Combining multiplex metabolic engineering with adaptive evolution strategies for high-level succinic acid production in Yarrowia lipolytica

Abstract

Succinic acid, an essential platform chemical with extensive utility in biodegradable materials, pharmaceuticals, and the food industry, faces challenges of high energy consumption and environmental pollution in traditional chemical synthesis. Here, we employed multiplex metabolic engineering and adaptive laboratory evolution to enhance succinic acid biosynthesis in Yarrowia lipolytica. By attenuating succinate dehydrogenase (Sdh) activity, mitigating by-product accumulation, and enhancing the succinate synthesis pathway, engineered strains showed efficient succinic acid production from glycerol. The titer reached 130.99 g/L under unregulated pH conditions, translating to a yield of 0.35 g/g and a productivity of 0.70 g/(L·h). Subsequently, transporter engineering and adaptive evolution strategies were applied to enhance glucose utilization for succinic acid synthesis, yielding an evolved strain that eliminated the growth lag phase and produced 106.68 g/L succinic acid from glucose, which translated to a yield of 0.32 g/g and a productivity of 0.64 g/(L·h). Additionally, transcriptomic analysis and inverse metabolic engineering revealed that 4-hydroxyphenylpyruvate dioxygenase (4-Hppd) in the tyrosine degradation pathway partially restored the growth of Sdh-deficient strains on glucose, offering new insights for subsequent succinic acid biomanufacturing using Y. lipolytica.

1. Introduction

Succinic acid is an essential platform compound, with its primary applications lying in areas such as biodegradable plastics, food additives, and drug synthesis precursors [1,2]. Furthermore, as environmental regulations become increasingly strict globally, there is a rising demand for succinic acid as a critical precursor for the biodegradable polymer polybutylene succinate (PBS) [3,4]. Moreover, due to the bifunctional nature of succinic acid, its numerous derivatives, such as tetrahydrofuran, 2-pyrrolidone, γ-butyrolactone, and 1,4-butanediol, widely applied across pharmaceuticals, materials, and food industries [5]. Consequently, the US Department of Energy has listed succinic acid among the twelve bio-based platform chemicals that are given priority for development and exploitation [6,7]. Current succinic acid production relies on either chemical synthesis from petrochemicals or microbial fermentation [8]. Due to its advantages of environmental friendliness and utilization of renewable feedstocks, microbial fermentation is increasingly replacing traditional chemical synthesis and emerging as the dominant approach for succinic acid production [9].

Currently, various microorganisms are employed for succinic acid bioproduction [10]. Yarrowia lipolytica, distinguished by its exceptional acid resistance and well-established genetic tools, has become a prominent candidate in this field [11,12]. Early studies revealed that α-ketoglutaric acid undergoes oxidation via hydrogen peroxide to yield succinic acid [13]. Via a two-step decarboxylation procedure, Y. lipolytica-derived α-ketoglutaric acid from ethanol was converted into succinic acid, with the final titer amounting to 63.4 g/L [14]. Since then, researchers have begun to utilize Y. lipolytica as the chassis strain for succinic acid synthesis. Y. lipolytica possesses two primary endogenous metabolic pathways for succinic acid production: the mitochondrial oxidative tricarboxylic acid (oTCA) cycle and the peroxisomal glyoxylate cycle [37]. Additionally, the glyoxylate cycle supplies the metabolic intermediates needed for the oTCA cycle under certain conditions, such as when using fatty acids as substrate [15]. Yuzbashev et al. were the first to inactivate succinate dehydrogenase (Sdh) to facilitate succinic acid synthesis in Y. lipolytica. Specifically, upon knockout of the Sdh2-encoding gene, the modified strain exhibited a succinic acid production of 4 g/L [16]. Notably, as a consequence of Sdh deletion, the engineered strain could not regenerate FADH₂ when glucose served as the exclusive carbon source, leading to decreased oxidative phosphorylation and reduced ATP production [17]. However, glycerol-3-phosphate dehydrogenase (Gut2) can regenerate FADH₂ when using glycerol as the substrate. Thus, glycerol-mediated FADH₂ regeneration via Gut2 maintains oxidative phosphorylation activity, alleviating the ATP insufficiency caused by Sdh deletion [18]. Building on this foundation, additional studies on succinic acid production in Y. lipolytica primarily focused on metabolic engineering, adaptive evolution, expansion of the substrate spectrum, and fermentation optimization [5,19]. Notably, Cui et al. pioneered the construction of an active reductive tricarboxylic acid (rTCA) pathway in Y. lipolytica, and anchored rTCA pathway-related genes in the mitochondria to couple with the oTCA pathway for cofactor rebalancing of NADH. The final strain ultimately synthesized 111.9 g/L succinic acid without pH regulation, underscoring its viability as a high-performance microbial chassis for industrial succinic acid biosynthesis [20].

However, most research efforts on succinic acid biosynthesis in Y. lipolytica initiate genetic modification of strains with the inactivation or attenuation of Sdh activity, thereby causing severe impairment to the glucose growth capacity of engineered strains since Y. lipolytica is an obligately aerobic microorganism [21]. Adaptive laboratory evolution (ALE) is a strategy for optimizing specific properties of microorganisms without the need for specific knowledge of the underlying genetic mechanisms [[22], [23], [24]]. By culturing strains under stressful conditions, microbial strains exhibiting desired characteristics can ultimately be generated. Yang et al. developed an immobilization approach using cotton as a support material. After 14 passages, the cell growth rate in glucose culture medium was 2.3-fold higher than at the beginning. Under oxygen-limited conditions, the evolved strain produced 65.8 g/L succinic acid [18]. Zhong et al. employed ALE with Y. lipolytica Hi-SA2 as the parental strain to improve its succinic acid tolerance. After 27 generations of adaptation, the optimal evolved strain E501 yielded 102.1 g/L of succinic acid [25]. In addition, combining adaptive evolution with transcriptome or genome resequencing can further unravel the genetic mechanisms underlying the improved phenotypes, offering insights for more rational metabolic engineering approaches [26].

