High-yield production of hydroxytyrosol through metabolic engineering of Yarrowia lipolytica

Abstract

Hydroxytyrosol (HT) is a natural antioxidant widely employed in the food production, pharmaceutical, and cosmetic industries. Microbial fermentation for HT production holds significant potential for industrial applications. In this study, we constructed and optimized a biosynthetic pathway for HT in Yarrowia lipolytica for the first time. Initially, HT biosynthesis was achieved by introducing hpaBC, along with ARO10 and ADH6 from Saccharomyces cerevisiae. By employing strategies such as relieving feedback inhibition in the shikimate pathway, optimizing cofactor supply, weakening competing pathways, and balancing metabolic flux, the HT titer was elevated to 2438.27 mg/L. Finally, the engineered strain HT43 produced 11.9 g/L of HT in a 1.3 L bioreactor, which was 917.5 times higher than the initial level. This work sets a new record for the de novo production of HT in a microbial host, achieving the highest titer reported to date, and it offers a novel fermentation strategy with potential for industrial HT production.

1. Introduction

Hydroxytyrosol (HT) mainly exists in the form of esters in the fruits and leaves of olive trees and is a potent natural antioxidant [[1], [2], [3], [4]]. HT has been designated as Generally Recognized as Safe (GRAS) by the U.S. Food and Drug Administration (FDA), thereby qualifying it as a novel food ingredient suitable for incorporation into a variety of food products, including fish oil, vegetable oil, margarine, and beverages [1,5]. In addition to its potent antioxidant capacity, HT possesses diverse pharmacological activities such as anti-aging, neuroprotection, and amelioration of metabolic syndrome [[6], [7], [8]]. Consequently, it has been widely used in the fields of functional foods, pharmaceuticals and cosmetics.

With the growing international consumer demand for beneficial and natural ingredients, the worldwide market size of HT reached 7.51 billion US dollars in 2024 and is expected to grow to 14.25 billion US dollars by 2034 [9], demonstrating strong market potential. The current industrial supply of HT relies on both chemical synthesis and extraction of olive leaves [[10], [11], [12]]. However, plant extraction is constrained by raw material limitations, while chemical synthesis faces challenges such as strict reaction conditions and poor environmental sustainability [13,14]. With the development of green biomanufacturing and the enhancement of metabolic strategies, engineered microorganisms are being positioned as a viable route for HT biosynthesis [1,15,16]. This approach offers advantages including shortened synthesis cycles, freedom from geographical and seasonal constraints, targetable improvement of yield and purity, and alignment with the trend of green biomanufacturing.

Currently, microbial platforms for the direct production of HT have been successfully established in Escherichia coli (9.87 g/L) [1], Saccharomyces cerevisiae (6.97 g/L) [17], and Bacillus licheniformis (9.48 g/L) [2]. However, existing engineered strains still face the issue of low product titers, which hinders their industrial-scale production. To enhance economic viability, it is imperative to overcome key rate-limiting steps through chassis cell optimization and metabolic engineering, thereby achieving efficient redirection of carbon flux toward the HT biosynthetic pathway.

Yarrowia lipolytica, a prototypical unconventional oil-producing yeast, has demonstrated significant advantages in biomanufacturing due to its well-established genetic toolkit and robust tolerance to solvent stress [[18], [19], [20]]. Notably, this yeast has been successfully engineered for high-yield production of various valuable terpenoids and phenolic compounds [4,21], including gastrodin (13.22 g/L) [22], resveratrol (22.5 g/L) [23] and naringenin (8.65 g/L) [24]. These successful cases share the common feature of utilizing the shikimate pathway for target compound biosynthesis, which is also essential for HT production. Therefore, leveraging these shared metabolic features, Y. lipolytica holds great promise as an ideal host for reconstructing high-yield HT biosynthesis systems.

In this work, Y. lipolytica was engineered for high-yield HT production via a multi-level engineering approach (Fig. 1). Initially, we established the HT biosynthetic pathway by introducing HpaBC and the Ehrlich pathway enzymes from S. cerevisiae (ScARO10 and ScADH6). Subsequently, by relieving feedback inhibition of the shikimate pathway, enhancing cofactor recycling and blocking competitive pathways, we significantly improved the HT titer to 2438.27 mg/L. The final engineered strain HT43, achieved an impressive HT titer of 11.9 g/L under fed-batch cultivation conditions in a 1.3 L fermenter, marking the highest titer reported to date. Collectively, this work establishes Y. lipolytica as a viable chassis for HT biosynthesis and provides a foundational framework for future industrial-scale implementation.

