Shikimate crosstalk enables record-level homogentisic acid production in Yarrowia lipolytica

Abstract

Homogentisic acid (HGA; 2,5-dihydroxyphenylacetic acid) is an aromatic compound with diverse applications in pharmaceuticals, food, and functional materials. Industrial HGA production still relies mainly on multistep chemical synthesis routes, which are associated with safety risks and environmental burdens. Here, we report that truncated HMG-CoA reductase (tHMGR), a key enzyme of the mevalonate pathway, unexpectedly serves as a dominant metabolic driving force that enhances shikimate-derived HGA biosynthesis in Yarrowia lipolytica. Through a targeted single-gene screen of the mevalonate pathway, tHMGR emerged as the determinant of HGA overproduction, and copy-number optimization increased the shake-flask titer 2.65-fold to 1766.43 mg/L. Transcriptomic analysis and metabolic network evaluation revealed that the massive NADPH consumption driven by tHMGR perturbs the intracellular redox balance. This cofactor depletion triggers a compensatory upregulation of l-tryptophan biosynthesis and shikimate-associated genes, uncovering a previously unrecognized mevalonate-shikimate metabolic crosstalk. The generality of this tHMGR-driven effect was further validated using p-coumaric acid as a shikimate-derived reporter. Finally, fed-batch fermentation achieved 33.71 g/L HGA in a 2-L bioreactor, representing the highest titer reported to date, and establishing tHMGR as a powerful metabolic lever for enhancing shikimate-derived aromatic production in Y. lipolytica.

Graphical abstract

1. Introduction

Shikimate-derived products form a rapidly expanding class of high-value aromatics with broad applications and a multibillion-dollar global market [[1], [2], [3], [4]]. Homogentisic acid (HGA; 2,5-dihydroxyphenylacetic acid) is an aromatic compound with diverse applications in pharmaceuticals, food, and functional materials [[5], [6], [7]]. Clinically, HGA is a characteristic metabolite in urine and serum and serves as a diagnostic marker for alkaptonuria [8]. HGA has been studied as a natural antioxidant, with activity comparable to widely used antioxidants such as α-tocopherol, ascorbic acid, resveratrol, and butylated hydroxyanisole (BHA) in food [9,10]. Moreover, HGA can undergo auto oxidation to form pyomelanin, a brown black phenolic polymer with favorable dyeing properties that can be used as a colorant and antifouling agent and HGA derived pigments (HDPs) have been developed as biocompatible intracellular labels [[11], [12], [13]]. Despite these high value applications, industrial HGA production still relies mainly on multistep chemical synthesis routes, which are associated with safety risks and environmental burdens, underscoring the need for greener and more sustainable microbial production strategies [14].

HGA can be synthesized in microorganisms via two primary routes: (i) a shikimate-derived branch in which prephenate is converted to 4-hydroxyphenylpyruvate (4-HPP) and subsequently oxidized to HGA by 4-hydroxyphenylpyruvate dioxygenase (HPD), and (ii) a tyrosine catabolic route in which aromatic aminotransferases (Aro8/Aro9) convert tyrosine to 4-HPP, followed by HPD-catalyzed formation of HGA [15]. To date, HGA production has been demonstrated in engineered E. coli (336 mg/L with l-tyrosine supplementation) and S. cerevisiae (809.4 mg/L in shake flasks), while Y. lipolytica has achieved 2.8 g/L in a 2-L bioreactor through increasing erythrose-4-phosphate (E4P) supply and relieving tyrosine feedback inhibition [11,16,17]. However, further titer improvements have largely depended on iterative, multi-gene shikimate-pathway overexpression coupled with deletion of competing branches, and the reported ceiling remains 2.8 g/L [12]. This bottleneck highlights a critical need for orthogonal, mechanism-driven engineering strategies beyond conventional shikimate-pathway stacking. Accordingly, developing a new reinforcement strategy is critical for achieving high-titer HGA production.

Y. lipolytica is an attractive industrial chassis for HGA production because it combines an inherent capacity to form tyrosine-derived brown pigments with strong process robustness [12,18]. As an unconventional oleaginous yeast, Y. lipolytica can utilize a broad range of low-cost substrates, including waste oils, glycerol, and organic acids, and exhibits relatively weak glucose repression, which supports economical biomanufacturing [19,20]. Moreover, HGA is relatively stable under acidic conditions, and the high acid tolerance of Y. lipolytica (pH 3.0–6.0) provides a favorable production environment while helping to mitigate oxidative polymerization [21].

In this study, Y. lipolytica was engineered for high titer of HGA production, revealing a previously unrecognized coupling between the MVA pathway and shikimate pathway metabolism. By relieving feedback inhibition within the shikimate pathway, the HGA titer initially reached 484.41 mg/L from glucose without tyrosine supplementation. Targeted evaluation of MVA-pathway genes identified truncated HMG-CoA reductase (tHMGR) as a key driver, and increasing tHMGR copy number further elevated the HGA titer to 1766.43 mg/L, increased 2.65-fold compared to the control strain. Transcriptomic analysis indicated that the massive NADPH consumption catalyzed by tHMGR perturbed the intracellular redox balance. This cofactor depletion acts as a metabolic trigger, rewiring aromatic metabolism by driving a compensatory upregulation of l-tryptophan biosynthesis and the shikimate pathway. The novel reinforcement strategy can increase flux through the shikimate pathway and thus has the potential to improve the titers of shikimate-derived products. Consistently, the titer of the shikimate-derived reporter p-coumaric acid increased from 47.74 mg/L to 120.85 mg/L, increased 1.53-fold by this strategy. Ultimately, fed-batch fermentation achieved an HGA titer of 33.71 g/L in a 2-L bioreactor, representing the highest titer reported to date.