This study aimed to achieve efficient succinic acid production in Y. lipolytica by combining multiplex metabolic engineering with ALE. First, a multi-step iterative metabolic engineering strategy was used to enable engineered strains to efficiently accumulate succinic acid using glycerol as substrate. Subsequently, transporter protein engineering combined with adaptive evolution was employed to facilitate glucose utilization and succinic acid synthesis. Finally, transcriptomic analysis was implemented to investigate the underlying mechanisms of restored glucose utilization in the evolved strains, providing new insights for the subsequent biomanufacturing of succinic acid in Y. lipolytica.

2. Materials and methods

2.1. Strains and culture conditions

The strains utilized in this study are all enumerated in Table 1. The starting strain was Yarrowia lipolytica Po1f, in which the KU70 was disrupted by implementing the URA3-blaster strategy to enhance its homologous recombination (HR) efficiency [27]. As previously described, Escherichia coli DH5α was grown in Luria-Bertani (LB) liquid medium supplemented with 100 mg/L ampicillin, and positive transformants were selected using corresponding agar plates [28]. Y. lipolytica strains were grown in YPD liquid medium at 28 °C, 220 rpm for 14–16 h. For Y. lipolytica transformation, activation, and URA3 marker recovery, SC-Ura/Leu, YPD, and 5-FOA solid media were employed [28]. Glucose in the medium was substituted with glycerol as needed. The medium used for succinic acid fermentation was YPD/G60, in which the content of glycerol or glucose was increased to 60 g/L. YPD30-150 was employed for adaptive evolution, in which the glucose concentration of YPD liquid medium was modified to 30, 60, 90, 120, and 150 g/L. Yeast colonies were picked from YPD/G solid medium, cultivated in corresponding liquid medium for 10–14 h, and then transferred at an optical density (OD₆₀₀) of 0.02 to 250-mL shake flasks filled with 50 mL of the fermentation broth. The cultures were shaken at 220 rpm and 28 °C for 120 h to produce succinic acid.

Table 1.

Strains used in this study.

Strain	Genotype or property	Source
E. coli DH5α	supE 44 ΔlacU169 (φ80 lacZ ΔM15) hsdR17 recA1 endA1 gyrA 96 thi-1 relA1	Takara
Y. lipolytica Po1f (ATCC MYA-2613)	derived from W29, extracellular protease AEP deleted, grown on sucrose, MATa, leu 2–270, ura3-302, xpr2-322, axp1-2, pXPR2-SUC2 (URA3-, LEU2-)	ATCC
Y. lipolytica Po1f-Δku70	derived from Po1f, KU70 deleted, Δku70::HisG (URA3-, LEU2-)	28
Y. lipolytica Po1f-Δku70-LEU	derived from Po1f-Δku70 A08: LEU2 (URA3-, LEU2+)	This study
Y. lipolytica Po1f-Δku70-LEU-URA	derived from Po1f-Δku70-LEU ΔIntC3:: HisG-URA3-HisG (URA3+, LEU2+)	This study
Y. lipolytica ST-101	derived from Po1f-Δku70-LEU Δsdh5:: HisG-URA3-HisG (URA3+, LEU2+)	This study
Y. lipolytica ST-102	derived from Po1f-Δku70-LEU tP30SDH5:: HisG-URA3-HisG (URA3+, LEU2+)	This study
Y. lipolytica ST-103	derived from Po1f-Δku70-LEU tP50SDH5:: HisG-URA3-HisG (URA3+, LEU2+)	This study
Y. lipolytica ST-104	derived from Y. lipolytica ST-101 Δach1:: HisG-URA3-HisG (URA3+, LEU2+)	This study
Y. lipolytica ST-105	derived from Y. lipolytica ST-104 ΔIntF:: PEXP-PYC-Tcyc1t, HisG-URA3-HisG (URA3+, LEU2+)	This study
Y. lipolytica ST-106	derived from Y. lipolytica ST-105 Δpck:: HisG-URA3-HisG (URA3+, LEU2+)	This study
Y. lipolytica ST-107	derived from Y. lipolytica ST-106 ΔIntC:: PEXP-SCS2-Tmig1t, HisG-URA3-HisG (URA3+, LEU2+)	This study
Y. lipolytica ST-108	derived from Y. lipolytica ST-107 ΔIntC:: PTEF-KGDH-Txpr2t, HisG-URA3-HisG (URA3+, LEU2+)	This study
Y. lipolytica ST-109	derived from Y. lipolytica ST-106 ΔIntC1:: PTEFin-ICL-Tmig1t, PTDH-MLS-Txpr2t, HisG-URA3-HisG (URA3+, LEU2+)	This study
Y. lipolytica ST-110	derived from Y. lipolytica ST-109 ΔIntE2:: PEXP-SpMAE1-Tmig1t, HisG-URA3-HisG (URA3+, LEU2+)	This study
Y. lipolytica ST-111	derived from Y. lipolytica ST-110 adaptive evolution (URA3+, LEU2+)	This study
Y. lipolytica ST-112	derived from Y. lipolytica ST-111 Δjen1:: HisG-URA3-HisG (URA3+, LEU2+)	This study
Y. lipolytica ST-113	derived from Y. lipolytica ST-110 Δ7h:: PTEFin-4HPPD-Tmig1t, HisG-URA3-HisG (URA3+, LEU2+)	This study
Y. lipolytica ST-114	derived from Y. lipolytica ST-110 Δ7h:: PTEFin-DAK-Tmig1t, HisG-URA3-HisG (URA3+, LEU2+)	This study
Y. lipolytica ST-115	derived from Y. lipolytica ST-110 Δ7h:: PTEFin-THI-Tmig1t, HisG-URA3-HisG (URA3+, LEU2+)	This study
Y. lipolytica ST-116	derived from Y. lipolytica ST-113 ΔIntC3:: PTEFin-HMG2-Tmig1t, PH3-FAHA-Tcyc1t, HisG-URA3-HisG (URA3+, LEU2+)	This study

