To help chemists oxidize certain carbon-hydrogen bonds in complex organic molecules, researchers have developed a strategy to evolve enzymes capable of the task (J. Am. Chem. Soc., DOI: 10.1021/ja3073462). These enzymes can, with high yield and stereoselectivity, add hydroxyl groups at specific carbons in the antimalarial drug artemisinin.
When organic chemists want to convert a molecule’s carbon-hydrogen bonds into other chemical groups, they often turn to small inorganic catalysts. But these catalysts usually target the molecule’s most reactive C-H bond, says chemist Rudi Fasan at the University of Rochester. It’s not easy, he says, to steer the catalysts toward less reactive bonds. To target these less reactive bonds, chemists use enzymes. In particular, the family of cytochrome P450 enzymes oxidize carbons, replacing a hydrogen with a hydroxyl group. However, methods to generate bond-specific enzymes are time-consuming. The resulting enzymes also struggle to add the hydroxyl groups in a stereoselective manner.
To overcome these limitations, Fasan and his colleagues developed a faster, three-step approach to generate P450s that oxidize certain C-H bonds in artemisinin. They wanted to hit the C-H bond in one of the drug’s methyl groups, as well as each of the two C-H bonds on a carbon in one of the molecule’s rings. Medicinal chemists would like to add hydroxyl groups at these positions so they can eventually convert them into other functional groups, such as fluorines, to improve the drug’s effectiveness.
The researchers selected a P450 from Bacillus megaterium as the starting point, because previous studies had shown that the enzyme could oxidize all of these C-H bonds in artemisinin. In their method’s first step, they generated a library of 100,000 P450 mutants, focusing on varying amino acids in the enzyme’s active site. In the next step, they took 12,500 of those proteins and determined which general shapes of organic molecules they preferred to react with. The chemists did so by watching how each enzyme oxidized five molecules with rings like artemisinin has. Based on these data, the team then whittled down the mutant library to 522 enzymes with minimal overlap in shape preference.
To find the enzymes that reacted readily with artemisinin, without checking each of the 522 P450 mutants, the researchers built a computer algorithm to predict enzyme reactivity. The algorithm took into account how each mutant reacted with the five molecules from the previous step and data on how well 20 randomly selected mutants oxidized artemisinin. Of the 50 enzymes the algorithm suggested were most likely to react with the drug, 40 actually oxidized artemisinin when the researchers tested them.
Three evolved enzymes oxidized one of the researcher’s target C-H bonds without affecting other bonds. When running a reaction with each enzyme, the researchers obtained yields greater than 90% of artemisinin with a hydroxyl group at the desired location. With the enzymes that modified the C-H bonds on artemisinin’s ring, each enzyme produced only the desired stereoisomer.
David Sherman, a medicinal chemist at the University of Michigan, Ann Arbor, says the stereoselectivity of the P450 mutants is impressive and unprecedented. He says the researchers should next determine if their method can produce enzymes that can react with C-H bonds in other complex molecules.
Fasan’s group is now attempting to do just that with a substrate molecule that has promising anticancer properties.