Enzymatic Promiscuity and the Evolution of Bioluminescence

• The defining framework: firefly luciferase is a bifunctional enzyme, and luciferase activity is latent in many ACSLs. Photinus pyralis firefly luciferase (Fluc) catalyzes both light emission from D-luciferin and ligation of CoA to long-chain fatty acids ≥12 carbons, the same enzyme, the same active site, two different chemistries selected by substrate identity. This is not vestigial; the ACSL activity is robust enough that Fluc has been characterized as a legitimate fatty acyl-CoA synthetase in its own right (Oba 2003). The review's organizing argument is that this bifunctionality is the key to understanding bioluminescence evolution: luciferase activity did not arise de novo in fireflies but was a latent capability already present in ancestral ACSLs, only needing the right luciferin substrate to manifest. The evolutionary problem then becomes biosynthesizing the luciferin, not engineering the luciferase.
• Latent luciferase activity in non-luminous insects is experimentally demonstrated. This is the most striking section. Drosophila CG6178 (a non-luminous fruit fly's ACSL with ~40% identity to Fluc) cannot use natural D-luciferin but produces light efficiently with synthetic luciferin analogs like CycLuc2 (Mofford 2014). Same result with AbLL, the ACSL ortholog from the non-luminescent click beetle Agrypnus binodulus. The interpretation: many insect ACSLs are “latent luciferases,” they can run the bioluminescent chemistry but lack a productive natural substrate. The chemistry of light emission is inherent to the luciferin ester (treating any luciferin ester with base in DMSO under O₂ produces light without any enzyme at all), so any ACSL that can adenylate a luciferin should give bioluminescence. The selectivity barrier between non-luminescent ACSLs and luciferases is much lower than the sequence divergence suggests.
• A single-residue conversion of an ACSL into a luciferase. Oba 2009 mutated AbLL at one position (restoring serine 347, which forms a water-mediated hydrogen bond to the benzothiazole nitrogen of D-luciferin) and converted it into a weakly bioluminescent enzyme on natural D-luciferin. One residue. The evolutionary distance from latent-luciferase ACSL to real-luciferase is in principle a single-mutation walk, given a substrate. This has direct implications for the ESM2 deep mutational scanning and ProteinMPNN inverse-folding work in the project, it suggests that the ACSL/luciferase functional landscape is gradient-rich and that engineered variants should not be expected to require many simultaneous changes.
• Domain-swap chimeras define the functional logic of the two-domain architecture. Oba 2006 connected Fluc's N-terminal domain (residues 1 to 437) to CG6178's C-terminal domain (residues 436 to 544) and got a functional luciferase. The reciprocal chimera (CG6178 N-terminal + Fluc C-terminal) had only ACSL activity. The N-terminal domain is what defines luciferase activity; the C-terminal domain can be substituted with a homologous adenylating-enzyme C-terminus without losing function. This makes the substrate-binding pocket on the N-terminal domain (specifically the residues that bind the luciferin benzothiazole ring) the key engineering target, and explains why the Conti 1996 architecture is the right structural prior for any luciferase engineering work.
• Detailed mechanism with the SET (single electron transfer) light-emitting step. D-luciferyl-AMP is deprotonated at C4 to form an enolate-stabilized carbanion, which reduces molecular O₂ via single-electron transfer to give superoxide and a C4 radical; the radicals recombine to form a C4-peroxide; intramolecular nucleophilic attack of the peroxide on the AMP-ester carbonyl displaces AMP and forms a dioxetanone; the O-O bond breaks, releasing CO₂ and excited-state oxyluciferin; oxyluciferin radiatively relaxes to ground state. Two key catalytic lysines (K443 for oxidation, K529 for adenylation) are both contributed by the C-terminal domain and rotate in/out of the active site as the domain rotates. Worth knowing for two reasons: it pins down which residues matter for which half-reaction, and it explains why in vitro luciferase assays add CoA (to scavenge dehydroluciferyl-AMP, the dark-reaction byproduct that accumulates and inhibits the enzyme).
• The dark reaction, dehydroluciferyl-AMP, is a practical concern for any luciferase assay. In step 4 of the mechanism, instead of forming the C4-peroxide, the C4 radical can lose a hydrogen at C5 to give dehydroluciferyl-AMP, a non-luminescent byproduct that accumulates in the active site and inhibits further catalysis. Adding CoA to luciferase reactions converts dehydroluciferyl-AMP to dehydroluciferyl-CoA, which dissociates and frees the enzyme. This is why the canonical firefly luciferase reaction buffer (Promega's “Luciferase Assay Reagent” and similar formulations) contains CoA, and why endogenous CoA pools in plant peroxisomes are a non-trivial parameter for in planta luminescence intensity.