2.2. Plasmid construction and strain transformation

pUC-57 served as the backbone for the vector plasmids, with Ura functioning as the selection marker. The integrating plasmids used for gene knockout and expression comprised pUC-57, the 1500-bp homologous arms up- and downstream of the target site, and the URA3 expression cassette. The knocked-out genes were as follows: SDH5 (YALI0F11957g), ACH1 (YALI0E30965g), PCK (YALI0C16995g), JEN1 (YALI0C15488g) [29,30]. The overexpressed genes were PYC (YALI0C24101g), SCS2 (YALI0E24013g), KGDH (YALI0E16929g), ICL (YALI0C16885g), MLS2 (YALI0D19140g), SpMAE1 (SPAPB8E5.03), 4-HPPD (YALI0B21846g), DAK (YALI0E20691g), and THI (YALI0A09768g). As previously described, when knocking out genes or truncating the SDH5 gene promoter, the up- and downstream homologous arms were amplified from Y. lipolytica Po1f genomic DNA via PCR and then placed at both ends of the URA3 expression cassette through single-step cloning [31]. The constructed plasmids included pUC-HUH-ΔSDH5, pUC-HUH-tP30SDH5, pUC-HUH-tP50SDH5, pUC-HUH-ΔACH1, pUC-HUH-ΔPCK, and pUC-HUH-ΔJEN1. For the overexpression of target genes, one or two expression cassettes were inserted into the integrating plasmids that had already been constructed in our laboratory [32]. The endogenous genes, promoters, and terminators were obtained from the Y. lipolytica Po1f genome, while SpMAE1 was subjected to codon optimization and synthesized by General Biol (Chuzhou, China). The corresponding expression cassette elements were inserted into the integrating plasmids at different sites to construct pUC-HUH-IntF-PYC, pUC-HUH-IntC-SCS2, pUC-HUH-IntC-SCS2-KGDH, pUC-HUH-IntC1-ICL-MLS2, pUC-HUH-IntE2-SpMAE1, pUC-HUH-7h-4HPPD, pUC-HUH-7h-DAK, pUC-HUH-7h-THI, and pUC-HUH-IntC3-HMG2-FAHA.

URA-deficient yeast strains were grown in YPD/G liquid medium to an OD₆₀₀ of 0.8–1.0, centrifuged at 500 g to remove the culture supernatant, and transformed with linearized plasmids using the Frozen-EZ Yeast Transformation Kit II (Zymo Research, Orange, CA). The transformed mixture was spread onto SC-Ura solid medium to allow single colony formation, followed by colony picking, genomic DNA extraction, PCR verification, and culture of validated recombinant strains in YPD/G medium for subsequent fermentation validation or streaking on 5-FOA plates to eliminate the URA3 marker.

2.3. Adaptive laboratory evolution

The initial strain for adaptive evolution was ST-110, and three parallel ALE experiments were conducted. Specifically, a single colony picked from YPG solid medium was transferred into liquid YPG and cultured overnight (12–14 h). The culture broth after overnight cultivation was centrifuged, washed twice with sterile water, and transferred into YPD medium for adaptive evolution. The glucose was initially set to 30 g/L, which was gradually elevated to 150 g/L in 30 g/L increments at each step. Once the cells reached the mid-log phase, they were used to inoculate the next generation at an OD₆₀₀ of 0.1. The glucose concentration was increased when the maximum OD₆₀₀ remained unchanged for two consecutive rounds of adaptive evolution. Each transfer interval was approximately 24–36 h, with a total of 20 transfers performed. When the glucose concentration reached 150 g/L and remained stable for three generations, single colonies were picked from YPD150 solid medium. Subsequently, succinic acid yield and glucose consumption of the single colonies were measured to evaluate their glucose utilization ability. The strain with the best succinic acid production performance was named ST-111 and used in subsequent experiments.

2.4. RNA extraction, sequencing, and analysis

The strains ST-110 and ST-111 described in Section 2.1 were seeded into YPD60 medium at an initial OD₆₀₀ of 0.02. After 24 h of cultivation, the cultures were centrifuged at 5000×g and the cell pellets collected for transcriptomic analysis. Three parallel samples were prepared for each strain. RNA was extracted via a protocol reported earlier [32]. DEGs (Differentially expressed genes) were identified via statistical analysis with |log₂ FC| > log₂(2) and FDR < 0.05.