Fig. 1.

Schematic representation of the biosynthetic pathway for HT in Y. lipolytica. E4P: erythrose 4-phosphate, PEP: phosphoenolpyruvate, DAHP: 3-deoxy-D-arabino-heptulosonate-7-phosphate, 4-HPP: 4-hydroxyphenylpyruvate, 4HPAA: 4-hydroxyphenylacetaldehyde, HT: hydroxytyrosol, TCA: tricarboxylic acid cycle, PAH: phenylacetaldehyde, 2-PE: 2-phenylethanol, YlPYK: pyruvate kinase, YlARO3: pentafunctional, YlARO4: 3-deoxy-7-phosphoheptulonate synthase, YlARO7: chorismate mutase, YlTRP: anthranilate synthase, YlTYR1: prephenate dehydrogenase, YlPHA2: prephenate dehydratase, YlHPD: 4-hydroxyphenylpyruvate dioxygenase, YlARO10: phenylpyruvate decarboxylase, YlPAR4: phenylacetaldehyde reductases; ScARO10: phenylpyruvate decarboxylase, ScADH6: alcohol dehydrogenase, HpaBC: 4-hydroxyphenylacetate 3-monooxygenase, YlARO1: pentafunctional aromatic synthase, YlARO2: bifunctional chorismate synthase, YlALD2,3: aldehyde dehydrogenase, YlARO8: l-phenylalanine transaminases, YlARO9: aromatic amino acid aminotransferase.

2. Results

2.1. Construction of HT synthesis route in Y. lipolytica

The biosynthesis of HT from tyrosol was achieved through a single hydroxylation step catalyzed by HpaBC (4-hydroxyphenylacetate 3-hydroxylase) [2,17]. Previous studies have screened highly efficient HpaB/HpaC catalytic pairs, including PaHpaB from Pseudomonas aeruginosa and EcHpaC from Escherichia coli [17]. In this work, the genes PaHpaB and EcHpaC were first integrated into the genome of the W29 to obtain strain HT1 (Fig. 2A). In the fermentation broth of HT1 strain fed with 2 g/L tyrosol, a characteristic peak was detected at 5.7 min (Fig. 2B), with mass spectrum data consistent with the HT standard (Fig. S1), corresponding to a quantified titer of 1.92 g/L (Fig. 2C), which confirms functional expression of the heterologous hydroxylase module in Y. lipolytica. However, no new chromatographic peaks were observed in the control group without tyrosol supplementation (Fig. 2B). Given that the Ehrlich pathway from S. cerevisiae (ScARO10 and ScADH6) has been widely employed for efficient synthesis of tyrosol and HT in other microbial hosts [16,25,26], we introduced this heterologous Ehrlich pathway into Y. lipolytica. The genes ScARO10 and ScADH6 were integrated into the genome of HT1 to obtain strain HT2. As expected, strain HT2 produced HT after 96 h fermentation in a shake flask, but at a mere 12.97 mg/L (Fig. 2C).

Fig. 2.

Construction of HT synthesis route in Y. lipolytica. (A) Schematic diagram of strain HT1. (B) HPLC chromatograms of HT and tyrosol. (C) HT production of strains HT1 in shake flasks supplemented with 2 g/L tyrosol and de novo biosynthesis of HT by strain HT2. (D) HT production of strains HT1 in shake flasks supplemented with varying concentrations of tyrosol.

Subsequently, the ability of the engineered strain HT1 to produce HT was investigated by feeding varying concentrations of tyrosol. The highest HT titer achieved was 9.56 g/L with the addition of 10 g/L tyrosol (Fig. 2D). These results suggest that the bottlenecks lie in the upstream pathway. Concurrently, observable growth suppression was monitored by OD₆₀₀ (Fig. 2D), suggesting the presence of cytotoxic effects linked to elevated HT concentrations [2].

