Pseudomonas putida CBB5 has evolved the ability to metabolize caffeine and other methylxanthines to xanthine using a set of five enzymes, NdmABCDE. NdmABC are N1-, N3-, and N7-specific N-demethylases, respectively, that are part of the multicomponent Rieske oxygenase family. Structural investigations of NdmA, NdmB, and NdmD were conducted along with determining the applicability of harnessing this N-demethylation system for the bioproduciton of methylxanthines. Protein co-expression and purification of NdmA and NdmB confirmed the presence of an NdmAB complex that is constructed to perform N-demethylation. The interactions of NdmD and NdmAB were then elucidated because NdmD serves as the sole Rieske reductase for this set of enzymes and transfers electrons to each catalytic site. NdmD is unique amongst Rieske reductases because it contains an extra Rieske [2Fe-2S] cluster at the N-terminal end. The hypothesis is the Rieske [2Fe-2S] cluster on NdmD is used in conjunction with the Rieske-less NdmC enzyme and structural subunit NdmE to carry out the N7-demethylation of 7-methylxanthine to xanthine, but is not required for activity with the NdmA and NdmB enzymes. The results support the hypothesis and expand it by suggesting that the extra Rieske [2Fe-2S] cluster can be used as a secondary electron transport pathway to NdmAB. NdmA converts caffeine to theobromine, which is further N3-demethylated by NdmB, resulting in 7-methylxanthine. However, NdmA exhibits a slight promiscuity toward the N3-methyl group, resulting in 1.5% of caffeine being convertedto paraxanthine. Analysis of the NdmA and NdmB structures identified that only two of the nine amino acids in the binding pocket differ between NdmA and NdmB. Mutation of the two unique amino acids in NdmA to mimic the NdmB active site produced a mutant enzyme with a paraxanthine:theobromine ratio of at least 3:1, over a 100-fold improvement from the wild-type ratio (1:39). Additionally, a peptide loop near the active sites also differs between NdmA and NdmB. Mutation of the NdmA loop sequence to match that of the NdmB loop further increased the yield of paraxanthine. This research confirms that biocatalytic production of paraxanthine from caffeine is achievable and begins to optimize the starting reaction conditions.