2.5. Bioreactor fermentation

To assess the engineered strains' capability to synthesize succinic acid, fermentation was conducted in a 5-L bioreactor. Glycerol was used as the carbon source for strain ST-109, both in the seed culture and the bioreactor. The seed culture was prepared by transferring the overnight culture into a 250-mL shake flask containing 50 mL YPG and incubating it for 24 h. Thereafter, the culture was moved to a 500-mL shaking flask filled with 100 mL YPG for another 36 h. The bioreactor contained an initial 2 L of medium with 300 g glycerol, 40 g peptone, and 20 g yeast extract. When the glycerol concentration fell below 10 g/L, 900 g/L glycerol was dosed into the system at 3–10 mL/h. The culture was maintained at a gas flow rate of 2 vvm, a rotational speed of 300–800 rpm, 28 °C, and 20 % DO, without pH regulation. For strain ST-112, glucose was used instead of glycerol as substrate in both the seed culture and the bioreactor. The bioreactor was initially charged with 50 g/L glucose, and 900 g/L glucose was fed at 5–15 mL/min when the glucose concentration dropped below 10 g/L. The pH was controlled at 5.5 by adding 10 M NaOH. Samples were taken from the fermentation broth every 12 h to monitor the concentrations of the substrate and succinic acid.

2.6. Analytical methods

In this study, the RID-1260 differential refractive index detector of a 1260 Infinity II Prime HPLC instrument (Agilent, USA) was used to detect glycerol, glucose, and succinic acid. Sample pretreatment, mobile phase composition, and detection conditions were in accordance with previous studies [33]. OD₆₀₀ was measured following the method of Wang et al. [28].

For lipid determination, 1 mL of the fermentation broth was pelleted via centrifugation, followed by washing the cells twice with pure water, lyophilized, and freeze-dried to measure the dry weight. Concurrently, the fermentation broth was centrifuged, and approximately 0.02 g (dry weight equivalent) of wet bacterial paste was placed into a disruption tube, to which 250 μL of pure water and 750 μL of chloroform-methanol solution (2:1 ratio) were added. After adding 0.02 g of 0.5 mm diameter glass beads for disruption, the mixture was vortexed at 1200 rpm for 5–10 min, centrifuged, and the lower layer aspirated and dried using an automatic quantitative concentrator (SpeedVac SPD121P; Thermo Fisher Scientific, USA). Finally, the total lipid proportion was calculated based on the lipid weight corresponding to 1 mL of the original fermentation broth.

3. Results and discussion

3.1. Weakening succinate dehydrogenase activity promoted succinic acid accumulation

Yarrowia lipolytica features an exceptionally active tricarboxylic acid (TCA) cycle, enabling the high-yield biosynthesis of diverse intermediate metabolites, including succinic acid, α-ketoglutaric acid, isocitric acid, citric acid, and malic acid, via synthetic biology methods [12,34,35]. To produce succinic acid, downstream conversion of succinate in mitochondria is typically prevented by weakening succinate dehydrogenase (Sdh) activity, such as by truncating the Sdh subunit gene promoter or knocking out the Sdh subunit gene [36]. Accordingly, we constructed the engineered strains ST-101, ST-102 and ST-103 by knocking out the encoding gene of Sdh subunit 5 or truncating its promoter to 30 or 50 bp upstream of the initiation codon using the URA-Blaster technique (Fig. 1a). Since inactivation of Sdh disrupts FADH₂ regeneration, the engineered strains were fermented with glycerol as the substrate. Among them, ST-101 generated 2.49 g/L succinic acid at the highest yield, with acetic acid formed at 4.72 g/L as the dominant by-product (Fig. 1b). The OD₆₀₀ of this strain peaked at 24 h and then declined progressively as fermentation time increased, which was primarily attributed to the gradual accumulation of acetic acid (Fig. 1c). By contrast, succinic acid was only detected in strains ST-102 and ST-103 at 48 h and 72 h, reaching concentrations of 1.04 g/L and 0.67 g/L, respectively (Fig. 1b). Notably, the by-product accumulation of acetic acid went undetected in the two strains, which explains why they reached a higher OD₆₀₀ which subsequently did not decrease (Fig. 1c). Based on the yield comparison, strain ST-101 was selected for subsequent experiments.

Fig. 1.

Weakening of Sdh activity. a. Schematic representation of the applied strategy. Sdh5: Succinate dehydrogenase subunit 5; P: The promoter region of the SDH5; T: The terminator region of the SDH5; HUH: HisG-URA3-HisG expression cassette; tP30-SDH5: Truncation of the SDH5 promoter to 30 bp upstream of the initiation codon; tP50-SDH5: Truncation of the SDH5 promoter to 50 bp upstream of the initiation codon; HR: Homologous recombination. b. Succinic acid production kinetics of strains ST-101, ST-102, and ST-103 in YPG medium. c. Biomass accumulation of strains ST-101, ST-102, and ST-103 (OD₆₀₀) in YPG medium.

