Engineering Autonomously Luminescent Plants Using the Fungal Bioluminescence Pathway

Note: this is a review article, and the correct attribution is Yu et al. with Du as last/corresponding author, Yu, Wang, Kong, Du in author order. Hao Du is the senior author behind a 2025 cluster of fungal-bioluminescence-in-plants reviews including Du 2025 Trends in Biotechnology “Biotechnologies based on the fungal bioluminescence pathway” and Zhang/Du/Lu 2025 Trends in Plant Science “Illuminating plants: autoluminescence through big data mining and metabolic optimization.” Treating this paper as one entry in a coordinated review program coming out of Du's group is the right framing, it's the New Phytologist-format version of a thesis the group has been articulating across several journals in the same year.

• The state-of-the-art review for fungal bioluminescence in plants as of late 2025. This is the most recent comprehensive review of the field's progress since Mitiouchkina 2020, with five years of subsequent optimization work synthesized in one place. The main thesis: the fungal bioluminescence pathway (FBP), Luz, HispS, H3H, CPH from Neonothopanus nambi, has been optimized “by orders of magnitude” since the 2020 baseline through a combination of metabolic engineering (boosting caffeic acid supply) and protein engineering (improving enzyme quantum yield and thermal stability). For any project working in adjacent space, this is the cleanest single citation for “where the competing fungal platform actually is right now,” as opposed to where it was when Light Bio's products first launched.
• Caffeic acid availability is the binding constraint, and the field has converged on this as the central engineering problem. Caffeic acid is the substrate the FBP draws on, and it sits at a phenylpropanoid pathway branch point where it competes with lignin biosynthesis and flavonoid biosynthesis for shared precursor pool. The implication is that the FBP has to “steal” carbon from structural and pigment metabolism, and there's a hard ceiling on how bright the system can get without compromising plant fitness. The review covers two strategies: heterologous pathway engineering (introducing tyrosine ammonia-lyase / hydroxylase routes from non-plant sources to bypass the native bottleneck) and endogenous pathway enhancement (overexpressing rate-limiting steps in the existing plant phenylpropanoid pathway). Zheng et al. 2023's BnC3′H1 (4-coumaroyl shikimate/quinate 3′-hydroxylase from Brassica napus) is the canonical example of the second strategy, overexpression alongside NPGA (null-pigment mutant from A. nidulans) achieved a threefold increase to 3 × 10¹¹ photons/min/cm², bright enough to read text by, and this is the brightness benchmark to know.
• Protein engineering is now standard practice on Luz / HispS / H3H. Shakhova et al. 2024 (cited in the review, separate paper) used directed evolution and rational design to improve quantum yield and thermal stability of the FBP enzymes, contributing further orders-of-magnitude enhancement. The review explicitly frames protein engineering as a complementary axis to metabolic engineering, meaning the field's playbook is now: (a) enhance precursor supply, (b) engineer the enzymes, (c) combine. This is a generalizable lesson for any luciferase-based system, including firefly. luc2 is structurally well-characterized (Conti 1996), there is extensive existing engineering literature on improving its quantum yield, thermal stability, and substrate specificity (the Akaluc lineage being the famous example for in vivo imaging), and analogous enzyme engineering on luc2 should be considered as Phase 4 work if the four-TU baseline construct is dim.
• Application framings: bioluminescent plants as biosensors, not just decorations. The review's section on applications goes beyond ornamental use and explicitly discusses stress-responsive promoters driving FBP gene expression, with specific use cases including soil moisture reporting (drought-stress promoter), pathogen detection (PR-protein promoter, localized light at infection sites before visible symptoms), salinity reporting, and heavy metal contamination biosensors. The framing is “bioluminescent plant as the readout layer for plant biosensors,” and the review notes synthetic biological amplifiers can be used to enhance sensitivity to weak stimuli. This is a useful framing for any commercial pitch in the bioluminescent plant space because it positions the technology as a class of bio-sensors / agricultural monitoring tool rather than a novelty consumer product.
• The implicit field consensus: fungal pathway has won the “first generation” plant bioluminescence race. Yu et al. 2025 does not actually treat the firefly system as a serious competitor for autonomous plant bioluminescence, the review is structured as if the fungal pathway is the platform, with optimization being an ongoing program rather than a competition between routes. This reflects the reality of where the commercial space is: Light Bio's Firefly Petunia product is fungal-pathway-based, the published optimization literature is fungal-pathway-based, and the authoritative reviews now treat firefly as primarily a substrate-dependent reporter system rather than an autonomous platform. For any project pitching firefly autonomous bioluminescence, this framing is the headwind, and recognizing it is necessary. The counter-framing has to be specific: different photonics (yellow-green ~560 nm vs fungal green ~520 nm), different substrate provenance (cysteine-quinone vs caffeic acid pool), different brightness ceiling (potentially higher per-photon energetics, less competition with lignin biosynthesis since L-cysteine is not a carbon-skeleton building block), and different IP landscape (Planta LLC / Light Bio do not have firefly autonomous bioluminescence in their patent portfolio).
• Phenylpropanoid pathway competition is a generalizable lesson, not a fungal-specific issue. The review's central engineering problem, heterologous pathway “stealing” carbon from native lignin and flavonoid biosynthesis, applies in modified form to firefly bioluminescence in plants too. The firefly construct's BGL2/BGLU46 step releases hydroquinone from arbutin, which sits adjacent to the same phenylpropanoid pool. The PPYR_02911 step then oxidizes HQ to BQ, and AtLAC17 in TU5 is itself a laccase that interfaces with native plant phenolic metabolism. So the “your luminescence flux competes with native lignin biosynthesis” issue is not avoided by switching to a firefly route, it just gets remapped to a different point in the same pool. This is worth flagging: any pitch that frames the firefly route as “doesn't depend on the phenylpropanoid pool” is wrong, because it does. The honest framing is: the firefly route depends on a different part of the phenylpropanoid pool, and the specific bottleneck might be different (cysteine availability or hydroquinone availability rather than caffeic acid availability), but the structural problem of “luminescence flux vs lignin flux” is shared.