2.2. Enhancing HT production by relieving feedback inhibition and improving cofactor regeneration

Previous studies have demonstrated that the mutant allele YlARO3^K225L was used to alleviate the feedback inhibition of l-phenylalanine, while YlARO4^K221L and YlARO7^G139S effectively alleviate the feedback inhibition of l-tyrosine on DAHP synthase [23,24,27]. To mitigate feedback inhibition in the shikimate pathway and increase HT production, we engineered the metabolic pathway (Fig. 3A). Copies of YlARO3^K225L, YlARO4^K221L, and YlARO7^G139S were integrated independently or in combination into HT2 to obtain strain HT3-HT8. As shown in Fig. 3B, the HT titer significantly increased when any one of the three alleles was expressed. Moreover, when the combination of YlARO3^K225L, YlARO4^K221L and YlARO7^G139S was expressed in strain HT8, the HT titer reached 388.82 mg/L (Fig. 3B). Modifying the shikimate pathway resulted in a 30-fold increase in HT titer compared to the initial strain HT2. Thus, increasing shikimate-pathway flux substantially improved HT production.

Fig. 3.

Increasing HT production by relieving feedback inhibition and enhancing cofactor regeneration. (A) Schematic diagram of the shikimate module and the HT biosynthetic pathway. (B) HT production of strains HT2-8 in shake flasks. (C) Schematic diagram of the cofactor module and the HT biosynthetic pathway. (D) HT production of strains HT8-13 in shake flasks. PmLAAD: l-amino acid deaminase, EcTyrA: chorismate mutase/prephenate dehydrogenase.

A recent study successfully achieved the de novo biosynthesis of HT in E.coli using a temperature-inducible system. Researchers tested various phenolic acid decarboxylases and alcohol reductases from different sources and identified YlARO10 and YlPAR4 from Y. lipolytica as the most efficient catalysts [4]. Subsequently, we compared the performance of the ScARO10 and ScADH6 with the endogenous enzymes YlARO10 and YlPAR4 in a shikimate pathway-engineered strain S01, which carries YlARO3^K225L, YlARO4^K221L, and YlARO7^G139S. Results showed that ScARO10 and ScADH6 demonstrated superior catalytic activity in this context, and phenylpyruvate decarboxylase was identified as the major rate-limiting step in the pathway (Fig. S2).

HpaB catalyzes substrate oxidation by utilizing FADH₂, establishing a strict cofactor dependency [28]. Additionally, cytosolic NADH is consumed by HpaBC and ScARO10 during the conversion of tyrosol to HT [17] (Fig. 3C). Building on this principle, our metabolic engineering strategy employed PmLAAD from Proteus mirabilis [29], a validated FADH₂-regenerating enzyme that was previously shown to increase HT titer in E. coli could be increased from 1.52 g/L to 3.12 g/L, which increased by 105.26 % [1]. In addition, we introduced EcTyrA^M53I,A354V to enhance the supply of cytosolic NADH [17] (Fig. 3C). In this separate experiment, overexpression of PmLAAD in strain HT9 resulted in 424.69 mg/L HT, an increase of only 9.22 % compared to the strain HT8 (Fig. 3D). Overexpression of the EcTyrA^M53I,A354V mutations in strain HT10 significantly enhanced HT production to 884.22 mg/L. Notably, co-expressing of PmLAAD and EcTyrA^M53I,A354V in HT13 increased HT to 1013.88 mg/L, representing a 160.75 % increase relative to HT8 (Fig. 3D). These results indicate that enhancing cofactor supply effectively augments HT biosynthesis in Y. lipolytica. Additionally, we overexpressed the tyrosol synthesis pathway genes (ScARO10 and ScADH6) in strains HT11 and HT12, but this did not increase HT production. Therefore, we speculate that the rate-limiting step is due to insufficient upstream flux.

2.3. Deletion of competing pathways to further enhance HT production

Chorismate serves as a key central branch-point precursor in the biosynthesis of aromatic amino acids and derived compounds. During HT production, the biosynthetic pathways for l-phenylalanine and l-tryptophan compete with the HT pathway for this common precursor. Previous research has employed knockout of the tryptophan synthesis genes (YlTRP1/2/3) and the aromatic aminotransferases genes (YlARO8/9) to reduce carbon flux diversion toward l-tryptophan and l-phenylalanine, respectively [22,30]. Similarly, the knockout of the aldehyde dehydrogenase genes (YlALD2/3) has been used to inhibit the oxidation of phenylacetaldehyde to phenylacetate, while the knockout of the 4-hydroxyphenylpyruvate dioxygenase gene (YlHPD) can prevent the conversion of phenylpyruvate to 2-hydroxyphenylacetate [31]. Furthermore, the biosynthetic branch of 2-phenylethanol (2-PE), mediated by YlPHA2, YlARO10, and YlPAR4, represents another shikimate-derived pathway that competes for carbon flux [22]. Separately, knocking out the pyruvate kinase gene (YlPYK) can expand the phosphoenolpyruvate (PEP) pool, a strategy previously shown to boost precursor supply for aromatic compound synthesis [32]. To redirect metabolic flux toward HT synthesis, we targeted and knocked out key genes involved in these competing pathways (Fig. 4A).