2. Materials and methods

2.1. Strain and plasmids construction

Yarrowia lipolytica PO1f (MYA-2613) and strain FY10 were used as hosts for genetic engineering. The FY10 strain, a laboratory collection strain engineered for β-farnesene production, features a reinforced mevalonate (MVA) pathway, including two additional copies of MVA pathway genes and three genomic copies of tHMGR. The strains and plasmids used in this study are listed in the Supplementary Materials (Tables S1 and S2). Escherichia coli Trans10 (TransGen Biotech, China) was used for plasmid propagation. The codon-optimized heterologous genes were synthesized by BGI (China) and are listed in the Supplementary Materials (Tables S3 and S4).

The genomic integration loci utilized in this study, specifically intB1, intD1, and intE1, were selected based on previously reported neutral integration sites in Y. lipolytica [22]. The integration of heterologous gene expression cassettes into these specific loci ensures stable and high-level gene expression without disrupting essential endogenous genes or causing observable growth penalties to the engineered strains.

To construct strains with varying tHMGR copy numbers (from one to four), a stepwise genomic integration strategy targeting the intE1 and intD1 loci was employed. Linearized DNA fragments, harboring either a single tHMGR expression cassette or tandem dual tHMGR cassettes, were amplified from the corresponding constructed plasmids. Sequential transformation of these purified linear fragments enabled the precise genomic assembly of specific tHMGR copy numbers. Detailed genotypes of all resulting strains are provided in Supplementary Table S1.

2.2. Media and cultivation conditions

E. coli was used for plasmid propagation and grown in LB medium (10 g/L tryptone, 5 g/L yeast extract, and 5 g/L NaCl). For solid plates, 20 g/L agar was added. Y. lipolytica was cultivated in YPD medium (20 g/L glucose, 20 g/L peptone, and 10 g/L yeast extract) or synthetic SC medium containing glucose as the carbon source (40 g/L glucose, 5 g/L ammonium sulfate, and 1.7 g/L yeast nitrogen base). Transformants were selected on SC-Ura or SC-Ura-Leu agar plates supplemented with 20 g/L glucose and 1.2 g/L amino acid dropout mixture. Shake-flask cultivations were performed in 100 mL flasks containing 30 mL SC-glucose medium, inoculated to an initial OD₆₀₀ of 0.1, and incubated at 30 °C and 200 rpm for 96 h.

2.3. Fed-batch fermentation

For bioreactor experiments, seed cultures were prepared in SC-glucose medium and grown for 24 h. Cells were harvested, washed, and inoculated into a 2-L bioreactor (xCUBIO twin, bbi-biotech, Germany) to an initial OD₆₀₀ of 0.8–1.0. Fermentations were performed in 800 mL SCY-glucose medium (80 g/L glucose, 10 g/L ammonium sulfate, 3.5 g/L yeast nitrogen base, and 1.5 g/L yeast extract) at 30 °C. Fed-batch feeding was initiated when residual glucose decreased below 30 g/L by supplying an 800 g/L glucose solution supplemented with 30 g/L ammonium sulfate and 1.5 g/L yeast extract. The pH was maintained at 5.0 by automatic addition of 5 M KOH, and agitation was set to 600 rpm.

2.4. Analytical methods

HGA was quantified by HPLC with UV detection at 290 nm using a Kromasil 100-5-C18(W) column (4.6 × 250 mm, 5 μm). The mobile phase consisted of 10 mM KH₂PO₄: Methanol (90:10, v/v), delivered at 0.8 mL/min, with the column maintained at 40 °C.

For p-coumaric acid quantification, 0.1 mL of fermentation broth was mixed with 0.9 mL anhydrous methanol (100%, v/v). After vortexing for 30 s, samples were centrifuged at 12,000 rpm for 10 min, and the supernatant was analyzed by HPLC. The mobile phase was 45% (v/v) methanol in water at a flow rate of 0.6 mL/min, the column temperature was maintained at 40 °C, and UV detection was performed at 304 nm.

The intracellular levels of total NADP(H) and NADPH were quantified using the Enhanced NADP⁺/NADPH Assay Kit with WST-8 (Beyotime Biotechnology, Shanghai, China), strictly following the manufacturer's instructions.

For RNA extraction and transcriptome sequencing, the strain was cultured in SC medium for 72 h. After incubation, the cells were harvested, washed twice with ddH₂O, and then quickly frozen in liquid nitrogen. Total RNA extraction and transcriptome sequencing were subsequently performed (Tsingke Biotechnology, China). Sequencing data analysis was conducted using a cloud-based platform.

2.5. Data analysis

All shake-flask experiments were performed in biological triplicates. Data are expressed as the mean ± standard deviation (SD). Differences between experimental groups and the control were evaluated using a two-tailed Student's t-test. A P-value of less than 0.05 was considered statistically significant.

