Genetic encoding of a highly photostable, long lifetime fluorescent amino acid for imaging in mammalian cells

Acridonylalanine (Acd) is a fluorescent amino acid that is highly photostable, with a high quantum yield and long fluorescence lifetime in water. These properties make it superior to existing genetically encodable fluorescent amino acids for monitoring protein interactions and conformational changes through fluorescence polarization or lifetime experiments, including fluorescence lifetime imaging microscopy (FLIM). Here, we report the genetic incorporation of Acd using engineered pyrrolysine tRNA synthetase (RS) mutants that allow for efficient Acd incorporation in both E. coli and mammalian cells. We compare protein yields and amino acid specificity for these Acd RSs to identify an optimal construct. We also demonstrate the use of Acd in FLIM, where its long lifetime provides strong contrast compared to endogenous fluorophores and engineered fluorescent proteins, which have lifetimes less than 5 ns.

The blueprint was modified such the rotamer, identity, and secondary of all residues solved in the structure would remain fixed. The identity of missing residues was defined and the secondary structure of them and those residues immediately up and downstream was defined as loop. An excerpt illustrating this is shown below: In this segment, a four-residue loop (ISLN) is missing from the structure. Its secondary structure and that of the residues immediately up and downstream was defined as loop (L).
The identities of all residues included in the structure were kept constant (NATAA).

§2-Modeling of PylRS variants 41 & 82
PylRS variants were modeled in PyRosetta4 (Mac version 203 for Python 3.6). A script was written which first mutated all relevant residues to their mutant identities. The pose was then FastRelaxed with coordinates constrained to using the following settings: The relax_taskfactory included all residues with the IncludeCurrent(), RestrictToRepacking(), and InitializeFromCommandLine() task operations with the following CommandLine imported at the start of the script from a txt file: The relax_movemapfactory allowed all chi angles, backbone, bond angles, and jumps to move and prohibited bond lengths from moving.
Then, the Ala ligand for each was deleted and ligand-free structures were output to PDB files (Relaxed_41mutations.pdb & Relaxed_82mutations.pdb).

§3-Docking of Acd into PylRS variants 41 & 82
The following docking protocol was repeated 3 separate times from start-to-finish for enzyme 82.
To initially position the Acd ligand in the binding pocket, DARC was used. The positioned ligand pdb output was edited so that the LIG name was changed back to ACD. The LIG residue from the pdb output of the enzyme/ligand complex was replaced with ACD and the merged pdb (DARC_82_merge.params) was prepared for further modeling: /Applications/rosetta_bin_mac_2018.33.60351_bundle/main/source/build/src/release/macos/10.13/64/x86 /clang/9.0/static/score_jd2.static.macosclangrelease -renumber_pdb -ignore_unrecognized_res -s DARC_82_merge.pdb -out:pdb Finally, the position of Acd in the pocket was optimized in PyRosetta using the beta_nov16. Firstly, the ligand position was adjusted. A FastRelax of the ligand constrained to starting coordinates without ramping-down was performed in DualSpace, allowing chi angles, backbone, bond angles, and jumps to minimize only at the ligand residue (i.e., the enzyme remained unmodified).
Next, the ligand and all enzyme residues within 10Å of it were packed using the PackRotamersMover and the beta_nov16_soft scorefunction. The full structure was then minimized in Cartesian space using the beta_nov16_cart scorefunction and the MinMover.
Finally, the full structure was relaxed as above via FastRelax two times sequentially. The first time it was constrained to starting coordinates with ramping down of constraints. The second time all constraints were removed.
NB-due to steric constriction of the pocket, Acd could not be initially positioned well in the pocket of enzyme 41 using DARC (see Fig. 1). Therefore, to model enzyme 41, all steps of the docking protocol following DARC were performed after changing C313G back to C313 in the fully docked 82 structure.

Fig. 1. A)
The C313G mutation opens up extra space in the binding pocket of AcdRS (82) as compared to AcdRS (41). For this reason, Acd could be positioned well in the pocket of (82) but not of (41) with DARC. B) Three full docking runs were performed for AcdRS (82) and then for AcdRS(41) starting with the (82) structures. In all cases, Acd was positioned in the (82) pocket such that it could make hydrogen bonds with both W382T and with N311S. This was enabled by the steric relief from the C313G mutation. On the other hand, Acd adopted a variety of states in the (41) pocket, the first of which is productive for tRNA aminoacylation. In this state, Acd acts as a donor to the W382T residue. Note, it is unable to interact with N311S due to the steric clash from C313.