The Chemical Mechanism and Evolutionary Development of Beetle Bioluminescence

Keith V. Wood

Photochem. Photobiol., Vol. 62, pp. 662–673 (1995)

Luciferin is unique to luminous beetles. Unlike fungal bioluminescence where the substrate (caffeic acid) is a common metabolite, D-luciferin doesn't exist anywhere else in biochemistry. It has to be actively synthesized by dedicated enzymes. That's why the substrate problem is so much harder for the firefly system than for the fungal one.
The stereochemistry is non-negotiable. Luciferase only uses D-luciferin. L-luciferin gets adenylated but can't proceed to the light-producing oxygenation step because the C4 proton is misaligned.
One ATP per photon, 0.88 quantum yield. The energy for light comes from O₂ oxidation, not ATP. ATP just activates the luciferin. Metabolic cost to your plant is essentially one ATP per photon. Remarkably cheap, and nearly every luciferin molecule produces a photon. For fungal: ~1–2 out of 100 → quantum yield of ~0.017. The vast majority of the chemical energy is wasted as heat.
No post-translational modifications needed. No prosthetic groups, no disulfide bonds, no glycosylation. Luciferase is functional the instant it's translated. It folds into a functional enzyme straight off the ribosome. Clean heterologous expression in plants.
Luciferase evolved from plant 4-coumarate:CoA ligase. Luciferase is essentially a CoA synthetase that learned to oxidize instead of form thioesters. The first half of the reaction (adenylation) is identical to 4CL. Plants are full of these enzymes. Luciferase isn't alien to plant metabolism — it's a cousin of enzymes the plant already uses.
This evolutionary relationship validates your pathway chemistry. If luciferase evolved from aromatic acid CoA ligases, the luciferin biosynthetic pathway likely evolved in the same metabolic neighborhood: aromatic intermediates, CoA chemistry, cysteine conjugation with quinones.
Luciferin existed before luciferase. Wood argues the substrate came first, probably from pigment chemistry. Benzothiazole structures appear in pigmentation pathways. An ancestral CoA synthetase encountered this pre-existing molecule, accidentally adenylated it, and the faint resulting glow was selectable. This means the biosynthetic enzymes are older than the bioluminescence function and may have had (and still have) other metabolic roles.
Emission color is already optimal. Wild-type P. pyralis emits yellow-green at ~560 nm, near-optimal for human scotopic (dark-adapted) vision. No color engineering needed. A few point mutations can shift emission from green to red.