3. Results

3.1. Chassis evaluation for HGA production under l-tyrosine supplementation

HGA can be natively produced from l-tyrosine in Y. lipolytica [23]. l-tyrosine is converted to 4-hydroxyphenylpyruvate (4-HPP) by aromatic aminotransferases (Aro8/Aro9), and 4-HPP is subsequently oxidized to HGA by 4-hydroxyphenylpyruvate dioxygenase (HPD) (Fig. 1A). HPD homologs are widespread among microorganisms, and the HPD from Pseudomonas putida KT2440 (ppHPD) was selected as a heterologous catalyst to enhance HGA formation [17]. To enhance the conversion of 4-HPP to HGA, ppHPD was integrated into the ku70 locus of the chassis strain Y. lipolytica PO1f (MYA-2613), generating strain XYL01 (Fig. 1B). l-tyrosine supplementation is commonly used to enhance HGA accumulation, as tyrosine biosynthesis in Y. lipolytica is constrained by feedback inhibition [15]. Accordingly, HGA production was evaluated in shake-flask cultures containing 0, 0.5, or 1.0 g/L l-tyrosine to quantify the contribution of heterologous ppHPD.

Fig. 1.

(A) l-tyrosine–derived biosynthesis of homogentisic acid (HGA) in Y. lipolytica. L-Tyr, l-tyrosine; 4-HPP, 4-hydroxyphenylpyruvate; HGA, homogentisic acid; ARO8, aromatic aminotransferase; ARO9, aromatic aminotransferase. (B) Construction of strains XYL01 and XYL02. The heterologous ppHPD expression cassette was integrated into the ku70 locus to generate XYL01 (based on strain 2613) and XYL02 (based on the MVA-reinforced background FY10). (C) HGA titers of strains 2613 and XYL01 in shake-flask cultures supplemented with different concentrations of l-tyrosine. (D) Growth curves (OD₆₀₀) of strains 2613 and XYL01 under the same l-tyrosine supplementation conditions. (E) HGA titers of strains XYL01 and XYL02 in shake-flask fermentation with 1.0 g/L l-tyrosine. (F) Growth curves (OD₆₀₀) of strains XYL01 and XYL02 in shake-flask fermentation with 1.0 g/L l-tyrosine.

Without l-tyrosine supplementation, HGA was not detected in the parental strain 2613, whereas strain XYL01 reached an HGA titer of 3.62 mg/L. With 0.5 g/L l-tyrosine supplementation, the HGA titer increased to 17.38 mg/L in strain 2613 and 61.43 mg/L in strain XYL01. Increasing l-tyrosine to 1.0 g/L further elevated the HGA titer to 25.77 mg/L in strain 2613 and 101.10 mg/L in strain XYL01 (Fig. 1C). Although 1.0 g/L l-tyrosine slightly reduced OD₆₀₀ relative to the no-supplement condition, it substantially increased the HGA titer, indicating that heterologous ppHPD expression enhances the conversion of l-tyrosine to HGA in Y. lipolytica (Fig. 1D).

HGA is a key precursor for tocotrienol biosynthesis, while the other precursor, geranylgeranyl diphosphate (GGPP), is supplied by the mevalonate (MVA) pathway [24]. FY10, a laboratory collection strain for β-farnesene production, carries reinforced MVA pathway expression, including two additional copies of MVA pathway genes and three genomic copies of tHMGR [25,26]. To investigate whether MVA pathway reinforcement affects HGA formation, ppHPD was integrated into the ku70 locus of FY10 to generate strain XYL02. Using the strain XYL01 as the control, shake-flask cultures was performed with 1.0 g/L l-tyrosine supplementation. The strain XYL02 produced 263.78 mg/L HGA at 72 h, whereas the strain XYL01 reached 113 mg/L at 96 h (Fig. 1E). Thus, the strain XYL02 achieved a 2.33-fold higher HGA titer than the strain XYL01, although its OD₆₀₀ was slightly lower (Fig. 1F).

3.2. Relief of feedback inhibition strengthens the shikimate pathway and enables high HGA production from glucose

Although l-tyrosine supplementation substantially increases HGA titers, its cost limits practical implementation. HGA can be synthesized de novo from glucose by the shikimate pathway in Y. lipolytica, making shikimate pathway reinforcement a practical alternative [12]. Common approaches include strengthening flux through endogenous ylARO1 and ylARO2, as well as expressing specific feedback-insensitive variants such as scARO4^K229L (a feedback-resistant 3-deoxy-7-phosphoheptulonate synthase from S. cerevisiae) and corresponding ylARO7 mutants [1,27,28] (Fig. 2A).

Fig. 2.

Shikimate-pathway reinforcement enables de novo HGA biosynthesis from glucose. (A) Schematic of the de novo HGA biosynthetic route from glucose via the shikimate pathway in Y. lipolytica. ARO4, 3-deoxy-d-arabino-heptulosonate 7-phosphate synthase; ARO1, pentafunctional AROM polypeptide; ARO2, chorismate synthase; ARO7, chorismate mutase; scARO4^K229L, feedback-resistant S. cerevisiae Aro4 variant (K229L); ylARO7^G139S, feedback-resistant Y. lipolytica Aro7 variant (G139S). (B) HGA titers of XYL01-derived strains expressing shikimate-pathway genes or feedback-resistant variants in shake-flask cultures. (C) Growth curves (OD₆₀₀) of the corresponding XYL01-derived strains. (D) HGA titers of XYL02-derived strains expressing the same shikimate-pathway genes or feedback-resistant variants in shake-flask cultures. (E) Growth curves (OD₆₀₀) of the corresponding XYL02-derived strains.