3.2. Optimizing metabolic pathways further enhanced succinic acid synthesis

As previously studied, disruption of the oxidative TCA cycle causes metabolic disturbances and cofactor imbalances, which in turn lead to the spillover of multiple byproducts (Fig. 2a). Specifically, acetyl-CoA hydrolase (Ach1) cleaves acetyl-CoA, leading to acetic acid spillover. Knockdown of the Ach1 gene reduced acetic acid titer to 0.2 g/L and boosted succinic acid titer to 5.84 g/L (Fig. 2b). The conversion of pyruvate to oxaloacetate via pyruvate carboxylation represents the first step of gluconeogenesis. Subsequent decarboxylation of oxaloacetate yields phosphoenolpyruvate, which can then proceed along the gluconeogenic pathway to form upstream metabolites. To enhance the supply of oxaloacetate, the precursor of the oxidative TCA cycle, overexpression of the endogenous pyruvate carboxylase (Pyr) gene and knockdown of the phosphoenolpyruvate carboxykinase (Pck) gene further increased the succinic acid titer to 7.71 g/L (Fig. 2b).

Fig. 2.

Metabolic engineering of Y. lipolytica to improve succinic acid production from glycerol. a. Schematic diagram of metabolic pathway optimization. Glu: Glucose; GAP: Glyceraldehyde 3-phosphate; Cit: Citrate; SA: Succinate; 3 PG: 3-phosphoglycerate; Fum: Fumarate; Icit: Isocitrate; PEP: Phosphoenolpyruvate; Succ: Succinyl-CoA; OAA: Oxaloacetate; Pyr: Pyruvate; DHAP: Dihydroxyacetone phosphate; Ac-CoA: Acetyl-CoA; Gly: Glycerol; G3P: 3-phosphoglycerol; α-KG: α-ketoglutarate. Pck: Phosphoenolpyruvate carboxykinase; Kgdh: α-ketoglutarate dehydrogenase; Pyrc: Pyruvate carboxylase; Ach1: Acetyl-CoA hydrolase; Icl: Isocitrate lyase; Scs2: Succinyl-CoA synthase. b. Optimization of metabolic pathways. “-” indicates the absence of metabolic engineering manipulation. “+” indicates gene integration into the genome; “Δ” indicates gene knockout. c. Production of succinic acid by strain ST-109 in a 5-L bioreactor. d. The succinic acid titer and yield of strain ST-109 were measured at different time points in a 5-L bioreactor.

Y. lipolytica has two endogenous pathways for succinic acid generation—the oxidative TCA cycle and the glyoxylate cycle. Notably, the glyoxylate cycle is located in the peroxisomes, and its key enzyme, isocitrate lyase (Icl), requires acetyl-CoA precursors, such as fatty acids and ethanol, to be activated and to replenish the oxidative TCA cycle with succinic acid. Therefore, the endogenous succinic acid synthesis pathway was enhanced to improve the product yield. The succinic acid titers of strain ST-107 constructed by overexpressing SCS2 and strain ST-108 constructed by co-expressing SCS2 and KGDH were 8.58 g/L and 7.79 g/L, respectively (Fig. 2b). Meanwhile, the MLS targeting sequences in the peroxisome-anchored proteins isocitrate lyase and malate synthase (Mls) were removed for expression in the cytoplasm, and the P_TEFin promoter was utilized to control Icl gene expression. The succinic acid yield of the engineered strain ST-109 was increased to 9.07 g/L (Fig. 2b). In addition, strain ST-109's fermentation performance was evaluated in a 5-L bioreactor without pH regulation, yielding a succinic acid titer of 130.99 g/L, with a corresponding yield of 0.35 g/g and productivity of 0.70 g/(L·h) (Fig. 2c).

3.3. Transporter engineering combined with adaptive laboratory evolution promoted glucose utilization and succinic acid production

Although the strain with Sdh5 gene knockout retained the ability to utilize glucose, it exhibited a longer lag period when using glucose as substrate compared to glycerol (Fig. 3c). Therefore, the strain was engineered to express cell membrane transporter proteins and adaptively evolved to shorten its lag phase and enhance glucose utilization for succinic acid production (Fig. 3a).

Fig. 3.

Transporter protein engineering combined with adaptive evolution promoted glucose utilization and succinic acid production. a. Diagrammatic sketch of the strategy to boost glucose utilization. Gly: Glycerol; SA: Succinic acid; Mae1: C4-dicarboxylic acid transporter protein; Glu: Glucose. b. The OD₆₀₀ of strains ST-109, ST-110, and ST-111 cultured in YPD medium. c. The glucose consumption curve of strains ST-109, ST-110, and ST-111 cultured in YPD medium. d. The 3D scatter plot shows the distribution relationships among OD₆₀₀, succinic acid titer, and yield in the 3D scatter plot of ST-109 evolved strains. The color of the points corresponds to succinic acid titer intervals (as shown in the legend; e.g., 14.4–16.0 g/L is red, etc.). e. The succinic acid titer and succinic acid yield of the strains after 120 h of cultivation in YPD medium. ALE: Adaptive laboratory evolution; Jen1: dicarboxylic acid transport protein. “-” indicates the absence of metabolic engineering manipulation. “+” indicates implementation of metabolic engineering; “Δ” indicates gene knockout.

Several studies have shown that C4-dicarboxylic acid transporter proteins can promote the efflux of malate, fumarate as well as succinate [37]. Among them, Mae1 derived from Schizosaccharomyces pombe classifies as a voltage dependent SLAC1 transporter, which operate without relying on proton or Na⁺ gradients for motive force, making them more energetically advantageous [37]. After heterologously expressing SpMAE1 in strain ST-109, the engineered strain ST-110 exhibited notable improvements in succinic acid titer and yield, which increased from 5.75 g/L and 0.16 g/g to 8.37 g/L and 0.20 g/g, respectively, while OD₆₀₀ at 120 h also increased from 6.52 to 9.72 (Fig. 3e). However, strain ST-110 only exhibited an increase of the OD₆₀₀ from 0.42 to 0.89 at 24 h (Fig. 3c), indicating that overexpression of SpMAE1 alone did not fully alleviate the glucose utilization defect of Sdh5-deficient strains.

