Firefly Genomes Illuminate Parallel Origins of Bioluminescence in Beetles

The first comprehensive firefly genome paper. Three reference genomes: Photinus pyralis (your source organism), Japanese Aquatica lateralis, and the Caribbean click beetle Ignelater luminosus. The P. pyralis assembly is chromosome-scale (Hi-C scaffolded into 11 linkage groups, including the X) and BUSCO-complete at 97.2%. Every comparative firefly analysis since this paper traces back to these genesets.

• Bioluminescence evolved independently at least twice in beetles. Both firefly Luc and click beetle Luc derive from peroxisomal fatty acyl-CoA synthetase (PACS) ancestors, but via independent neofunctionalization events, confirmed by synteny analysis and a positive-selection signal (aBSREL, dN/dS = 3.98) on the click beetle ancestral branch. Convergent solution from a shared raw material. Implication for protein engineering: luciferase activity is a comparatively easy novelty to evolve from PACS scaffolds, which matters if you ever want to do directed evolution on the enzyme.
• Two firefly luciferases, Luc1 and Luc2, and a nomenclature trap. Luc1 is the bright adult/larval lantern luciferase originally cloned by de Wet 1985. Luc2 is a paralog producing the dim glow of eggs, ovaries, and pupae. Important: iGEM's part-name “luc2” for BBa_K389004 is unrelated to Fallon's Luc2, BBa_K389004 is Fallon's Luc1. Worth knowing before citing this paper, since calling your TU1 enzyme “luc2” in a manuscript will read as the wrong gene to anyone working from the genome paper.
• The lantern HE+DE enzyme list, your candidate mining ground. Figure 5 of this paper is where PPYR_02911 came from. The authors filtered for enzymes that are simultaneously (a) highly expressed in the adult lantern (top 90th percentile), (b) differentially expressed lantern-vs-fatbody, and (c) directly orthologous between P. pyralis and A. lateralis. The intersection is the conserved core of firefly lantern biochemistry. PPYR_02911 (your CYP4C P450), PPYR_05464 (your peroxiredoxin backup), cystathionine gamma-lyase, adenylate kinase, and peroxisomal biogenesis factors all live here.
• Peroxisomal targeting is structural to the bioluminescence reaction. Luc1 has a C-terminal PTS1 (SKL); the combined adenylyl-sulfate kinase / sulfate adenylyltransferase (ASKSA) that produces the PAPS cofactor for luciferin sulfonation also has a PTS1 specifically in fireflies (and not in their non-luminous orthologs); PEX11 (peroxin 11, peroxisome biogenesis) is HE+DE in the lantern. The peroxisome is where the chemistry happens. This justifies the C-terminal SKL tags you're appending to luc2 in TU1 and BGLU46 in TU3, the cofactor supply and organelle machinery are already set up for peroxisome-localized substrates.
• Cystathionine gamma-lyase HE+DE supports a closed-loop cysteine pool. The enzyme cleaves cystathionine to cysteine + α-ketobutyrate + ammonia. Its lantern enrichment fits Okada 1974's old claim that cysteine is regenerated from oxyluciferin and recycled, meaning the photocyte runs a tighter cysteine economy than de novo synthesis alone would suggest. Connects directly to Oba 2013's two-cysteine incorporation model: the lantern is set up to reuse the cysteines it consumes.
• The luciferin biosynthesis pathway is explicitly still open. Fallon notes that “the enzymes of the de novo biosynthetic pathway for firefly luciferin remain unknown,” and explicitly raises the possibility, given Kanie 2016's spontaneous benzoquinone+cysteine condensation, that part of the pathway may not even be enzymatic. This paper assembles the candidate list but does not solve it. That gap is the project space your work occupies.

Bottom line for the project: This is the genome paper. PPYR_02911 (TU2), BGLU46 (TU3), ACOT9 (TU4), every Twist order you have outstanding was either pulled directly from this paper's geneset or identified via its lantern expression analysis. The SKL-targeting strategy on TU1 and TU3 is justified by the peroxisomal localization evidence here. If Oba 2013 is the biochemical ground truth, Fallon 2018 is the genetic source code, and the unresolved candidate list it leaves behind is exactly what you're building constructs to test.