Bottom line for the project: This review is the elephant-in-the-room context citation for any pitch, grant proposal, or paper introduction. Three concrete uses for the bibliography. First, it is the cleanest demonstration that the fungal pathway has continued to advance significantly past Mitiouchkina 2020, meaning the brightness benchmark to compete with is not Mitiouchkina's original 2020 results but Zheng 2023's 3 × 10¹¹ photons/min/cm² eFBP plus Shakhova 2024's protein-engineering enhancements on top of that. Any honest framing of the project's competitive position has to acknowledge this. Second, the caffeic acid / phenylpropanoid competition lesson is a generalizable warning, the firefly route's analogous bottleneck is most likely either L-cysteine availability or hydroquinone availability, and Phase 2 troubleshooting should include flux-monitoring of the relevant pools rather than assuming they are unlimiting. The native plant 4CL upregulation noted in Wang 2025 P. pectoralis lantern transcriptomics, combined with this review's emphasis on phenylpropanoid pool competition, suggests that N. tabacum native phenylpropanoid metabolism is going to be a relevant variable in Phase 2 even if the engineering doesn't directly target it. Third, the stress-responsive promoter / biosensor application framing is a useful template, if the firefly platform achieves autonomous bioluminescence in N. tabacum, the same biosensor applications (stress-responsive promoters driving luc2+SKL or driving the BGL/PPYR/ACOT genes individually for kinetic-readout applications) translate over directly. This means the project's commercial space is not just “bioluminescent plant as decoration” but “bioluminescent plant as sensor platform,” with the firefly route potentially having advantages around dynamic range and decoupling between substrate availability (controlled by Phase 1 to 2 genes) and signal generation (controlled by luc2). Cite Yu/Du 2025 as the field-context citation for any document that needs to position the project relative to the fungal-pathway state-of-the-art, alongside Mitiouchkina 2020 as the historical anchor and Zheng 2023 as the brightness benchmark.