Therefore, we evolved strain ST-110 to shorten its lag period when grown on glucose. Specifically, YPD30 served as the starting medium and the glucose concentration was gradually increased to 60, 90, 120, and 150. The next generation was seeded when the OD₆₀₀ reached 3–4 and the glucose concentration was increased after 3 generations of stable growth. Adaptive evolution was divided into 3 independent parallel experiments. After 20 rounds, the fermentation broths from the three parallel experimental groups were spread on YPD150 solid medium and a total of 60 single clones underwent selection via colony size comparison. The 60 strains derived from adaptive evolution were fermented in YPD60 medium for 120 h for initial screening (Fig. 3d; Fig. S1), and 5 were selected for rescreening based on the initial results. Among them, the evolved strain ST-110E30 showed the highest succinic acid yield of 0.31 g/g, whereas the evolved strain ST-110E34 displayed the highest titer of 15.39 g/L, with the OD₆₀₀ and yield were also increased to 11.07 and 0.28 g/g, respectively (Fig. 3e). Therefore, this evolved strain ST-110E34 was named ST-111 and used for subsequent experiments.

In Y. lipolytica, the Jen1 protein is responsible for the uptake of dicarboxylic acids, including succinic acid. By knocking out the Jen1 gene to prevent the reabsorption of succinic acid, strain ST-112 achieved a product titer of 16.02 g/L and a yield of 0.29 g/g (Fig. 3e). In Issatchenkia orientalis, knocking out the gene encoding the succinic acid uptake transporter protein enhanced succinic acid production [38]. By analogy, the gene encoding the main succinic acid transporter was knocked out in Y. lipolytica, but this only moderately increased the product titer. This may result from catabolite repression, where the presence of glucose suppresses the expression of genes responsible for the uptake and metabolism of secondary carbon sources, such as organic acids [39].

3.4. Transcriptomic analysis of the evolved strain

The best strain obtained through adaptive laboratory evolution showed a significant shortening of the lag phase on glucose. Nevertheless, the remodeling of cellular metabolism that occurred during this evolutionary process remained unknown. To comprehensively understand the genetic changs that promoted glucose utilization and succinic acid production, we performed transcriptome sequencing on the parental strain ST-110 and its evolved derivative ST-111 at the 24-h time point. According to the transcriptomic analysis, a total of 1445 genes were significantly differentially expressed (|log₂FC| > log₂(2), FDR < 0.05), among which 1242 were up- and 203 downregulated (Fig. 4a; Fig. S2). Based on KEGG annotation, these genes were classified into 20 distinct biological processes, among which pathways related to ribosome biogenesis, tyrosine metabolism, and phenylalanine metabolism were significantly perturbed (Fig. 4b). During the logarithmic growth phase, ribosome biogenesis is typically highly active, so it was unsurprising that the evolved strain exhibited high expression levels of ribosome biogenesis genes [40]. Tyrosine and phenylalanine degradation pathways share the same catabolic route, whereby phenylalanine is first converted into tyrosine by phenylalanine hydroxylase (Pah), and tyrosine is then gradually degraded into fumarate and acetoacetate, which enter the TCA cycle and ketone body metabolism [41]. Additionally, Gene Ontology (GO) analysis of the relevant genes demonstrated significant enrichment in multiple biological processes (Fig. 4c), such as detoxification, rhythmic processes, inter/intraspecies interactions, carbon/nitrogen utilization, and metabolic/regulatory pathways. In addition, molecular functions and cellular components including transporter activity, catalytic activity, binding, virion components, and protein-containing complexes were predominantly represented.

Fig. 4.

Transcriptomic analysis of the evolved strain. a. Volcano plot comparing the transcriptomes of ST-110 and ST-111. b. KEGG enrichment bar chart for ST-110 and ST-111 displaying the top 20 pathways with the lowest q-values. c. Bubble chart showing GO enrichment of differentially expressed genes between strains ST-110 and ST-111.

Subsequently, we focused on transcriptional level changes in the central carbon metabolism of evolved strains based on KEGG analysis (Fig. 5). The upregulated genes encoded enzymes included dihydroxyacetone kinase (Dak, 2.7.1.28) and fructose 1,6-bisphosphatase (Fbp, EC 3.1.3.46) in the gluconeogenesis pathway, citrate synthase (Cs, EC 2.3.3.1) in the TCA cycle, aspartate aminotransferase (Aat, EC 2.6.1.1) in the malate-aspartate shuttle pathway, nitrilase family member 2 (Nit2, EC 3.5.1.3), malate synthase (Mls, EC 2.3.3.9), as well as serine ammonia-lyase (Sal, EC 4.3.1.17) and d-amino acid oxidase (Dao1, EC 1.4.3.3). The downregulated gene encoded transketolase (Tkl, EC 2.2.1.1), which functioned in the hexose monophosphate pathway (HMP) (Fig. S3). Transcriptional analysis showed that reactions catalyzed by enzymes encoded by the upregulated genes in the central carbon metabolism of evolved strains mainly supplied intermediates to the TCA cycle and directed metabolic fluxes toward the gluconeogenic pathway. Aat catalyzes the conversion of cytoplasmic aspartate (ASP) to oxaloacetate (OAA), and Cs catalyzes its condensation with acetyl-CoA into citrate. Glycine is converted into malate via a two-step reaction catalyzes by Dao and Mls, and malate enters the mitochondrial TCA cycle. In gluconeogenesis, Fbp and Dak generate Fructose 6-phosphate (F6P) and Glyceraldehyde 3-phosphate (GAP), which are can shuttled into the non-oxidative hexose monophosphate pathway by Tkl.

