19F NMR is a powerful tool for monitoring protein conformational changes and interactions; however, the inability to site-specifically introduce fluorine labels into proteins of biological interest severely limits its applicability. Using methods for genetically directing incorporation of unnatural amino acids, we have inserted trifluoromethyl-l-phenylalanine (tfm-Phe) into proteins in vivo at TAG nonsense codons with high translational efficiency and fidelity. The binding of substrates, inhibitors, and cofactors, as well as reactions in enzymes, were studied by selective introduction of tfm-Phe and subsequent monitoring of the 19F NMR chemical shifts. Subtle protein conformational changes were detected near the active site and at long distances (25 Angstrom). 19F signal sensitivity and resolution was also sufficient to differentiate protein environments in vivo. Since there has been interest in using 19F-labeled proteins in solid-state membrane protein studies, folding studies, and in vivo studies, this general method for genetically incorporating a 19F-label into proteins of any size in Escherichia coli should have broad application beyond that of monitoring protein conformational changes.

A straightforward protocol for the site-specific incorporation of a 19F label into any protein in vivo is described. This is done using a plasmid containing an orthogonal aminoacyl-tRNA synthetase/tRNA(CUA) that incorporates L-4-trifluoromethylphenylalanine in response to the amber codon UAG. This method improves on other in vivo methods because the 19F label is incorporated into only one location on the protein of interest and that protein can easily be produced in large quantities at low cost. The protocol for producing 19F-labeled protein is similar to expressing protein in Escherichia coli and takes 4 d to obtain pure protein starting from the appropriate vectors.

The ability to site-specifically incorporate a diverse set of unnatural amino acids (>30) into proteins and quickly add new structures of interest has recently changed our approach to protein use and study. One important question yet unaddressed with unnatural amino acids (UAAs) is whether they can improve the activity of an enzyme beyond that available from the natural 20 amino acids. Herein, we report the >30-fold improvement of prodrug activator nitroreductase activity with an UAA over that of the native active site and a >2.3-fold improvement over the best possible natural amino acid. Because immense structural and electrostatic diversity at a single location can be sampled very quickly, UAAs can be implemented to improve enzyme active sites and tune a site to multiple substrates.