Fig. 4.

Deletion of competitive pathways to further enhance HT production. (A) Schematic diagram of the competing pathway module and the HT biosynthetic pathway. (B) HT production of strains HT13-25 in shake flasks.

However, knockout of YlTRP1/2/3 or YlARO8/9 did not improve HT production and slightly decreased titers, likely because these perturbations affected chorismate allocation and cellular fitness. Deletion of YlPHA2 or YlPYK impaired growth (strains HT19 and HT25), consistent with reduced carbon flux into central metabolism in the YlPYK-deficient background [33] (Fig. 4B). In contrast, deletion of YlARO10, YlPAR4, YlALD2, YlALD3, or YlHPD increased HT titers. Among these targets, YlHPD deletion gave the best-performing strain (HT24), reaching 1481.39 mg/L (a 146.11 % increase). Therefore, careful selection of competing-pathway targets is critical for improving HT yield while avoiding severe growth defects. Based on these results, HT24 was used for subsequent engineering.

2.4. Enhancing the precursor supply improves HT yield

Eliminating competing pathways significantly boosted HT production. However, strategies that solely depend on enhancing endogenous pathway flux exhibit inherent limitations [33,34]. Given the inherent complexity and length of the shikimate pathway, we suspect that implementing targeted metabolic engineering approaches to amplify flux through this route could substantially improve HT biosynthesis efficiency.

By overexpressing the genes YlARO1 (encoding pentafunctional aromatic protein), YlARO2 (encoding bifunctional chorismate synthase) and DAHP synthase (YlARO5, YlDHS1, YlDHS2 and YlDHS3) in strain HT24 (Fig. 5A), we generated strains HT26-HT33. As shown in Fig. 5B, Overexpression of YlARO1, YlARO2, or YlARO5 alone did not significantly affect HT titer, whereas co-overexpression of YlDHS1, YlDHS2 and YlDHS3 consistently increased HT production across all tested combinations. The top strain from this effort HT33 produced 1683.33 mg/L of HT.

Fig. 5.

Systematic engineering of the shikimate and downstream pathways for enhanced HT Biosynthesis. (A) Schematic diagram of the competing pathway module and the HT biosynthetic pathway. (B) HT production of strains HT26-33 in shake flasks. (C) HT production of strains HT34-36 in shake flasks. (D) HT production of strains HT37-43 in shake flasks. YlARO1: pentafunctional aromatic synthase, YlARO2: bifunctional chorismate synthase, YlARO5: DAHP synthase, YlDHS1,2,3: 3-deoxy-7-phosphoheptulonate synthase, PcAAS: aromatic aldehyde synthase, EcAROG: 3-deoxy-D-arabino-heptulonate-7-phosphate synthase.

In E. coli, EcAroG accounts for about 80 % of the total DAHP synthase activity, and its feedback-resistant mutant EcAroG^G146N is often expressed to boost the production of target compounds [29,31]. In plants, tyrosine is converted into 4-hydroxyphenylacetaldehyde (4-HPAA) by aromatic aldehyde synthase (AAS), and 4-HPAA is subsequently reduced to tyrosol [35,36]. Based on this pathway, we introduced PcAAS and EcAroG^G146N into the HT33 strain, generating engineered strains HT34-HT36. As shown in Fig. 5C, the expression of PcAAS alone in HT34 enabled HT production to reach 1759.71 mg/L, indicating that this enzyme exhibits good catalytic functionality in the host. However, co-expression of EcAroG^G146N did not lead to a further increase in HT synthesis, suggesting that the metabolic flux in the upstream shikimate pathway is no longer the limiting factor under the current engineering background. Instead, the bottleneck likely lies in the downstream reduction step or cofactor availability. Therefore, we focused on enhancing the tyrosol synthesis pathway and supplying the essential cofactor FADH₂.