Fig. 5.

Comparison of gene expression levels within the central carbon metabolic network between strains ST-110 and ST-111. Glu: Glucose; 4-HPP: 4-Hydroxyphenylpyruvate; G6P: Glucose 6-phosphate; OAA: Oxaloacetate; DHA: Dihydroxyacetone; F6P: Fructose 6-phosphate; Mal: Malate; HGA: Homogentisate; F1,6BP: Fructose 1,6-bisphosphate; Glyo: Glyoxylate; MLAA: Maleylacetoacetate; GAP: Glyceraldehyde 3-phosphate; FA: Fumarylacetoacetate; G1,3P: 1,3-Bisphosphoglycerate; E4P: Erythrose 4-phosphate; Fum: Fumarate; PEP: Phosphoenolpyruvate; ASP: Aspartate; Pyr: Pyruvate; Glyc: Glycine; Ser: Serine; Ac-CoA: Acetyl-Coenzyme A; Cit: Citrate; Ru5P: Ribulose 5-phosphate; R5P: Ribose 5-phosphate; Xu5P: Xylulose 5-phosphate; S7P: Sedoheptulose 7-phosphate; SA: Succinate.

3.5. Identification and functional verification of genes restoring glucose utilization for succinic acid production

Following the initial transcriptomic analysis, we performed targeted screening for genes involved in FADH₂ regeneration or ATP biosynthesis with significantly elevated expression levels. However, no potential candidates were identified. Consequently, our focus shifted to genes exhibiting high |log₂FC| values and possessing NCBI annotations, specifically including 4-hydroxyphenylpyruvate dioxygenase (4-Hppd, YALI0B21846p), dihydroxyacetone kinase (Dak, YALI0E20691p), and thiamine thiazole synthase (Thi, YALI0A09768p) (Fig. 6a). To interrogate their functional roles, we individually overexpressed these genes in the parental strain ST-110, generating engineered strains ST-113, ST-114, and ST-115, respectively. Functional validation was subsequently conducted using glucose as the exclusive substrate. The fermentation profiles revealed that the three inverse metabolic engineering interventions all alleviated the prolonged lag phase observed in ST-110 to varying degrees. Notably, ST-113 with the 4-Hppd gene integration displayed the highest OD₆₀₀ (7.02 at 24 h) (Fig. 6c), accompanied by an enhanced succinic acid titer from 8.37 to 10.25 g/L and a yield improvement from 0.21 to 0.24 g/g (Fig. 6b). By contrast, although ST-114 and ST-115 moderately restored the lag phase duration on glucose, their succinic acid production and yield either remained unchanged or decreased versus the parental strain.

Fig. 6.

Inverse metabolic engineering and theoretical assessment of the principles underlying its mechanism. a. Transcriptional changes of the YALI0B21846g, YALI0E20691g, and YALI0A09768g genes with high |log₂FC| values at 24 h. b. Succinic acid titer and yield of the control strain, recombinant strains obtained through inverse metabolic engineering, and evolved strains at 120 h. c. Growth performance of the control strain, recombinant strains obtained through inverse metabolic engineering, and evolved strains. d. Metabolic map of the tyrosine degradation pathway. 4-Hppd: 4-hydroxyphenylpyruvate dioxygenase; Hmga: homogentisate 1,2-dioxygenase; Faha: fumarylacetoacetase. e. Metabolic diagram of succinic acid production from glucose via glycolysis or the hexose monophosphate pathway. EMP: Embden-Meyerhof-Parnas pathway; TDP: Tyrosine degradation pathway; HMP: Hexose monophosphate pathway. f. Lipid content of strains ST-110 and ST-113 relative to dry cell weight. g. The TDP activated in the process of adaptive evolution is capable of enhancing carbon flux within the HMP pathway. ALE: Adaptive laboratory evolution.

Given the superior performance of engineered strain ST-113, we further focused on the metabolic pathways associated with 4-Hppd. As a pivotal enzyme within the tyrosine and phenylalanine degradation pathways, the 4-Hppd enzyme catalyzes 4-hydroxyphenylpyruvate to form homogentisate, which is subsequently metabolized into acetoacetate and fumarate (Fig. 6d). In yeast, the tyrosine biosynthesis pathway starts from erythrose 4-phosphate (E4P) and phosphoenolpyruvate (PEP), which are condensed into 3-deoxy-d-arabino-heptulosonate 7-phosphate (DAHP) by DAHP synthase. DAHP is then converted into 4-hydroxyphenylpyruvate through a series of catalytic steps, and the latter is transformed into l-tyrosine by tyrosine aminotransferase (Tat) [42,43]. This process relies on E4P from the hexose monophosphate pathway (HMP) and requires NADPH as a reducing agent (Fig. S4). Thus, we hypothesized that the degradation of 4-hydroxyphenylpyruvate continuously activates the HMP pathway, alleviating metabolic inhibition caused by succinate dehydrogenase (Sdh) deficiency (Fig. 6e). Additionally, the degradation of 4-hydroxyphenylpyruvate yields the end-products acetyl-CoA and fumarate, which can be channeled into the TCA cycle. Given that Y. lipolytica is celebrated for its lipid-producing capacity and NADPH serves as a key cofactor for fatty acid biosynthesis, we quantified lipid accumulation in strains ST-110 and ST-113. The findings revealed a substantial lipid content increase from 5.6 % to 14.9 % of dry cell weight in ST-113, supporting the hypothesis of HMP pathway activation (Fig. 6f) [44]. This activation may also have partially reduced the flux of the glycolytic pathway, thereby alleviating the cofactor imbalance caused by excessive NADH accumulation (Fig. 6g).

