Discovery of Red-Shifting Mutations in Firefly Luciferase Using High-Throughput Biochemistry

Abstract

Photinus pyralis luciferase (FLuc) has proven a valuable tool for bioluminescence imaging, but much of the light emitted from the native enzyme is absorbed by endogenous biomolecules. Thus, luciferases displaying red-shifted emission enable higher resolution during deep-tissue imaging. A robust model of how protein structure determines emission color would greatly aid the engineering of red-shifted mutants, but no consensus has been reached to date. In this work, we applied deep mutational scanning to systematically assess 20 functionally important amino acid positions on FLuc for red-shifting mutations, predicting that an unbiased approach would enable novel contributions to this debate. We report dozens of red-shifting mutations as a result, a large majority of which have not been previously identified. Further characterization revealed that mutations N229T and T352M, in particular, bring about unimodal emission with the majority of photons being >600 nm. The red-shifting mutations identified by this high-throughput approach provide strong biochemical evidence for the multiple-emitter mechanism of color determination and point to the importance of a water network in the enzyme binding pocket for altering the emitter ratio. This work provides a broadly applicable mutational data set tying FLuc structure to emission color that contributes to our mechanistic understanding of emission color determination and should facilitate further engineering of improved probes for deep-tissue imaging.

Figure 1

Scheme detailing our combination of bioinformatics-guided library design (selected positions are shown in magenta on the left) with deep mutational scanning to identify novel red-shifting mutations.

Figure 2

(A) Mutability of 20 SCA-identified amino acid positions, as determined from a top-plating activity screen. Each library was transformed into bacteria and embedded in a thin layer of top agar containing IPTG and d-luc. After overnight incubation, dark- and bright-field images of each plate were captured and were digitally superimposed. The numbers of total and light-emitting (active) colonies were counted, and the ratio of these values was taken to give the percent active proteins in each library. 200 total colonies were assessed per library. (B) SCA-identified positions were mapped onto the FLuc protein structure (PDB: 4G36, visualized with PyMol), where color corresponds to the mutability of that residue, and in cyan is shown the luciferyl adenylate intermediate analogue 5′-O-[(N-dehydroluciferyl)-sulfamoyl]adenosine (DLSA).

Figure 3

Color scores of each library from a screen for red-shifted mutants. (A) Active mutants were expressed in a 96-well format. The cells were lysed and combined with d-luc and ATP, then screened for luminescence using 620/40 and 528/20 nm optical filters. The ratio of these values is the color score, corresponding to a red-shifted emission. Each plate contained WT FLuc and COL2 (a known red-shifted mutant) as internal controls; all experiments and controls are shown here. (B) FLuc protein structure (PDB: 4G37, visualized with Chimera) is shown, with libraries displaying the five highest median color scores in red (N229 and T352) and orange (I237, L286, and H431), all other SCA-identified positions shown in gold, and DLSA (a luciferyl adenylate intermediate analogue) shown in cyan. Also shown is the numbered phenol–enol form of oxyluciferin. (C) The boxed region in (B) is enlarged to illustrate the proximity of strongly red-shifting positions N229 and T352 (red) and moderately red-shifting positions I237, L286, and H431 (orange) to the 6′OH on DLSA (white, colored by the heteroatom).

Figure 4

Sequence-function map of FLuc mutations at 20 SCA-identified positions. Mutants displaying visible bioluminescence in a top-plating screen were collected into an active population and those displaying a color score at least two standard deviations above the WT mean were collected into a population enriched for red-shifted enzymes. The populations were quantified by HT-Seq. Mutants that were at least 2-fold de-enriched in the active population with respect to the prescreen population are shown in gray (“inactive”). Mutants predicted to be active are sorted by enrichment in the red-shifted population with respect to the active population, as white (de-enriched), pink (enriched), and red (at least 2-fold enriched).

Figure 5

Bioluminescence emission of red-shifted mutants. (A) Mutants displaying the highest color score from each red-enriched population were isolated and expressed. The cells were lysed and briefly incubated with d-luc and ATP, then the emission was recorded from 450 to 700 nm at room temperature and pH 7.5. (B) To quantify the red-shift displayed by each mutant enzyme, two Gaussian curves were fit to each spectrum corresponding to red (∼605 nm, red dotted line) and green (∼555 nm, green dotted line) oxyluciferin emitters. Mutant color scores and relative activities were determined in independent screens of the recovered mutants (8–12 replicates per sample).