Beet red food colourant can be produced more sustainably with engineered *Yarrowia lipolytica*

Colour strongly influences consumer perception of foods, driving widespread use of colourants in processed products. Rising concerns about synthetic additives have increased demand for natural pigments such as betanin (E162) from red beet. However, beetroot contains only ~20–210 mg betanin per 100 g fresh weight on average, extraction yields are modest (≈60% recovery), and expanding production by agriculture would increase land and resource use. Prior engineering in crops (tomato, rice, Nicotiana) improved pigment levels but remains land-dependent and seasonal. Microbial fermentation offers a potentially more sustainable and cost-effective alternative. Previous work in Saccharomyces cerevisiae achieved only tens of mg l−1 betanin, insufficient for competitiveness. This study aims to engineer Yarrowia lipolytica—a robust, Crabtree-negative industrial host with demonstrated high shikimate-pathway flux—to produce betanin at high titres and to evaluate environmental and economic performance versus plant extraction using life cycle and techno-economic assessments.

Plant systems have been engineered for betanin production (e.g., Nicotiana benthamiana, tomato, rice) achieving up to ~270 mg betanin per 100 g fresh weight, exceeding typical beet content yet still constrained by land, climate and seasonality. In microbes, S. cerevisiae engineering progressed from ~17 mg l−1 to ~31 mg l−1 betanin, far below industrially relevant levels. Y. lipolytica has previously yielded high titres of shikimate-derived products (e.g., resveratrol), indicating potential for strong flux to aromatic amino acid derivatives such as l-tyrosine, the precursor to betalains. Regulatory and market pressures increasingly favour natural colours, with the natural colours market estimated at US$1.5–1.75 billion (2022). These studies motivate a microbial, shikimate-pathway-based route in a suitable industrial host.

Biological engineering: The betalain pathway from l-tyrosine was reconstituted in Yarrowia lipolytica. Enzyme variants were selected from prior high-throughput screening in S. cerevisiae. DOPA-4,5-extradiol dioxygenases (DOD) and tyrosine hydroxylases (TYH, CYP76AD family) tested included MjDOD (Mirabilis jalapa), BgDOD2 (Bougainvillea glabra), and TYHs from Beta vulgaris (including an engineered W13L variant), Ercilla volubilis (EvTYH), and Abronia nealleyi. Combinatorial expression identified MjDOD + EvTYH as a superior pair for betanidin formation. To stabilize betanin, glycosyltransferases were tested: BuSGT2 (B. vulgaris scopoletin GT) outperformed DbBSGT (Dorotheanthus bellidiformis), yielding ~9.9 mg l−1 betacyanins (8.6 mg l−1 betanin; 1.3 mg l−1 isobetanin). Localization showed ~40% intracellular and ~60% extracellular betanin; isobetanin mainly extracellular. Flux enhancement: Feedback-insensitive alleles YlARO4K221L and YlARO7G1415 were expressed to improve l-tyrosine supply, shifting accumulation from anthranilate-betaxanthins to betanidin and increasing betacyanins to ~21.7 mg l−1 (17.7 mg l−1 betanin; 4.0 mg l−1 isobetanin). Copy-number optimization of the heterologous pathway (up to three copies) and testing native shikimate-pathway gene amplifications (YlARO1–3) were performed; a third pathway copy improved titres by ~33%. Supplementing l-tyrosine increased titres but with poor conversion (2 g l−1 needed to double production), suggesting upstream bottlenecks. By-product pathway elimination: Media pH and buffering influenced browning and betalamic acid. To prevent pyomelanin formation via homogentisic acid (HGA), Yl4HPPD was deleted, which increased betacyanin ~2.1-fold to 118.1 mg l−1 betanin and 17.4 mg l−1 isobetanin despite persistent browning in buffered media, hypothesized to result from eumelanin formation via cyclo-DOPA/L-dopachrome decomposition at pH >4. Deglycosylation control: Betanin degradation post-peak suggested native β-glucosidases were active. A screen of 14 candidates identified YALI1_B18845g and YALI1_B18887g as key; their disruption in the production strain (ST14284) reduced betanin deglycosylation and increased retention 3.6-fold at 66 h. Fermentations: Controlled fed-batch bioreactor cultivations (AMBR250) used mineral media with glucose (pH 4 or 6) or glycerol (pH 6). Exponential feeds were initiated pre-carbon depletion; samples were frozen automatically. Betacyanins were quantified primarily in total (whole-cell) samples. Analytics: Betanin and isobetanin were quantified by HPLC-UV at 540 nm using a C18 column and solvent gradients; identification referenced commercial beet extract standards and Beer–Lambert calculations. LC–MS/MS was used for untargeted identification of intermediates (anthranilate-betaxanthin, betanidin). UV–vis spectra and fluorescence were used for screening. Sustainability assessment: Life cycle assessment (LCA) used SimaPro with Ecoinvent 3.8 and ReCiPe 2016 (H perspective). System boundary was cradle-to-gate; functional unit, 1 kg E162 colourant (0.4% betanin). Scenarios included feedstocks (glucose, glycerol, sucrose, molasses), varying titres (50% to 400% of base), and plant locations (Germany, Brazil, China). Hotspot and sensitivity analyses were performed; uncertainty via Monte Carlo (1,000 iterations). Techno-economic assessment (TEA): Process simulated in SuperPro Design; upstream fermentation and downstream (cell removal, evaporation, drying with maltodextrin to 0.4% betanin). Baseline constraints: 3-year payback, capacity ≤755 t yr−1, selling price US$34.75 kg−1 E162. Financial metrics (IRR, ROI) and cost breakdowns were computed; sensitivity to titre, capacity, feedstock and selling price assessed. TEA accuracy anticipated ±30–50%.