• Plant adenylating enzymes are explicitly proposed as candidate luciferases. This section is short but consequential: Adams & Miller note that Arabidopsis thaliana expresses peroxisome-targeted ACSL homologs that adenylate bulky jasmonate pathway intermediates (e.g., At5g63380, which makes OPDA-CoA from 12-oxo-phytodienoic acid). They suggest these enzymes “may be capable of adenylating molecules capable of acting as luciferins” or could “serve as models to develop new luciferases.” The implication is that the firefly luciferase chassis is not the only adenylating enzyme that could plausibly catalyze bioluminescence in a plant context, native plant peroxisomal ACSLs might independently produce some baseline light from D-luciferin or a luciferin analog. This is both an opportunity (plant ACSLs as alternative luciferase chassis) and a complication (native plant ACSLs in the same peroxisomal compartment as luc2 might compete for substrate or generate dehydroluciferyl-AMP that inhibits luc2).
• The L → D luciferin epimerization is mediated by Fluc's own ACSL activity. In vivo, fireflies make L-luciferin first (the BQ + L-cysteine chemistry from Kanie 2016 / Oba 2013) and then epimerize to D-luciferin. The proposed mechanism (Niwa 2006, Maeda 2017): Fluc enantioselectively thioesterifies L-luciferin to L-luciferyl-CoA via its own ACSL activity → the L-luciferyl-CoA epimerizes to D-luciferyl-CoA → thioester hydrolysis releases D-luciferin. This means luc2 itself is part of the racemization pathway in vivo, alongside whatever thioesterase performs the final hydrolysis (Maeda 2017 used bacterial TESB; Zhang 2020 demonstrated AteACOT1 as the firefly version). Direct project relevance: in tobacco, TU1 (luc2+SKL) and TU4 (ACOT9) cooperate not only to use D-luciferin but to make it from L-luciferin produced by the spontaneous BQ + L-cys chemistry, two roles for the luc2 enzyme, both required.
• The benzoquinone-detoxification origin hypothesis is anticipated here. Adams & Miller speculate two years before de Souza 2022 that “some luciferin [may] have been fortuitously produced by the reaction of cysteine with benzoquinone in related beetles,” citing Bombardier beetles' use of benzoquinone for defensive sprays as evidence that benzoquinone chemistry is widespread in non-luminous beetles. The de Souza 2022 cysteine-rescue / quinone-detoxification thesis is essentially the experimental confirmation of this speculation.

Bottom line for the project: This is the foundational review for the framework that the entire firefly luciferase engineering and biosynthesis literature operates within. Three concrete uses for the bibliography. First, in any “why firefly luciferase as the chassis” pitch context, Adams & Miller is the most authoritative single review of why ACS-family enzymes are uniquely well-positioned to function as luciferases, the bifunctionality, the latent activity in non-luminous orthologs, the single-residue evolutionary path, and the broad ANL-family substrate flexibility together make a much stronger “why this enzyme family” argument than any single primary paper. Second, the plant-ACSL section flags a genuine experimental issue worth keeping in mind for Phase 2/Phase 3: native N. tabacum peroxisomal ACSLs (the tobacco orthologs of Arabidopsis At5g63380 and related jasmonate-pathway adenylating enzymes) co-occupy the same compartment as TU1 luc2+SKL and might either contribute baseline D-luciferin → light activity in luc2-free controls or generate dehydroluciferyl-AMP byproducts that inhibit luc2, either way, native peroxisomal ACSL transcripts in agroinfiltrated tobacco are worth checking against the construct's predicted background, especially when interpreting low-luminescence phenotypes. Third, the L → D racemization story confirms that luc2 itself is part of the stereochemical conversion pathway, which means TU1 and TU4 are functionally coupled in a way that affects design priorities: the ratio of TU1:TU4 expression matters more than each gene's expression in isolation, because both are needed to drive the L-luciferyl-CoA → D-luciferyl-CoA → D-luciferin → photon cycle. Cite Adams & Miller 2020 alongside Conti 1996 (architecture), Nakatsu 2006 (substrate-bound structures), Gulick 2009 (conformational dynamics), and Fallon 2018 (parallel-origins genomics) as the canonical five-paper foundation for the firefly luciferase enzymology side of any project document.