Controlling the Metabolic Labeling of Proteins with Non-Natural Amino Acids Using Engineered Ligand-Responsive Amino Acyl tRNA Synthetases

A wide variety of amino acyl tRNA synthetases (aaRSs) have had their substrate specificities altered through protein engineering to create mutant aaRSs that can charge tRNA with non-canonical amino acids in cells. While the activities of these mutant aaRSs can be dynamically controlled in cells using conditional promoters, aaRS can also be synthesized using non-natural amino acids as they accumulate in cells, which in some cases could negatively impact their ability to metabolically label other proteins. In addition, our ability to quickly switch aaRS-mediated metabolic labeling on and off remains limited because we have not yet discovered ways to directly control aaRS activity post-translationally in response to environmental changes. To develop prototype aaRSs with activities that are dependent upon direct binding to a chemical, we have engineered a first generation of ligand-responsive aaRSs. For these efforts, we targeted a mutant Escherichia coli methionyl tRNA synthetase (MetRS), which was previously developed to charge tRNA with azidonorleucine (Anl) in Escherichia coli. We first used a combinatorial approach to discover split MetRS whose fragments cooperatively function when each fragment is fused to a pair of proteins that interact. We then examined whether the activity of split MetRS can be regulated by either fusing the fragments to a pair of proteins whose interaction is stabilized by ligand binding or by fusing the fragments to the termini of a single protein domain that exhibits a ligand-dependent conformational change. We have found that both approaches can be used to regulate Anl labeling of newly synthesized proteins in a ligand-dependent manner. When MetRS fragments were fused to FKBP12 and the rapamycin binding domain of mTOR, metabolic labeling was significantly enhanced in growth medium containing rapamycin, which stabilizes the FKBP12-mTOR complex. In addition, fusion of MetRS fragments to the termini of the ligand-binding domain of the estrogen receptor yielded a protein whose metabolic labeling was significantly enhanced when 4-hydroxytamoxifen was added to the growth medium. Ongoing efforts are focused on assessing whether the ligand-dependent metabolic labeling mediated by our engineered proteins can be switched on faster than labeling achieved using transcriptional regulation, improving the metabolic labeling efficiencies that can be achieved with our prototype aaRS switches, and investigating the diversity of ligands that can be used to directly regulate MetRS activity. The protein switches arising from these studies are expected to be useful for extending our control over protein metabolic labeling with Anl by allowing for fast regulation of this labeling using post-translational reactions. Furthermore, this approach can be applied to metabolic labeling with additional non-natural amino acids by extending our designs to structurally-related aaRS with distinct substrate specificities. Ligand-responsive aaRSs are expected to have applications in the production of protein biomaterials, proteomic studies, and protein pharmaceutical biosynthesis.