To evaluate key shikimate pathway nodes, ylARO1, ylARO2, scARO4, ylARO7, scARO4^K229L and ylARO7^G139S were individually constructed on the pMO plasmid, and the empty vector was used as a control. Shikimate pathway reinforcement was tested in both chassis because strain XYL01 and strain XYL02 showed markedly different HGA production with the l-tyrosine supplementation. The plasmids were transformed to strain XYL01,generating strains XYL03 to XYL09. The control strain XYL03 with empty plasmid produced 2.32 mg/L HGA, and the strain XYL08 with expression of scARO4^K229L reached the highest titer of 20.64 mg/L in shake-flask cultures. Expression of ylARO7^G139S (strain XYL09) increased HGA titer to 9.58 mg/L (Fig. 2B), and the strain XYL09 showed a slightly reduced OD₆₀₀ (Fig. 2C). The same plasmids were transformed to the strain XYL02, generating strains XYL10 to XYL16. The control strain XYL10 produced 5.79 mg/L HGA, whereas scARO4^K229L expression (strain XYL15) increased the titer of HGA to 247.02 mg/L. Expression of ylARO7^G139S (strain XYL16) increased the titer of HGA to 43.04 mg/L (Fig. 2D), and strain XYL16 showed a slightly reduced OD₆₀₀ (Fig. 2E).

Overall, shikimate pathway reinforcement promoted HGA accumulation in both backgrounds. The superior performance of scARO4^K229L aligns with its biological role as the first rate-limiting enzyme in the pathway; alleviating its feedback inhibition effectively redirects carbon flux from central metabolism into the aromatic network. Furthermore, the substantially higher titers achieved in the XYL02 background highlight a pronounced chassis dependence. Considering that strain XYL02 harbors a pre-reinforced MVA pathway, this synergistic effect suggests a potential metabolic link between the MVA and shikimate pathways, which prompted a deeper investigation into the specific MVA pathway components driving this synergy.

3.3. MVA pathway gene screening identifies tHMGR as a key driver of increased HGA titer

A significant difference in HGA titer was consistently observed between strains XYL01 and XYL02. Strain XYL01 was derived from strain 2613, whereas strain XYL02 was constructed from strain FY10 with the MVA pathway reinforced. To dissect the genetic basis of this difference, the shikimate pathway reinforcement was first implemented in the XYL01 lineage. Preliminary plasmid-based evaluations indicated that co-expressing ylARO1, ylARO2, scARO4^K229L, and ylARO7^G139S exerted a substantial synergistic effect, yielding an HGA titer of 336.99 mg/L. This significantly outperformed the expression of the feedback-resistant scARO4^K229L and ylARO7^G139S alone (106.08 mg/L) or the expression of ylARO1 and ylARO2 alone (3.79 mg/L) (Fig. S1). Based on this rationale, the cassette ylARO1, ylARO2, scARO4^K229L, and ylARO7^G139S was integrated into the chromosomal intB1 locus (a neutral integration site located on chromosome B) to generate strain XYL17, which reached an HGA titer of 484.41 mg/L HGA in shake-flask cultures.

The MVA pathway supplies the universal isoprenoid precursors isopentenyl diphosphate (IPP) and dimethylallyl diphosphate (DMAPP) from acetyl-CoA through a series of conserved enzymatic steps [29]. Acetyl-CoA acetyltransferase (ERG10) and HMG-CoA synthase (ERG13) condense acetyl-CoA to form HMG-CoA, which is subsequently reduced to mevalonate by HMG-CoA reductase (HMGR) [30]. Wild-type HMGR is anchored to the endoplasmic reticulum and is subject to strict native degradation. Therefore, a truncated variant (tHMGR) lacking the N-terminal membrane anchor was employed in this study, as its enhanced cytosolic stability effectively bypasses endogenous regulation and maximizes catalytic efficiency [19,31]. Mevalonate is then sequentially phosphorylated by mevalonate kinase (ERG12) and phosphomevalonate kinase (ERG8), followed by decarboxylation catalyzed by mevalonate diphosphate decarboxylase (ERG19) to generate IPP. IPP is interconverted with DMAPP by isopentenyl diphosphate isomerase (IDI), and geranylgeranyl diphosphate synthase (GGPPS) subsequently condense these C5 units to generate longer-chain prenyl diphosphates [32], thereby supporting the biosynthesis of isoprenoids and related products [33,34] (Fig. 3A).

Fig. 3.