By co-expressing PmLAAD and EcTyrA^M53I,A354V, the HT titer in strain HT40 was increased to 2071.13 mg/L. A further increase in the copy numbers of these two genes led to HT production of 2438.27 mg/L in the engineered strain HT43 (A total of 3 copies of PmLAAD and EcTyrA^M53I,A354V) (Fig. 5D). These results demonstrate that when sufficient upstream shikimate flux is achieved, the downstream reducing capacity and cofactor supply become the critical constraints for efficient HT biosynthesis.

2.5. Fed-batch production of hydroxytyrosol

To verify the feasibility of industrial production, strain HT43 was employed to produce HT in a 1.3 L fermenter. During the entire cultivation process, the pH was maintained at 5.0, while the stirring speed was dynamically adjusted between 200 and 1200 rpm to keep the dissolved oxygen level at 30 % [18].

Strain HT43 exhibited delayed growth during fed-batch cultivation process, with HT accumulation beginning after 36 h. The biomass peaked at 84 h, reaching an OD₆₀₀ of 70.8. At 132 h, the HT titer reached 11.9 g/L (Fig. 6), corresponding to a productivity of 0.09 g/L/h. This titer was 4.88-fold higher than that in shake flasks and 917.5 times higher than the initial level. Meanwhile, the residual tyrosol concentration was only 0.99 g/L. To our knowledge, this is among the highest de novo HT titers reported in a microbial host.

Fig. 6.

Fed-batch production of HT in a 1.3 L bioreactor.

3. Discussion

HT holds significant market potential in the food, pharmaceutical, and cosmetic industries, attributed to its potent antioxidant capacity and diverse bioactivities [37,38]. Although microbial production of HT has been explored in several chassis organisms, achieving titers sufficient for industrial-scale implementation remains a challenge [39]. In this study, we successfully established Y. lipolytica as a highly efficient microbial cell factory for the de novo biosynthesis of HT. The final engineered strain HT43 produced 11.9 g/L in a fed-batch bioreactor, which represents the highest titer reported to date.

Despite reports indicating that YlARO10 and YlPAR4 from Y. lipolytica showed the highest in vivo catalytic efficiency when screened in E. coli [4], our findings revealed that the ScARO10 and ScADH6 pair from S. cerevisiae produced more HT in the Y. lipolytica system than the YlARO10 and YlPAR4 combination (Fig. S2). This discrepancy can be attributed to the influence of different host backgrounds and expression conditions on gene functional efficacy [17,40]. A critical factor is the temperature-inducible system employed in the previous study, where high-level expression of heterologous enzymes in E. coli was triggered at 37 °C, a temperature exceeding the optimal growth range for yeast. Conversely, ScARO10 and ScADH6 may be better adapted to the Y. lipolytica system. Moreover, further analysis confirmed that phenylpyruvate decarboxylase (ARO10) serves as the major rate-limiting step in the HT synthesis pathway (Fig. S2).

The shikimate pathway is crucial for the biosynthesis of aromatic compounds [41,42]. Our initial engineering efforts focused on relieving feedback inhibition in the shikimate pathway via expression of feedback-resistant mutants (YlARO3^K225L, YlARO4^K221L, and YlARO7^G139S). This modification enhanced carbon flux toward tyrosine, resulting in a 30-fold increase in HT production (Fig. 3B). Due to the inherent complexity and length of the shikimate pathway, we adopted a stepwise pathway engineering strategy (Fig. 5A). Overexpression of DAHP synthase (YlDHS1, YlDHS2, and YlDHS3) provided an additional enhancement, raising HT production from 1481.39 mg/L in strain HT24 to 1683.33 mg/L in strain HT33 (Fig. 5B). Overall, these results underscore the shikimate pathway as a central target for efficient biosynthesis of HT.

Carbon flux was efficiently redirected toward HT synthesis by eliminating competing pathways [31,32], most notably through the deletion of the competing gene YlHPD. This modification resulted in a 146.11 % increase in HT production while maintaining robust cell growth (Fig. 4B). In contrast, deletions in the phenylalanine branch or PEP consuming pathways led to growth defects, this result consistent with the previous findings [22,32]. emphasizing the importance of careful target selection to balance metabolic flux and cell fitness. A key finding was the critical role of cofactor supply in limiting HT biosynthesis. Increasing the copy numbers of PmLAAD and EcTyrA^M53I/A354V led to a significant enhancement in HT production, achieving a final titer of 2438.27 mg/L (Fig. 5D). This result underscores the necessity of concurrently boosting the regeneration of both FADH₂ and NADH to effectively drive the HpaBC catalyzed hydroxylation and ADH6 mediated reduction steps [1,43].