• Enzyme and pathway engineering in Y. lipolytica enabled gram-per-litre betanin production. The best combination MjDOD + EvTYH with BuSGT2 produced initial ~9.9 mg l−1 betacyanins; adding YlARO4K221L and YlARO7G1415 increased to ~21.7 mg l−1; increasing pathway copy number and further engineering raised titres substantially.
• Eliminating HGA formation by deleting Yl4HPPD increased betacyanin 2.1-fold (to 118.1 mg l−1 betanin, 17.4 mg l−1 isobetanin), outperforming an additional pathway-copy integration (only +19%).
• Fed-batch results: with glucose at pH 6, achieved 1,197 ± 42 mg l−1 betacyanin in 48 h (1,168 mg l−1 betanin; 29 mg l−1 isobetanin). At pH 4, 734 ± 22 mg l−1 betacyanin (718 mg l−1 betanin; 15 mg l−1 isobetanin) in 48 h, with reduced betaxanthins/betalamic acid. With glycerol at pH 6, 911 ± 31 mg l−1 betacyanin (872 mg l−1 betanin; 39 mg l−1 isobetanin) at 36 h.
• Deleting two β-glucosidases (YALI1_B18845g, YALI1_B18887g) improved betanin retention 3.6-fold at 66 h and enabled a record titre: 1,326 ± 148 mg l−1 betacyanin in 51 h (1,271 mg l−1 betanin; 55 mg l−1 isobetanin).
• LCA: Fermentation-based betanin exhibits substantially lower environmental impacts than extraction across most midpoint categories (notably terrestrial ecotoxicity, land use, ionizing radiation, human carcinogenic toxicity, climate change, fossil resource scarcity). Traditional extraction impacts are ~5× higher for human health and ~3× higher for ecosystem quality and resources at endpoints. Process-material inputs dominate impacts; utilities drive climate impact; glycerol scenarios show higher land/ecotoxicity due to upstream rapeseed oil.
• Sensitivity: Increasing fermentation titre markedly reduces impacts; glucose feedstock performs best at higher titres (e.g., ~4.4 g l−1).
• TEA: At commercial scale, production of E162 colourant via fermentation is economically feasible under constraints. Estimated costs: glycerol US$20.62 kg−1 (capacity 688 t yr−1; IRR ~50%); glucose US$21.08 kg−1 (capacity 550 t yr−1; IRR ~50%); molasses US$23.12 kg−1; sucrose US$23.22 kg−1. Doubling titre reduces cost by ~22%; tripling by ~34.5%. Minimum viable selling price ≈ US$21.5 kg−1 formulated product, depending on capacity and market.

The study demonstrates that Y. lipolytica can be engineered to produce betanin at industrially relevant titres and productivities, addressing the need for a sustainable, natural red food colourant alternative to plant extraction. Key strategies—optimal enzyme selection and copy number, enhanced l-tyrosine supply, elimination of competing flux to HGA, and suppression of betanin deglycosylation—collectively increased titres by orders of magnitude over previous yeast efforts. Fermentation pH and feedstock affected both product formation and stability, consistent with betalain and cyclo-DOPA chemistry. LCA confirms a superior environmental profile of fermentation versus extraction even at early-stage performance, and TEA indicates that current titres can already be cost-competitive, with further titre and productivity gains improving both sustainability and economics. Trade-offs between feedstock choices (e.g., glycerol better economically, glucose better environmentally) suggest optimization should prioritize titre and process efficiency while considering regional supply chains and utilities.

Engineering Y. lipolytica enabled gram-scale, fed-batch production of betanin/isobetanin, achieving up to 1.326 g l−1 betacyanins (1.271 g l−1 betanin) within ~51 h—about a 42-fold improvement over previous S. cerevisiae reports. LCA shows fermentation dramatically lowers environmental impacts relative to beet extraction, and TEA indicates economic feasibility at modeled scales and prices, with strong sensitivity to titre and capacity. The work establishes a sustainable, scalable route for natural red colourant production and provides a framework integrating metabolic engineering with LCA/TEA-guided process design. Future research should: (i) further increase titres and productivities; (ii) minimize by-products (betaxanthins, betalamic acid) and pH-dependent degradation; (iii) refine downstream processing tailored to microbial betanin; and (iv) expand to diversified betalain structures (acylated or further glycosylated variants) to deliver improved hues and stability.

• Process maturity: The fermentation and downstream schemes are early-stage; downstream was modeled based on extraction processes, and overall TEA accuracy is preliminary (±30–50%).
• Product stability and by-products: Betanin degradation post-peak and formation of betaxanthins/betalamic acid and browning/eumelanin at pH >4 indicate unresolved stability and side-reaction issues.
• Biological constraints: l-tyrosine supplementation showed poor conversion; precise flux control in the shikimate/tyrosine pathway remains challenging. Reported YlARO7 mutation notation varies across sections.
• Analytics: Lack of commercially available pure betanin standard required indirect quantification (using beet extract and extinction coefficients).
• LCA/TEA assumptions: Results depend on feedstock sourcing, regional energy mixes, market sizes, prices, and scaling rules; system boundaries are cradle-to-gate and exclude end-of-life. Hotspots indicate materials and utilities as major contributors subject to optimization.
• Experimental scale: Many experiments were small-scale (n=2–3) and bioreactor runs in duplicate; broader scale-up validation is needed.