Effects of mevalonate (MVA) pathway genes on HGA biosynthesis. (A) Schematic overview of HGA biosynthesis via the shikimate pathway and the MVA pathway module. Part I, l-tyrosine–to-HGA conversion module; Part II, shikimate pathway module; Part III, MVA pathway module converting acetyl-CoA toward GGPP. ERG8, phosphomevalonate kinase; ERG10, acetyl-CoA acetyltransferase; ERG12, mevalonate kinase; ERG13, HMG-CoA synthase; ERG19, mevalonate diphosphate decarboxylase; IDI, isopentenyl diphosphate isomerase; tHMGR, truncated HMG-CoA reductase; SaGGPPs, geranylgeranyl diphosphate synthase from Sulfolobus acidocaldarius. (B) HGA titers of strains XYL18 to XYL26 generated by individual overexpression of MVA pathway genes. (C) Relative changes in HGA titer (%) for each single-gene overexpression strain compared with the control. (D) Growth curves (OD₆₀₀) of strains XYL18 to XYL26 under the same conditions as in (B). (E) HGA titers of strains overexpressing different genomic copy numbers of tHMGR.

Individual genes from the MVA pathway were evaluated in the strain XYL17 to identify components that positively contribute to HGA formation. Specifically, ERG8, ERG10, ERG12, ERG13, ERG19, saGGPPs, IDI, and tHMGR were cloned on the pMO plasmid and transformed to the strain XYL17, generating strains XYL18 to XYL26, with the empty plasmid strain XYL18 used as the control. The strain XYL18 produced 295.64 mg/L HGA, whereas tHMGR overexpression (strain XYL26) increased the titer of HGA to 411.02 mg/L, corresponding to a 39.03% improvement over the control. In contrast, ERG19 (strain XYL23) and IDI (strain XYL25) showed no appreciable change, yielding 269.71 mg/L and 278.31 mg/L HGA, respectively. Notably, overexpression of ERG12 (strain XYL21), ERG13 (strain XYL22), and saGGPPs (strain XYL24) substantially reduced the titer of HGA to 137.53 mg/L, 95.21 mg/L, and 84.15 mg/L, representing decreases of 53.48%, 67.80%, and 71.5% relative to the control (Fig. 3B and C). This substantial reduction suggests that the overexpression of these specific mevalonate pathway enzymes intensifies the competition for central carbon precursors and ATP, thereby restricting the parallel carbon flux necessary for shikimate and HGA biosynthesis. Across these strains, OD₆₀₀ remained comparable, indicating that the observed effects were not primarily driven by growth defects. The higher HGA titer of strain XYL02 relative to strain XYL01 is likely attributable to the presence of three genomic copies of tHMGR, which exhibited positive effect on HGA titer in the gene screen.

Although three genomic copies of tHMGR in strain XYL02 can increase the HGA titer, the strain also carries genomic overexpression of several MVA pathway genes that negatively affect HGA production. It was therefore hypothesized that overexpressing tHMGR alone may yield a higher HGA titer. Accordingly, different copies of tHMGR were integrated stepwise to generate XYL27 to XYL30. Strains with one to four copies tHMGR produced 813.18 mg/L, 1481.97 mg/L, 1766.43 mg/L, and 1589.59 mg/L HGA, respectively. These results identify tHMGR as a key driver of HGA overproduction and demonstrate that three genomic copies provide the optimal benefit, increasing HGA to 1766.43 mg/L, which corresponds to a 2.65-fold improvement over the control strain XYL17 (Fig. 3E). The subsequent decline in production observed with four genomic copies suggests that exceeding this optimal expression level likely imposes an unsustainable metabolic burden.

To explore the industrial applicability of the engineered strain, its capacity to utilize low-cost carbon sources was evaluated. Strain XYL29 was cultured in media containing glucose, glycerol, or oleic acid as the carbon source (Fig. S2). The strain successfully synthesized HGA from all tested substrates. At 96 h, glucose yielded the highest HGA titer of 1706.90 mg/L. Cultivation on oleic acid and glycerol resulted in HGA titers of 519.73 mg/L and 334.97 mg/L, respectively. The corresponding growth profiles indicated that glucose supported the most rapid growth. Oleic acid induced a slower initial growth phase but eventually supported a high cell density comparable to that of glucose after 72 h, whereas glycerol resulted in the lowest maximum biomass. These results confirm the feasibility of utilizing alternative, low-cost substrates for HGA biomanufacturing using the engineered Y. lipolytica chassis. Furthermore, the HGA titers of strain XYL29 from two independent fermentations using glucose as the carbon source exhibited no significant difference, confirming the robust genetic stability of this chromosomally integrated strain (Fig. S3).

3.4. Transcriptomic analysis reveals that tHMGR enhances HGA titer by upregulating l-tryptophan biosynthesis

tHMGR has been widely used to enhance flux through the MVA pathway by catalyzing the reduction of HMG-CoA to mevalonate, a reaction that consumes two equivalents of NADPH [35]. However, leveraging tHMGR reinforcement to increase the titer of the shikimate pathway product HGA has not been reported previously, to the best of current knowledge. To elucidate the metabolic mechanisms by which tHMGR overexpression improves the HGA titer, transcriptomic analysis was performed using strain XYL17 as the control and strain XYL29 as the experimental group [36]. Volcano plot analysis identified 646 upregulated and 603 downregulated genes (Fig. 4A). KEGG enrichment analysis indicated that pathways involved in amino acid biosynthesis were among the most prominently altered. Among the differentially expressed genes (DEGs), the most statistically significant enrichment was observed for phenylalanine, tyrosine, and tryptophan biosynthesis, followed by lysine biosynthesis; proline, leucine, and isoleucine biosynthesis; and glycine, serine, and threonine metabolism. Expectedly, terpenoid backbone biosynthesis was also significantly enriched among the upregulated pathways, consistent with the role of tHMGR as a key enzyme in the mevalonate pathway (Fig. 4B).