Notably, the scale-up from shake flasks to a 1.3 L bioreactor in the Y. lipolytica system increased by 388 % (2.44–11.9 g/L), confirming its superior scalability. This improvement significantly surpasses those observed in other hosts, such as a 50 % increase in B. licheniformis and a 216 % increase in E. coli [1,2]. Despite these advancements, there is still substantial potential for further process optimization. Future research should concentrate on refining fermentation parameters and employing adaptive evolution strategies to enhance host tolerance.

In conclusion, this study showcases a systematic multi-modular engineering strategy that includes pathway construction, feedback inhibition relief, cofactor regeneration, competitive pathway knockout, and precursor supply enhancement. This approach not only demonstrates the superior capacity of Y. lipolytica for high-level HT production but also provides a versatile metabolic engineering framework applicable to other value-added phenolic compounds. These findings provide valuable insights for developing sustainable microbial platforms for the industrial production of natural antioxidants.

4. Materials and methods

4.1. Strains, plasmids and genes

The original strain of Y. lipolytica used was the W29 strain, which harbors Cas9 in the KU70 locus [18,44]. E. coli DH5α was used for plasmid construction. The plasmids, strains and primers used in this study are summarized in Supplementary Table S1–S3. To construct integrative strains of Y. lipolytica, a single gRNA plasmid and a linearized homologous donor plasmid were introduced into the cells using the lithium acetate transformation method as described previously [18,45]. The integrative plasmid was constructed using pMD-19T as the backbone with different integration sites, such as IntE3, IntD1, IntC2, IntE1, IntC3, D17, IntE4, IntE14, IntA1, and IntA2 [45]. Promoters (TEFin, GPD), terminators (Lip2, Pex20), and homologous arms were PCR-amplified from the genomic DNA of Y. lipolytica strain W29. Optimal target-specific sgRNA sequences (20 bp) were identified with the help of the online tool “CHOPCHOP” (http://chopchop.cbu.uib.no/) [18]. For gene deletions, DNA fragments consisting of 500–1000 bp up and downstream of the target gene were used as repair template (1000–2000 bp total) [27]. All constructed plasmids were verified by Sanger sequencing (Tsingke Biotechnology Co.).

The following heterologous genes were codon-optimized for Y. lipolytica and synthesized by GenScript (China): PaHpaB (Gene ID: PKG21040.1) from Pseudomonas aeruginosa, EcHpaC (Gene ID: ARI00016.1) from Escherichia coli, ScADH6 (Gene ID: WP_123902793.1) from Saccharomyces cerevisiae, ScARO10 (Gene ID: EF059264.1) from Saccharomyces cerevisiae, PmLAAD (Gene ID: AXQ04983.1) from Proteus mirabilis, EcTyrA^M53I,A354V (Gene ID: WP_103849370.1) from Escherichia coli, EcAROG^D146N (Gene ID: WP_175086260.1) from Escherichia coli, and PcAAS (Gene ID: AAA33860.1) from Petroselinum crispum.

4.2. Cultivation conditions

For cloning and plasmid amplification, E. coli and Y. lipolytica strains were cultured in LB medium (with 100 μg/mL ampicillin) at 37 °C and YPD medium (10 g/L yeast extract, 20 g/L peptone, 20 g/L glucose), respectively.

The yeast transformation was performed using the standard lithium acetate method [45]. Transformants were selected using natMX and hphMX resistance markers. The yeast transformants were plated on YPD agar (YPD medium supplemented with 10 g/L agar) with antibiotics at concentrations of 250 mg/L of nourseothricin and 400 mg/L hygromycin.

4.3. Flask fermentation

Culture was initiated by inoculating a single Y. lipolytica colony into 3 mL of YPD, followed by 24 h growth at 30 °C and 220 rpm. This preculture was transferred to 30 mL of YNB medium (6.7 g/L YNB, 5 g/L (NH₄)₂SO₄, 20 g/L glucose) in a 250 mL shake flask with an initial OD₆₀₀ of 0.1. Fermentation continued for 96 h under identical conditions.

