyme catalyzed reactions usually has a beneficial influence on environmental pollution, as illustrated in the principles of Green Chemistry. The major drawback of the use of enzymes for these purposes is their natural selectivity towards a small group of possible substrates and products, which often do not have the desired composition or conformation for an industrial application. In order to improve an enzyme for industrial purposes, the alcohol dehydrogenase ADH-A, from Rhodococcus ruber, was subjected to a semi-rational approach of directed evolution, using iterative saturation mutagenesis (ISM), in the second project of this thesis. This resulted in different enzyme variants that showed the desired improvements in activity. Most functional improvements could be rationalized with the help of structural information and molecular dynamics simulations. This showed that artificial protein design has the potential to produce enzyme variants capable of substituting many organic synthetic reactions, and that structural information can play a key role in the designing process.