Fig. 4.

Transcriptomic analysis of the tHMGR-reinforced strain reveals activation of aromatic amino acid biosynthesis and the NAD⁺ de novo pathway. (A) Volcano plot of differentially expressed genes (DEGs) between XYL17 and XYL29. (B) KEGG pathway enrichment analysis of DEGs between XYL17 and XYL29. (C) Differentially expressed genes in the phenylalanine, tyrosine, and tryptophan biosynthesis pathways in XYL29 relative to XYL17, with upregulated genes shown in red and downregulated genes shown in gray. (D) Schematic of NAD⁺, de novo biosynthesis from l-tryptophan via the kynurenine pathway. 3-HAA, 3-hydroxyanthranilic acid; ACMS, 2-amino-3-carboxymuconate semialdehyde; NAMN, nicotinic acid mononucleotide; NAAD, nicotinic acid adenine dinucleotide; NAD⁺, nicotinamide adenine dinucleotide. (E) Intracellular NADPH/NADP⁺ ratios of strains XYL17 and XYL29. (F) Fold changes of total NADP(H) and NADPH levels in strain XYL29 relative to strain XYL17.

Biosynthesis of the aromatic amino acids (phenylalanine, tyrosine, and tryptophan) converges at chorismate, which also serves as a key node in the HGA biosynthetic route, suggesting that increased flux through this hub could directly promote HGA formation [37]. The aromatic amino acid network was mapped using transcriptomic data, with upregulated genes shown in red and downregulated genes shown in gray. Most genes involved in phenylalanine, tyrosine, and tryptophan biosynthesis were upregulated, with only YALI2_C01099g downregulated. Notably, YALI2_C01099g and the upregulated YALI2_E01737g are annotated as isozymes (Fig. 4C). Among the upregulated genes, YALI2_C00536g, YALI2_C01057g, YALI2_F00730g, and YALI2_D00775g are involved in channeling phosphoenolpyruvate (PEP) and erythrose 4-phosphate (E4P) toward prephenate formation, a key control region for strengthening the shikimate pathway flux in Y. lipolytica [2]. Together, these transcriptional changes indicate a systems-level metabolic rewiring, where tHMGR overexpression indirectly drives the upregulation of the shikimate derived aromatic network, thereby contributing to the marked increase in HGA titer.

As noted above, tHMGR catalyzes the reduction of HMG-CoA to mevalonate, a reaction that consumes substantial amounts of the cofactor NADPH [38]. While basal intracellular NAD(P)H levels are generally sufficient to sustain native NAD(P)H oxidoreductases, these cofactors can become rapidly depleted when tHMGR is strongly overexpressed in engineered cell factories [39]. This intense metabolic sink perturbs the global intracellular redox balance. To counteract this severe NADPH depletion and restore cofactor homeostasis, the cell must activate metabolic routes that expand the NAD(P)⁺/NAD(P)H pool.

This framework provides a compelling mechanistic explanation for the observed global upregulation of tryptophan biosynthesis and the shikimate pathway. In Y. lipolytica, the kynurenine pathway represents the sole de novo biosynthesis route for nicotinamide adenine dinucleotide (NAD⁺) [40,41]. In this pathway, l-tryptophan is converted through a multi-step enzymatic cascade to quinolinic acid (QA), which is subsequently converted to NAD⁺ (Fig. 4D) [42,43]. This newly synthesized NAD⁺ pool can then be phosphorylated to NADP⁺, which is subsequently reduced to regenerate the essential NADPH [44].

To validate this hypothesis, intracellular cofactor levels of the control strain XYL17 and the tHMGR-overexpressing strain XYL29 were quantified. The NADPH/NADP⁺ ratio in strain XYL29 decreased to 0.28, compared to 0.72 in strain XYL17 (Fig. 4E), confirming that tHMGR overexpression consumes substantial NADPH. Furthermore, while the absolute NADPH level in XYL29 was 0.91-fold that of the control, its total NADP(H) pool expanded to 1.52-fold (Fig. 4F).

Therefore, the tHMGR-induced redox imbalance acts as a metabolic trigger, driving a compensatory upregulation of the l-tryptophan pathway to replenish the exhausted cofactor supply. The cell upregulates the kynurenine pathway, the sole de novo NAD⁺ biosynthesis route from l-tryptophan in Y. lipolytica, to restore the NADP(H) pool. As the upstream shikimate pathway is globally enhanced to meet this heightened tryptophan demand, the highly expressed ppHPD effectively intercepts the amplified chorismate-derived intermediates. This cofactor-driven metabolic pull inadvertently redirects significant carbon flux through the shikimate pathway, thereby fueling HGA overproduction (Fig. 5).

Fig. 5.