4.4. Analytical methods

Cell density was monitored by measuring the optical density at 600 nm (OD₆₀₀) using a Shimadzu UV2450–Visible Spectrophotometer. To measure dry cell weight (DCW), a 1 mL sample was centrifuged at 12,000 rpm for 10 min to collect cells. The cells were resuspended in 1 mL of deionized water, centrifuged to wash cells. These cells were then dried at 60 °C for 72 h and weighed. To measure glucose, a 1 mL sample was diluted fivefold, and then centrifuged to obtain the supernatant. The glucose was detected by an SBA-40D biosensor from the Institute of Biology (Shandong Province Academy of Sciences, Jinan, China).

For HT and tyrosol analysis, culture broth was centrifuged at 12,000 rpm for 10 min to obtain the supernatant. The supernatant was diluted fivefold with methanol, filtered through a 0.22 μm membrane filter, and then analyzed by high-performance liquid chromatography (HPLC). Quantitative analysis was performed on a SHIMADZU (LC-2050C 3D) instrument equipped with a SHIMSEN SuperbII C18 column (4.6 × 250 mm, 5 μm) at 35 °C, with a 10 μL injection volume and a 1 mL/min flow rate. The separation was achieved using a gradient of 0.1 % formic acid in water (A) and acetonitrile (B) as follows: 10–40 % B (0–5 min), 40–60 % B (5–8 min), 60–100 % B (8–13 min), 100 % B (13–15 min), 100-10 % B (15–20 min), and 10 % B (20–25 min). The analyte was detected at 274 nm. Under these conditions, the samples were analyzed alongside a standard for quantification. Peaks corresponding to the target compounds were identified by comparison to prepared standards (Shanghai yuanye Bio-Technology Co., Ltd). Peak areas were used for compound quantification using external standard calibration method, R² value for the standard curve was >0.999.

Sample were further analyzed using Agilent 1290 UPLC/6540 Q-TOF LC/MS. The MS conditions were as follows: electrospray ionization in negative ion mode, voltage: 3500 V, fragmentation voltage: 135 V, taper hole voltage: 60 V, radio frequency voltage: 700 V, Scanning range: 100–1000 m/z. The chromatographic conditions were as described above.

4.5. Fed-batch fermentation

Y. lipolytica was first streaked from a glycerol stock onto a YPD agar plate and incubated at 30 °C for 48 h. A single colony was subsequently picked to inoculate 4 mL of minimal salt medium (7.5 g/L (NH₄)₂SO₄, 14.4 g/L KH₂PO₄, 0.5 g/L MgSO₄⋅7H₂O, 20 g/L glucose, 2 mL/L trace elements, and 1 mL/L vitamins), which was shaken at 220 rpm and 30 °C for 48 h. The culture was then transferred to 40 mL of minimal salt medium at an OD₆₀₀ of 0.2 for fermentation for 24 h. Subsequently, the culture was transferred to a 1.3 L bioreactor (T&J Bioscience, Shanghai, China) at an OD₆₀₀ of 0.2. The 1.3 L bioreactor contained 400 mL of minimal salt medium. When the glucose in the bioreactor was depleted, the feed composition was 50 g/L (NH₄)₂SO₄, 30 g/L KH₂PO₄, 5 g/L MgSO₄⋅7H₂O, 20 mL/L trace elements, 10 mL/L vitamins, and 400 g/L glucose. The trace elements and vitamins were prepared as described by Chen et al. [18].

Throughout the process, temperature was maintained at 30 °C and pH automatically controlled at 5.0 with 4 M KOH. Dissolved oxygen (DO) was held at 30 % saturation by varying the agitation from 200 to 1200 rpm, the aeration rate inside the bioreactor was 1.0 vol per volume per minute (vvm). After depletion of the initial glucose, a feeding solution was supplied to maintain the residual glucose concentration at 1 g/L. Samples were taken during the fermentation process to measure the cell density (OD₆₀₀), glucose, HT, and tyrosol.

CRediT authorship contribution statement

Bihuan Chen: Writing – original draft, Formal analysis, Data curation. Yuanyuan Wang: Conceptualization. Danlin Wang: Validation. Yina Wang: Validation, Conceptualization. Xiangyu Liu: Investigation. Guanghui Zhang: Supervision, Project administration. Shengchao Yang: Visualization, Supervision, Funding acquisition.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Data availability

Data will be made available on request.