Concurrently, we observed that the catalytic product of 4-Hppd is homogentisate, whose excessive accumulation generates reactive oxygen species (ROS), causing multisystem cellular damage, metabolic pathway imbalance, and growth inhibition [41]. This may explain the gradual decrease of OD₆₀₀ in strain ST-113 after 72 h (Fig. 6c). Therefore, we further explored whether some of the DEGs were associated with 4-hydroxyphenylpyruvate degradation or ROS mitigation. Notably, homogentisate 1,2-dioxygenase (Hmga, KAJ8053843.1) and fumarylacetoacetase (Faha, YALI1D36215p) genes involved in 4-hydroxyphenylpyruvate degradation, as well as catalase T1 (Ctt1, YALI0E34265p) gene related to oxidative stress alleviation, were significantly upregulated (Fig. S5) [45]. Accordingly, we co-overexpressed the HMGA and FAHA in ST-113 to generate strain ST-116. Unexpectedly, ST-116 accumulated only trace amounts of succinic acid in the fermentation broth, with improved growth being the dominant phenotype (Fig. 6b; Fig. 6c). Although three genes in the 4-hydroxyphenylpyruvate degradation pathway were significantly upregulated in the evolved strain, their integration into the control strain via inverse metabolic engineering failed to yield ideal results, potentially due to metabolic flux imbalances introduced during rational engineering, or the lack of additional, global regulatory effects of adaptive evolution. In the evolution-derived strain ST-111, the slight downregulation of the TKL in the HMP pathway might serve as a metabolic node to balance the two pathways, thereby enabling the evolved strain to enhance glucose utilization for succinic acid synthesis.

3.6. Fed-batch fermentation in a 5-L bioreactor

To evaluate the fermentation efficiency of strain ST-112 with glucose as the substrate, the process was expanded to a 5-L bioreactor. An initial concentration of 50 g/L glucose was used, followed by feeding with 900 g/L glucose to sustain carbon levels. In the absence of pH control, fermentation was inhibited as pH dropped to 3.5, halting growth and succinate accumulation (Fig. 7a). By stabilizing the pH at 5.5, the fermentation yielded 106.68 g/L succinic acid over 168.35 h, with a 0.32 g/g yield and 0.64 g/(L·h) productivity (Fig. 7c). The findings indicate that adaptive evolution contributes to the partial recovery of glucose utilization in Sdh-deficient mutants.

Fig. 7.

Succinic acid fermentation from glucose by strain ST-112 in a 5-L bioreactor. a. Fermentation profile without pH control. b. Fermentation profile with an initial glucose concentration of 150 g/L c. Fermentation profile with an initial glucose concentration of 50 g/L.

4. Conclusion

Overall, this study remarkably improved succinic acid production in Y. lipolytica via multiplex metabolic engineering combined with adaptive laboratory evolution. Without pH control, the engineered strain produced 130.99 g/L succinic acid from glycerol, with a yield of 0.35 g/g and productivity of 0.70 g/(L·h). Through transporter engineering and adaptive evolution, glucose utilization for succinic acid synthesis was significantly enhanced, yielding 106.68 g/L succinic acid with a yield of 0.32 g/g and a productivity of 0.64 g/(L·h). Additionally, transcriptomic analysis and inverse metabolic engineering demonstrated that overexpression of 4-hydroxyphenylpyruvate dioxygenase (4-Hppd) from the tyrosine degradation pathway partially restored the growth of Sdh-deficient strains on glucose. This work demonstrated the potential of multiplex metabolic engineering combined with ALE to enhance glucose utilization for succinic acid synthesis, and highlighted the metabolic rewiring of evolved strains revealed by transcriptomic analysis, offering new insights for subsequent succinic acid biomanufacturing using Y. lipolytica.

CRediT authorship contribution statement

Tao Sun: Writing – original draft, Conceptualization. Mei-Li Sun: Methodology, Formal analysis. Lu Lin: Resources, Investigation. Jian Gao: Writing – review & editing. Rodrigo Ledesma-Amaro: Writing – review & editing. Kaifeng Wang: Supervision, Funding acquisition. Xiao-Jun Ji: Writing – review & editing, Supervision, Project administration, Funding acquisition.

Declaration of interest statement

We declare that we have no financial and personal relationships with other people or organizations that can inappropriately influence our work, there is no professional or other personal interest of any nature or kind in any product, service and/or company that could be construed as influencing the position presented in, or the review of, the manuscript entitled, “Combining multiplex metabolic engineering with adaptive evolution strategies for high-level succinic acid production in Yarrowia lipolytica”.

Appendix A. Supplementary data

The following is the Supplementary data to this article.