Metabolic mechanism diagram illustrating the tHMGR-driven systemic rewiring for enhanced HGA production from glucose. The massive consumption of the cofactor NADPH by overexpressed tHMGR in the mevalonate (MVA) pathway triggers a compensatory upregulation of the de novo NAD⁺ biosynthesis route, systematically pulling the carbon flux toward HGA accumulation. Distinct metabolic modules are color-coded as follows: the glycolysis pathway (gray), the shikimate pathway (orange), the aromatic amino acid (tryptophan, tyrosine, and phenylalanine) biosynthesis pathways (yellow), the kynurenine pathway (blue), and the MVA pathway (green). The tricarboxylic acid (TCA) cycle is depicted within the mitochondrion.

Furthermore, lysine biosynthesis was upregulated, which is consistent with the elevated demand for acetyl-CoA as the precursor for the mevalonate pathway [45]. Alterations in glycine, serine, and threonine metabolism likely reflect systemic adjustments linking amino acid homeostasis to central carbon metabolism [46]. Given that intracellular amino acid pools rely on both de novo synthesis and interconversion, these coordinated responses help balance the precursor and cofactor demands essential for high-level HGA production.

3.5. p-Coumaric acid validates tHMGR-driven enhancement of shikimate pathway products

Shikimate pathway-derived products constitute a large class of high-value aromatics, and extensive efforts have been devoted to improving their biosynthesis in yeast [47,48]. To assess the generality of the tHMGR-driven enhancement strategy, p-coumaric acid was selected as a reporter. As a central phenylpropanoid precursor, it directly reflects the metabolic output of the shikimate and aromatic amino acid pathways [[49], [50], [51]] (Fig. 6A).

Fig. 6.

Validation of tHMGR reinforcement as a general strategy to enhance shikimate-derived production using p-coumaric acid as a reporter. (A) Schematic of the tHMGR-based strategy for p-coumaric acid biosynthesis through the shikimate pathway. rgTAL, tyrosine ammonia-lyase from Rhodotorula glutinis. (B) p-Coumaric acid titers of strains XYL31 to XYL33. (C) Growth curves (OD₆₀₀) of strains XYL31 to XYL33.

A basal p-coumaric acid-producing strain (XYL31) was established by integrating rgTAL (tyrosine ammonia-lyase from Rhodotorula glutinis) into the ku70 locus of strain 2613. To prevent native allosteric regulation from masking the tHMGR-induced metabolic pull, feedback-resistant mutants (scARO4^K229L and ylARO7^G139S) were integrated at the intB1 locus to generate strain XYL32. Subsequently, two copies of tHMGR were integrated into strain XYL32, yielding strain XYL33. Under identical shake-flask conditions, the p-coumaric acid titer increased from 47.74 mg/L in strain XYL32 to 120.85 mg/L in strain XYL33 (Fig. 6B). This represents a 2.53-fold improvement, with no obvious difference in cell growth observed between the two strains (Fig. 6C). These results demonstrate that the strong metabolic pull generated by tHMGR overexpression further enhances shikimate pathway output even after feedback inhibition is alleviated, supporting the general applicability of this cofactor-driven strategy to other shikimate derived products.

3.6. Fed-batch fermentation in a 2-L bioreactor

To further increase HGA production, strain XYL29 was evaluated in fed-batch fermentation in a 2-L bioreactor. Because HGA undergoes accelerated oxidation under alkaline conditions (pH > 7), which promotes pyomelanin formation, the fermentation pH was controlled at 5.0 to suppress oxidation and favor HGA accumulation. During cultivation, organic acids produced by Y. lipolytica led to pH decrease, and pH was maintained automatically by the addition of 5 M KOH.

Fed-batch feeding was initiated when the initial glucose concentration decreased to 30 g/L. Glucose was supplied by a constant-rate feeding strategy to maintain the residual glucose concentration below 30 g/L throughout the feeding phase. Under these conditions, HGA reached a maximum titer of 33.71 g/L at 216 h and remained essentially stable thereafter, measuring 33.36 g/L at 240 h (Fig. 7). The plateau in HGA production after 216 h likely results from physiological constraints typical of late-stage high-density fermentations, such as cellular aging, diminished oxygen transfer efficiency, or the accumulation of inhibitory byproducts. Nevertheless, this final titer represents the highest HGA production reported to date in any microbial host (Table 1).

Fig. 7.

Fed-batch production of HGA in a 2-L bioreactor. HGA titer profile during the fermentation (with glucose feeding initiated when the residual glucose decreased to 30 g/L and maintained below 30 g/L thereafter), with pH controlled at 5.0 by automatic addition of 5 M KOH.

Table 1.

HGA biosynthesis by engineered microorganisms.

Host	Product	Titer	Strategy	Fermentation mode	Reference
E. coli	HGA	336 mg/L	Express the HPD from Pseudomonas putida KT2440 and l-tyrosine was supplemented as a precursor	Shake flasks	[52]
E. coli	HGA	57 mg/L	Integrated HPD from Pseudomonas putida KT2440 and express HPD with a multicopy plasmid	Shake flasks	[16]
Paenibacillus sp. TKU036	HGA	60 mg/L	Squid pen powder (SPP) served as the sole C/N source	Shake flasks	[53]
Y. lipolytica	HGA	2.8 g/L	HGS1 and HPD1 were overexpressed; TKL1 and TAL1 were overexpressed to enhance E4P supply; the feedback-resistant ARO4K221L variant was overexpressed to relieve tyrosine feedback inhibition	2L Fermenter	[11]
S. cerevisiae	HGA	809.4 mg/L	ΔAro10, ΔAro3, and ΔPdc5; Integrated Aro4K229L and Aro7G141S; overexpression of Tal1, Eno2, Aro2, and Tyr1	Shake flasks	[54]
Y. lipolytica	HGA	33.71 g/L	Express the HPD from Pseudomonas putida KT2440; Integrated ylAro1, ylAro2, scAro4K229L and ylAro7G139S; Express three copies of tHMGR by improve NAD(P)H supply to support shikimate-pathway reinforcement.	2L Fermenter	This study

Alongside this record titer, the engineered strain achieved an HGA yield of 0.0885 g/g glucose (representing approximately 15.8% of the maximum theoretical yield of 0.56 g/g) and a volumetric productivity of 0.156 g/(L·h). Overall, this exceptional fermentation performance demonstrates that the combined engineering of the shikimate and mevalonate pathways, coupled with targeted cofactor balancing, can be effectively translated into high-density cultivation systems. Furthermore, the comparison with the theoretical maximum yield highlights both the remarkable efficiency of the current tHMGR-driven strategy and the clear trajectory for future metabolic optimization.

4. Discussion

This work establishes Y.lipolytica as a high-performance chassis for HGA production and identifies an unanticipated metabolic driving force to elevate shikimate-derived flux. Heterologous expression of ppHPD improved tyrosine-to-HGA conversion, and reinforcement of the shikimate pathway enabled de novo HGA biosynthesis from glucose. Beyond conventional shikimate-focused amplification, a targeted single-gene screen across the MVA pathway unexpectedly pinpointed truncated HMG-CoA reductase (tHMGR) as the dominant determinant of HGA overproduction. Copy-number tuning further revealed an optimum at three genomic copies, increasing the shake-flask titer to 1766.43 mg/L.

Transcriptome profiling and metabolic network evaluation provided a mechanistic basis for this phenotype. Rather than acting as a direct transcriptional regulator, the massive NADPH consumption by overexpressed tHMGR creates a severe intracellular redox imbalance. This cofactor depletion triggers a compensatory upregulation of aromatic amino acid biosynthesis, with the strongest activation in the phenylalanine, tyrosine, and tryptophan pathways, ultimately redirecting massive carbon flux to strengthen shikimate-associated metabolism.

Importantly, the generality of this tHMGR-driven enhancement was independently validated using p-coumaric acid as a shikimate-derived reporter, where tHMGR integration increased production 2.53-fold in a feedback-relieved background without an obvious growth penalty. Ultimately, the application of this strategy in fed-batch fermentation achieved an HGA titer of 33.71 g/L, representing the highest microbial HGA production reported to date. While the current metabolic rewiring proved effective, further efforts to narrow the gap toward the theoretical maximum yield remain necessary for future industrial strain optimization.

Furthermore, based on this elucidated cofactor balance mechanism, future synergistic reinforcement of tHMGR and the kynurenine pathway could be explored to systematically expand the total NAD(P)H pool. This combined approach offers a highly promising strategy to supply optimal cofactors for energy-intensive metabolic routes, such as MVA-derived terpenoid biosynthesis.

Collectively, this study delivers a scalable route to sustainable HGA biomanufacturing and establishes the tHMGR-induced metabolic sink as a transferable strategy to enhance shikimate pathway output and boost the production of diverse shikimate-derived products.

CRediT authorship contribution statement

Chenchen Xu: Writing – original draft, Methodology, Investigation, Conceptualization. Wanyue Zhang: Formal analysis, Data curation. Quanlu Zhao: Visualization, Formal analysis. Xinyi Lin: Investigation, Data curation. Kai Wang: Writing – review & editing, Formal analysis. Meng Wang: Writing – review & editing, Data curation. Tianwei Tan: 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.

Abbreviations

L-Tyr, l-tyrosine; 4-HPP, 4-hydroxyphenylpyruvate; HGA, homogentisic acid; ARO8, aromatic aminotransferase; ARO9, aromatic aminotransferase; ARO4, 3-deoxy-d-arabino-heptulosonate 7-phosphate synthase (DAHPS); ARO1, pentafunctional AROM polypeptide; ARO2, chorismate synthase; ARO7, chorismate mutase; scARO4^K229L, feedback-resistant S. cerevisiae Aro4 variant (K229L); ylARO7^G139S, feedback-resistant Y. lipolytica Aro7 variant (G139S); ERG8, phosphomevalonate kinase; ERG10, acetyl-CoA acetyltransferase; ERG12, mevalonate kinase; ERG13, HMG-CoA synthase; ERG19, mevalonate diphosphate decarboxylase; IDI, isopentenyl diphosphate isomerase; tHMGR, truncated HMG-CoA reductase; SaGGPPS, geranylgeranyl diphosphate synthase from Sulfolobus acidocaldarius; 3-HAA, 3-hydroxyanthranilic acid; ACMS, 2-amino-3-carboxymuconate semialdehyde; NAMN, nicotinic acid mononucleotide; NAAD, nicotinic acid adenine dinucleotide; NAD⁺, nicotinamide adenine dinucleotide; rgTAL, tyrosine ammonia-lyase from Rhodotorula glutinis.

Appendix A. Supplementary data

The following is the Supplementary data to this article: