The synergistic effect of formate dehydrogenase and carbonic anhydrase accelerates the ethanol fermentation process and improves carbon recovery

Abstract

Bioethanol is a significant product in the two-carbon strategy, but the CO₂ generated during ethanol fermentation incurs substantial carbon losses. To tackle this issue, we explored a prospective dual-enzyme approach, using formate dehydrogenase (FDH) and carbonic anhydrase (CA), aiming to convert CO₂ to formate in situ, transforming the process from carbon-neutral to carbon-negative. TsFDH was selected among four FDH candidates due to its high enzymatic activity. To overcome the low solubility of CO₂, CA was co-expressed with FDH, facilitating CO₂ hydration. The recombinant strain S. cerevisiae-tsfdh-ca efficiently converted CO₂ to formate and exhibited rapid glucose consumption, which produced 76.3 mg L⁻¹ formate during fermentation, with a 69% increase in the glucose fermentation rate and a 108% increase in formate yield. Additionally, it reduced CO₂ emissions significantly, and the ratio of CO₂ release and ethanol production decreased from 1.89 to 1.39. The dual-enzyme system, FDH/CA, as elucidated by transcriptomic analysis, amplifies fermentation efficiency. It achieves this by concurrently sustaining the integrity and optimal efficiency of the electron transport chain and optimizing metabolic fluxes, particularly those in the TCA cycle and oxidative phosphorylation. Consequently, it offers a promising strategy for mitigating carbon emissions and enhancing the production of biofuels and chemicals and thus holds immense potential for contributing to a more sustainable and environmentally friendly energy future.

---

1 Introduction

Energy supply dominated by fossil energy has always been one of the important factors influencing the global economy and international political relations. From a scientific perspective, the excessive reliance of industry, agriculture, and economic development on fossil fuels not only leads to energy supply issues but simultaneously releases a substantial amount of carbon dioxide and other greenhouse gases, resulting in an increasingly severe greenhouse effect. It is projected that by the year 2100, the concentration of CO₂ will further increase by 60% at the current rate.¹ Numerous environmental issues have emerged owing to the intensification of the greenhouse effect. To solve this problem, it is urgent to develop new clean renewable energy sources and also to reduce the production of CO₂.²

Bioethanol has gained prominence due to its high energy density, transportability, and regenerative properties. In particular, lignocellulosic bioethanol offers environmentally beneficial alternatives to fossil fuels and avoids the emission of greenhouse gases. The utilization of waste and residues (such as straw and corn cobs) for bioethanol production helps address the challenge of the “food vs. fuel” dilemma.³ A recent report indicates that the global demand for biofuels is expected to grow by more than 20% between 2022 and 2027.⁴ However, regardless of the raw material used, the bioethanol production process invariably releases an equal amount of CO₂. Notably, approximately one-third of the substrate is lost as CO₂, thereby resulting in a substantial and concerning loss of carbon resources.⁵ Therefore, utilizing CO₂ for various energy applications, including ethanol fermentation, provides a promising pathway to address environmental challenges and advance sustainable energy solutions.⁶

Low dissolved CO₂ from the gas phase diffuses into water, making it better to convert CO₂ to soluble one-carbon (C1) compounds.⁷ Among various C1 compounds like formaldehyde and methanol, formate shows distinct advantages over the lowest standard reduction potentials. On this basis, formate has been widely used as feedstock for the production of various C2–Cn compounds.^8,9 Formate dehydrogenase (FDH) is one of the most promising CO₂ fixation enzymes, which has received much attention for its ability to convert CO₂ to formate without any other organic compounds except for energy in the form of electrons. However, the problems of low reduction efficiency, poor enzyme stability, and high cost of NADH hinder the industrial scale-up of FDHs. Strategies such as immobilizing FDHs on metal–organic frameworks (MOFs),¹⁰ sol–gel carriers,¹¹ and membranes¹² have been investigated to enhance the recycling stability, conversion efficiency, and formate yield. Considering the high cost of NADH, cofactor regeneration is desperately necessary. Chen et al. established a bioelectrocatalytic system combining enzymatic reactions with the electrochemical regeneration of cofactors using an electrochemical reaction, converting CO₂ to formate at a rate of 79 ± 3.4 mM h⁻¹.¹⁰ The biocatalytic method, which can activate CO₂ and reduce the activation energy required for the reaction, is commonly preferred due to its superior selectivity, low energy consumption and relatively mild reaction conditions.¹³ In particular, the process of glucose fermentation for ethanol production generates excessive NADH which would integrate with CO₂ reduction. Researchers endeavored to fix CO₂ generated during bioethanol production and it was found that fixing CO₂ promoted ethanol yield.^14,15 However, the mechanism of CO₂ reduction contributing to ethanol production still remained unspecified. Additionally, due to the low solubility of CO₂ and the low reduction activity of FDH, the recovery efficiency of CO₂ and the yield of formate are relatively low. Carbonic anhydrase (CA) can efficiently catalyze 10⁶ molecules of CO₂ per second under ideal conditions,¹⁶ playing a pivotal role in maintaining equilibrium between CO₂ and HCO₃⁻ to facilitate cellular processes rather than just CO₂ fixation. Indeed, these advantages drive CA to be widely applied in assisting absorption, separation, and precipitation techniques.^17,18 Given the distinctive capabilities of both FDH and CA, the combination of these enzymes for CO₂ capture applications has gained significant attention.^19–21

To address the above issue, this work demonstrated an efficient CO₂ reductase FDH for constructing a feasible CO₂ utilization system in Saccharomyces cerevisiae to stimulate ethanol fermentation by accelerating CO₂ to formate. The combination of CO₂ fixation with ethanol production by the collaboration of FDH and CA enabled the efficient use of carbon from raw materials, promoted fermentation, and reduced carbon emissions in the environment. The mechanism of boosting glucose fermentation efficiency and ethanol yield was clarified. The present study provided a collaborative advantage strategy that enhanced production yield. Moreover, it put forward a novel first-step approach for biocatalytic carbon capture and utilization, aiming at both CO₂ reduction and the substitution of important commodity chemicals.

2 Materials and methods

2.1 Materials

The genes (scfdh from S. cerevisiae, tsfdh from Thiobacillus sp. KNK65MA, atfdh from Arabidopsis thaliana, mtfdh from Myceliophthora thermophila, and ca from Bos taurus) used in this study were codon optimized and synthesized by BBI Life Sciences (Shanghai, China). Plasmids pET-32a(+), pACYCDuet-1, and pRS424 and competent cells Escherichia coli DH5α/BL21(DE3) and S. cerevisiae S288C were purchased from Merck (Massachusetts, USA). Analytical grade sodium bicarbonate (A610482), sodium formate (A500851), reduced β-nicotinamide adenine dinucleotide (NADH, A600642), β-nicotinamide adenine dinucleotide (NAD⁺, A600641), yeast nitrogen base (YNB, A610507), and isopropyl-β-D-thiogalactopyranoside (IPTG, A100487) were purchased from BBI Life Sciences (Shanghai, China). Amino acids were purchased from Solarbio Life Sciences (Beijing, China). All molecular biology enzymes were from Takara Bio (Beijing, China). The His-Trap and PD-10 columns were purchased from GE Healthcare (Little Chalfont, UK).

2.2 Cloning, overexpression, and purification of FDHs and CA

The genes of fdh and ca with His-tag were PCR-amplified and ligated with pET-32a(+) and pACYCDuet-1 by restriction enzymes, respectively (Table S1†). Recombinant plasmids (pET32a(+)- scfdh/tsfdh/atfdh/mtfdh and pACYCDuet-ca) were then transformed into the competent E. coli BL21 (DE3). Expression and purification of the target proteins in recombinant strains were performed following reported methods.^22,23 His-tagged proteins were purified using Ni-NTA resin, and protein concentration was determined via the BCA assay. SDS-PAGE was used to assess protein expression and molecular weight, with gels stained with Coomassie Brilliant Blue R-250.

2.3 Activities assay of FDHs

The enzymatic activities of FDHs were assayed by measuring the changes in the absorbance of NADH at 340 nm. The standard curve of NADH is shown in Fig. S1,† characterizing the OD₃₄₀ values corresponding to different concentrations of NADH. The oxidative activity was defined as the amount of enzyme required to produce NADH, and the reductive activity was defined as the amount of enzyme required to consume NADH. To calculate the specific activity of formate oxidation, 200 mM sodium formate and 2 mM NAD⁺ were used as the substrate and electron acceptor at 30 °C with different pH values (5–8). The reduction assay was conducted in 50 mM sodium bicarbonate and 0.3 mM NADH at 30 °C with different pH values (5–8). For the optimal carbon source for FDH reduction, 200 mM sodium bicarbonate and purged CO₂ (100%) were used as substrates, respectively.

2.4 Construction of recombinant S. cerevisiae with FDHs and CA

After verifying the activity of FDHs and CA in the E. coli system, the selected formate dehydrogenase (TsFDH) was transformed into S. cerevisiae S288C. The PCR amplification, enzyme ligation, and other plasmid construction methods are described in Section 2.2 (Table S2†). Yeast transformation was performed using the standard lithium acetate transformation method.²⁴S. cerevisiae S288C trpΔ with pRS424 (S. cerevisiae-control, NC), S. cerevisiae-tsfdh, S. cerevisiae-scfdh, S. cerevisiae-tsfdh-ca and S. cerevisiae-tsfdh-ca-bica were constructed.

2.5 Fermentation experiments and analytical methods

The recombinant S. cerevisiae was activated and cultured in SD-trp medium (20 g L⁻¹ glucose, 6.7 g L⁻¹ YNB, and appropriate amino acids but without tryptophan) at 30 °C and 180 rpm. Then the cells were harvested and transferred to the fermentation medium according to the experimental needs. Cell growth was measured by measuring the optical density at 620 nm (OD_{620 nm}). Glucose and ethanol concentrations were measured using an HPLC system (Acchrom S6000) equipped with a BIO-RAD Aminex HPX-87H column and reflective index detector at 50 °C. Formate concentration was measured using a UV detector at 210 nm. The mobile phase was 0.01 N H₂SO₄ with a flow rate of 0.5 mL min⁻¹. All experiments were replicated three times, and the error bars denote the standard deviation. In this study, P < 0.05 was considered as a significant difference and P < 0.001 was a highly significant difference.

2.6 Off-gas analysis

Parallel bioreactors from T&J Bioscience Co., Ltd (Shanghai, China) were used to analyze the ratio between released CO₂ and produced ethanol r(C/E) of S. cerevisiae-tsfdh-ca and S. cerevisiae-control, with each connected to an off-gas analyzer to record CO₂ emissions. Fermentations were performed in SC media containing 40 g L⁻¹ glucose, with an initial inoculum of 1 OD₆₂₀. The process was conducted at 30 °C and 180 rpm, with aeration maintained at 0.5 vvm.

2.7 Transcriptome analysis

In order to explore the beneficial effects of carbon dioxide fixation on fermentation, transcriptome analysis was performed. Total RNA was extracted from the tissue using TRIzol® reagent according to the manufacturer's instructions (Invitrogen) and genomic DNA was removed using DNase I (TaKara). Then RNA quality was determined using a 2100 Bioanalyser (Agilent) and quantified using an ND-2000 (NanoDrop Technologies). RNA-seq transcriptome libraries were prepared using a TruSeq™ RNA sample preparation kit from Illumina (San Diego, CA), using 1 μg of total RNA. Paired-end libraries were sequenced by Illumina NovaSeq 6000 sequencing (150 bp × 2, Shanghai BIOZERON Co., Ltd). To remove any low-quality sequences, we assessed reads in FASTQC and filtered clean reads using Trimmomatic. The reads were then mapped to the S. cerevisiae S288C genome (NCBI accession no. PRJNA43747). TPM analysis was carried out using String Tie, and differential expression analysis was performed using the DESeq R package (1.18.0), where we assigned p ≤ 0.05 and log2|FoldChange| ≥ 1 as differentially expressed genes. All analyses had three biological replicates.

3 Results and discussion

3.1 Exploration and characterization of efficient FDHs for CO₂ reduction

FDHs are capable of reversibly interconverting CO₂ and formate (Fig. 1a), normally classified into two types: metal-dependent and metal-independent FDHs.^25,26 Metal-dependent FDHs generally exhibit higher activity toward CO₂ reduction than the metal-independent FDHs, but the oxygen-sensitive catalytic components limit the metal-dependent FDHs for industrial application.²⁷ In light of this, we selected previously reported metal-independent FDH candidates to explore FDHs with high CO₂ reduction activity to facilitate ethanol yield. Sequence comparison revealed that the absolute homology of metal-independent FDHs was more significant than 80%, with bacterial FDHs containing an N-terminal loop and C-terminal loop, which were more stable than eukaryotic derived FDHs (Fig. S2†). Indeed, after evaluating the catalytic turnover rate (k_cat), binding affinity (K_m) and the optimal pH (Table S3†) considering the reduction ability of CO₂, we selected four promising FDHs (ScFDH, TsFDH, AtFDH, and MtFDH) for further study. The SDS-PAGE gel exhibited the FDHs expressed correctly with a high purity (Fig. S3†).

Fig. 1 (a) CO₂ reduction and formate oxidation activity of FDHs. (b) 200 mM sodium bicarbonate and purged CO₂ (100%) were used as substrates, respectively. Enzyme activities of FDHs for (c) formate oxidation and (d) CO₂ reduction at different pH (5, 6, 7, and 8) values. The temperature was controlled at 30 °C. ScFDH: FDH from S. cerevisiae, TsFDH: FDH from Thiobacillus sp. KNK65MA, AtFDH: FDH from A. thaliana, and MtFDH: FDH from M. thermophila.

For the purpose of elucidating the effect of different CO₂ sources on the reduction and ethanol yield, purged CO₂ (100%) or 200 mM sodium bicarbonate was used as the substrate, respectively. As presented in Fig. 1b, the four FDHs exhibited the same trend and the reducing activity of FDHs was significantly higher with CO₂ (100%) as the substrate than sodium bicarbonate, supporting the discovery of Popov et al.²⁸ In particular, TsFDH has the highest CO₂ reduction activity and formate production, and the reducing activity was 0.46 U mg⁻¹ enzyme and 0.40 U mg⁻¹ enzyme when CO₂ and sodium bicarbonate were used as a substrate. Consistent with previous findings in electrochemistry and biocatalysis, the reaction rate of formate production by CbFDH increased in proportion to the concentration of CO₂ in the reaction solution.^29,30 However, CO₂ has limited solubility in water and is not easily manipulated, while sodium bicarbonate can slowly release CO₂. In subsequent in vitro experiments, we therefore chose sodium bicarbonate as the source of CO₂.

Reaction conditions are crucial for enzyme activity, as metal-independent FDH reportedly exhibited optimal reductive activity under acidic or neutral conditions. Therefore, considering the potential application in the yeast, we examined the reductive capacity of FDHs within the pH range of 5 to 8.³¹ The oxidative and reducing activities of the four FDHs at different pH values are shown in Fig. 1c and d, respectively. As previously reported, all the FDHs were able to successfully catalyze both the oxidation of formate and the reduction of CO₂. Additionally, they exhibited strong oxidizing abilities and relatively weak reducing abilities.²² The oxidative specific enzyme activities of the four FDHs ranged from 11.96 to 26.40 U mg⁻¹ (Table S4†) and the reducing specific enzyme activity ranged from 0.09 to 0.53 U mg⁻¹ (Table S4†). Promisingly, TsFDH exhibited outstanding CO₂ reduction activity regardless of the pH conditions. As shown in Fig. 1d, TsFDH exhibited the highest enzyme activity in terms of the reduction ratio at pH 6, reaching 0.53 U mg⁻¹. This activity was 2.1-fold higher than that of AtFDH, which was the enzyme with the second-highest activity. The effect of pH variation on reductase activity was non-proportional, and the activity of TsFDH in a mildly acidic environment (pH 5) was decreased but still statistically higher than the other FDHs. Alternatively, the reductase activity of ScFDH showed a trend consistent with that observed in the oxidation process. As the pH value increased, the reductase activity of MtFDH gradually decreased, dropping from 0.14 U mg⁻¹ to 0.09 U mg⁻¹. Although MtFDH demonstrated robust oxidative activity, its capacity to reduce CO₂ was rather weak, particularly at pH 5.

After considering the facts that TsFDH exhibited 4.32-fold higher reduction specific enzyme activity compared with ScFDH during in vitro enzyme assays, TsFDH has the potential as CO₂ reductase for ethanol production integrated with CO₂ fixation under acidic conditions, which might facilitate more carbon recycling subsequently.

3.2 TsFDH serves as a key mediator in recycling CO₂ obtained from ethanol production

Ethanol fermentation and formate production of the TsFDH overexpression strains S. cerevisiae-tsfdh (S. cerevisiae-scfdh as a control) were investigated. As was easily seen, TsFDH exhibited excellent CO₂ fixation capacity compared to S. cerevisiae's native FDH. As shown in Fig. 2b, formate produced by S. cerevisiae-tsfdh increased by 39.0%, from 22.3 to 31.0 mg L⁻¹ compared with S. cerevisiae-scfdh. Meanwhile, S. cerevisiae-tsfdh achieved better glucose fermentation performance in conjunction with ethanol productivity. In Fig. 2a, the glucose consumption rate of S. cerevisiae-tsfdh was significantly superior compared with S. cerevisiae-scfdh, increased by 20.5% from 1.27 g L⁻¹ h⁻¹ to 1.53 g L⁻¹ h⁻¹ within 36 h and produced 15.11 g L⁻¹ of ethanol, with a 6.0% increase in ethanol production. It proved that TsFDH exhibited more advanced CO₂ reduction activity than ScFDH, accompanied by improved glucose fermentation and ethanol yield.

Fig. 2 Fermentation properties of S. cerevisiae-tsfdh (TsFDH) and S. cerevisiae-scfdh (ScFDH) investigated at 30 °C and 180 rpm in SD media. (a–c) 40 g L⁻¹ glucose, initial inoculum was 1 OD₆₂₀; (d–f) 60 g L⁻¹ glucose, initial inoculum was 3 OD₆₂₀.

When initial inoculum and glucose concentration increased to 3 OD_{620 nm} and 60 g L⁻¹, formate and glucose consumption rates were both boosted (Fig. 2d–f). In the illustration of our previous work, a higher initial cell biomass meant that more target enzymes were added and allowed more FDH to fix CO₂ in a timely manner and reduce its overflow.¹⁵ Accordingly, more formate was produced for more FDH and more CO₂ from ethanol fermentation with initial inoculum 3 OD and a higher glucose concentration. Formate production of S. cerevisiae-tsfdh increased by 37.6%, from 46.3 to 63.7 mg L⁻¹. Beyond that, efficient CO₂ fixation also facilitated glucose fermentation, and more inoculation resulted in upgraded fermentation performance as shown in Fig. 2d. S. cerevisiae-tsfdh almost depleted 60 g L⁻¹ glucose at 24 h, with the production of 22.83 g L⁻¹ ethanol, while S. cerevisiae-scfdh only produced 19.29 g L⁻¹ ethanol. The glucose consumption rate increased by 14.6% from 2.12 g L⁻¹ h⁻¹ to 2.43 g L⁻¹ h⁻¹. The fermentation results showed that TsFDH could fix CO₂ released during ethanol fermentation in situ more efficiently, and the fixation of CO₂ could promote the glucose fermentation process and improve the productivity of ethanol. The promotion effect was positively correlated with the enzyme activity or enzyme dose.

Glycerol is a major by-product in ethanol production because yeast generates an excess of NADH during fermentation, and the production of glycerol helps to maintain the intracellular redox balance.³² In the present study, glycerol production of S. cerevisiae-tsfdh was noticed to decrease by 26.7% from 0.95 g L⁻¹ to 0.75 g L⁻¹ compared with S. cerevisiae-scfdh. This indicated that NADH-dependent TsFDH oxidized the excess NADH while fixing CO₂ produced from ethanol fermentation in situ, thereby reducing glycerol production and increasing the ethanol yield. Mitochondrial NADH dehydrogenase (NDE) is responsible for catalyzing the oxidation of NADH in the cytoplasm. It was reported that overexpressing NDE1/2 decreased the intracellular NADH/NAD⁺ ratio and glycerol yield.³³ Here, TsFDH fixed CO₂ requiring NADH, so that glycerol production was no longer needed to consume excess NADH for redox balance. Consequently, more CO₂-derived formate means less glycerol production, indicating that fixing CO₂ from ethanol production reduced the formation of the by-product and further improved carbon recovery.

Various applications highlight FDH-driven electroenzymatic and photoenzymatic CO₂ reduction systems in vitro, in which a whole-cell biocatalytic strategy to utilize CO₂ as a sustainable feedstock has also been attempted. Whole-cell catalysis has received widespread attention as it avoids enzyme purification, and provides optimum conditions for the enzyme and facilitates its maintenance. Alissandratos et al. overexpressed FDH from Pyrococcus furiosus (PfFDH) in E. coli JM109 for whole-cell catalysis of formate production, and this engineered strain produced 44 mM formate from sodium bicarbonate and gaseous hydrogen with a productivity of 22.7 mM h⁻¹.³⁴ Schwarz and Müller recently implemented efficient whole-cell biocatalytic CO₂ hydrogenation to formate by the TkFDH from Themoanaerobacter kivui with a yield of 130 mM and a productivity of 270 mM h⁻¹.³⁵ Although enzymatic synthesis of formate from CO₂ with electrons supplied from hydrogenase-driven H₂ oxidation exhibits many advantages, these reactions require highly purified and expensive hydrogen gas, leading to an increase in storage and transportation costs, as well as safety issues. In this work, glucose fermentation for ethanol production provided cofactor NADH for CO₂ fixation, which established a win–win strategy for ethanol production accompanied by CO₂ fixation with environmentally and economically beneficial effects. It not only reduced carbon emissions, but also provided insight into CO₂ conversion to valuable chemicals and fuels by whole-cell biocatalysis.

3.3 Harnessing FDH and CA powered CO₂ reduction

For whole-cell catalysis, the low solubility of CO₂ in water presents a significant challenge to CO₂ fixation. CA was reported to promote the hydration reaction of CO₂, so we first investigated the effect of different concentrations of purified CA on CO₂ reduction.²⁰ The fermentation performance of S. cerevisiae-tsfdh (TsFDH) with different concentrations of CA (0, 0.1, and 0.5 g L⁻¹) is shown in Fig. S4,† and both glucose consumption and ethanol yield increased after CA addition. Glucose consumption rates were 1.49, 1.53, and 1.69 g L⁻¹ h⁻¹, which increased by 2.7% and 13.4% compared with the control, respectively. Meanwhile, the ethanol yields were 0.36, 0.37, and 0.39 g g⁻¹, which increased by 2.8% and 8.3%, respectively. Obviously, S. cerevisiae-tsfdh had a higher glucose consumption rate and ethanol yield as enough CA was added. There were almost no differences in formate and glycerol yields with CA addition, perhaps because the enzyme was unstable and inactive in an acidic and ion-rich environment. As shown in Fig. S5,† the overexpression of TsFDH and CA individually resulted in a 51.1% and 50.3% increase in glucose fermentation rates, respectively, and a 71.3% and 54% increase in formate production. Overexpression of TsFDH increased the concentration of FDH, thereby enhancing glucose utilization, whereas overexpression of CA alone significantly improved the CO₂ recovery efficiency mediated by the endogenous FDH in S. cerevisiae. These results highlighted the critical roles both enzymes play in carbon recycling.

To further the research, a whole-cell biocatalysis system with FDH and CA harness was constructed to enable continuous production of FDH and CA and to maintain a relatively stable pH and environment that was more favorable for CO₂ capture. Formate production and ethanol fermentation of S. cerevisiae-tsfdh and S. cerevisiae-tsfdh-ca at different glucose concentrations were examined. As shown in Fig. 3, S. cerevisiae-tsfdh and S. cerevisiae-tsfdh-ca produced 26.7 mg L⁻¹, 49.7 mg L⁻¹, 63.3 mg L⁻¹ and 27.7 mg L⁻¹, 56.3 mg L⁻¹, and 76.3 mg L⁻¹ formate, respectively, while the glucose concentrations were 20 g L⁻¹, 40 g L⁻¹, and 60 g L⁻¹. S. cerevisiae-tsfdh-ca produced more formate at the same glucose concentration, and the trend became more pronounced with a higher sugar concentration. Dual enzymes of TsFDH and CA made great effort on the CO₂ hydration reaction and reduction. Consistent with this, S. cerevisiae-tsfdh-ca exhibited superior glucose metabolism, specifically a higher ethanol yield and a faster glucose consumption rate. As shown in Table 1, the glucose consumption rate of S. cerevisiae-tsfdh-ca was 1.15 g L⁻¹ h⁻¹ when 20 g L⁻¹ glucose was used, which was a 15.8% increase compared to S. cerevisiae-tsfdh. At 40 g L⁻¹ glucose, S. cerevisiae-tsfdh and S. cerevisiae-tsfdh-ca produced 12.12 g L⁻¹ and 14.18 g L⁻¹ of ethanol in 48 h. The yield increased from 0.31 g g⁻¹ to 0.34 g g⁻¹, with an increase of 9.68%. In addition, the glucose consumption rate of S. cerevisiae-tsfdh-ca increased from 1.10 g L⁻¹ h⁻¹ to 1.30 g L⁻¹ h⁻¹, an increase of 15.38%, compared to S. cerevisiae-tsfdh. Similar results were revealed at 60 g L⁻¹ glucose, S. cerevisiae-tsfdh and S. cerevisiae-tsfdh-ca with ethanol yields of 0.31 g g⁻¹ and 0.34 g g⁻¹, and glucose consumption rates of 1.32 g L⁻¹ h⁻¹ and 1.43 g L⁻¹ h⁻¹, an increase of 8.33%. The above results indicated that the co-expression of FDH and CA had meaningful and long-term effects on both CO₂ reduction and glucose fermentation.

Fig. 3 Fermentation properties of S. cerevisiae-tsfdh (TsFDH) and S. cerevisiae-tsfdh-ca (TsFDH + CA) at different glucose concentrations investigated at 30 °C and 180 rpm in SC media. (a) 20 g L⁻¹ glucose, (b) 40 g L⁻¹ glucose, and (c) 60 g L⁻¹ glucose.

Table 1 Fermentation parameters of S. cerevisiae-tsfdh (TsFDH) and S. cerevisiae-tsfdh-ca (TsFDH + CA)^a

Strain	Glucose (g L^−1)	Formate (mg L^−1)	Glucose consumption rate (g L^−1 h^−1)
a Fermentation was conducted in SC media using glucose as the carbon source.
TsFDH	20.96	26.67 ± 3.06	0.996 ± 0.022
	41.22	49.67 ± 4.04	1.100 ± 0.039
	60.69	63.33 ± 3.22	1.323 ± 0.060
TsFDH + CA	20.75	27.67 ± 4.04	1.153 ± 0.009
	41.38	56.33 ± 5.69	1.295 ± 0.005
	60.21	76.33 ± 10.97	1.428 ± 0.087

---

Formate is a useful source of microbial energy because it can be bio-electrochemically produced with high solar-to-product energy conversion efficiency and is easy to store and transport. Compared to soluble formate, the kinetic limitations of gaseous CO₂ supply result in lower intracellular CO₂ concentrations, reducing the CO₂ fixation rate. It is anticipated that the production of formate itself from CO₂ can be enhanced by using CA as previously mentioned. CA is attributed with the most rapid turnover frequency related to converting CO₂ to bicarbonate, speeding up metabolic processes involving dissolved CO₂ efficiently, and helping the pH balance through its buffering action. Integrating CA with microorganisms enables the efficient production of a broader range of chemicals, as metabolic pathway engineering expands product diversity, while CA-driven CO₂ hydration supplies more CO₂ for various conversion processes. Photoautotrophic microorganisms, like microalgae and cyanobacteria, offer valuable platforms for metabolic engineering, as they can sustainably produce target compounds such as carotenoids and lipids directly from CO₂ using light energy. It has been demonstrated that CA immobilized on microalgae cell surface increased its growth by 1.6-fold and lipid production by 1.7-fold without the need for the addition of other carbon sources.³⁶ In the context of CO₂ reduction, the synergy between CA and FDH has been found to enhance the reaction rate. As a result, the cascade system involving FDH and CA has attracted widespread attention, due to its potential implications in various industrial and environmental applications. Sato et al.³⁷ constructed a dual-enzyme system using FDH and CA, demonstrating that the use of CA to convert CO₂ into bicarbonate and the use of FDH to reduce CO₂ to formate in CCUS systems are coordinated. However, the activity and stability of enzymes in vitro present challenges, and there are issues with poor synergy when coupled with cofactor regeneration. Constructing microbial cell factories can provide a stable environment for enzymes, and the diverse endogenous or exogenous metabolic pathways in microorganisms can supply cofactors. Moreover, designing and optimizing metabolic pathways can also synthesize higher value-added products. In this study, we constructed a cellular factory that co-expresses FDH and CA to in situ fix the byproduct CO₂ from ethanol fermentation into formate, with glucose fermentation providing the cofactors. Formate production increased by 20.52%. Adding formate as a promising carbon-energy coupling carrier for bioproduction provides a new idea for the resourceful utilization of CO₂.

3.4 S. cerevisiae-tsfdh-ca exhibited the fastest fermentation rate and the highest formate production

Cyanobacteria possess an environmental adaptation known as a CO₂ concentrating mechanism (CCM) that evolved to improve photosynthetic performance, particularly under CO₂-limiting conditions, which contains inorganic carbon transporters that actively transport extracellular CO₂ and HCO₃⁻ into the cell.³⁸ To increase the intracellular CO₂ concentration, we attempted to overexpress the HCO₃⁻ transporter (BicA) from cyanobacteria in S. cerevisiae, and the S. cerevisiae-tsfdh-ca-bica strain was successfully constructed and the fermentation parameters are shown in Fig. 4. There was no significant difference in both glucose metabolism and formate production of S. cerevisiae-tsfdh-ca and S. cerevisiae-tsfdh-ca-bica. It was possible because the complex expression pattern of membrane proteins failed to enhance the affinity between distant species such as cyanobacteria and yeast. What's more, the expression of multiple heterologous genes brought a specific metabolic burden, which affected the fermentation ability of the strain. It needs to be stressed again that the overexpression of TsFDH and CA exhibited significant advantages in terms of glucose consumption rate and formate production capacity compared to S. cerevisiae-control. S. cerevisiae-tsfdh-ca exhibited the fastest fermentation rate and the highest formate production. Compared with S. cerevisiae-control, the glucose consumption rate, ethanol production, and formate production of S. cerevisiae-tsfdh-ca were 69%, 83%, and 108% higher, respectively. There was no doubt that the TsFDH screened in this study had strong power to simultaneously fix CO₂in situ during ethanol production and further achieve the conversion of CO₂ to formate by enhancing the rate of CO₂ hydration by CA. Considering that high concentrations of formate are harmful to cells, future work will focus on converting formate to less toxic methanol by formaldehyde dehydrogenase (FaldDH) and alcohol dehydrogenase (ADH).³⁹

Fig. 4 Fermentation properties of the S. cerevisiae-control, S. cerevisiae-tsfdh (TsFDH), S. cerevisiae-tsfdh-ca (TsFDH + CA) and S. cerevisiae-tsfdh-ca-bica (TsFDH + CA + BicA) investigated at 30 °C and 180 rpm in SC media. (a) Glucose and ethanol concentrations, (b) formate concentrations, and (c) glucose consumption rate.

3.5 Reducing CO₂ leakage during glucose fermentation

To verify the recovery of CO₂ during ethanol fermentation, a tail gas analyzer monitored the CO₂ released during the fermentation process in real time (Fig. S6†). Both strains exhibited a rapid increase in CO₂ production at the start of the fermentation process as shown in Fig. 5a, the green line reached its peak CO₂ emission faster and higher than the yellow line, indicating that the S. cerevisiae-tsfdh-ca strain had a stronger glucose metabolic performance during the initial fermentation phase. After reaching the peak, both strains showed a decline in CO₂ production, but consistent with the trend of formate production, the CO₂ of S. cerevisiae-tsfdh-ca dropped off more rapidly than that of the control. The statement highlighted the outstanding CO₂ reduction capability of S. cerevisiae-tsfdh-ca, which can promptly reduce the CO₂ produced by yeast cells back into formate, minimizing carbon loss. Throughout the entire fermentation, the S. cerevisiae-tsfdh-ca strain reduced CO₂ emissions by about 3.12% compared to the control strain.

Fig. 5 (a) Real-time measurement of CO₂ release during fermentation; (b) r(C/E, CO₂ released/ethanol produced) of S. cerevisiae-tsfdh-ca (Ts + CA), S. cerevisiae-tsfdh (Ts), S. cerevisiae-ca (CA) and the control (NC).

A new parameter, r(C/E), was also introduced to represent the ratio between the amount of CO₂ released and ethanol produced in the fermentation system. A lower r(C/E) indicated less CO₂ released per ethanol unit. During the initial 12 hours, both yeast strains were in the rapid growth phase. At this point, the r(C/E) values for S. cerevisiae-tsfdh-ca and the control strain were 1.13 and 1.30, respectively. S. cerevisiae-tsfdh-ca displayed superior glucose fermentation performance and CO₂ fixation capability. From 12 and 36 hours, both strains transitioned to the main fermentation phase, where ethanol production and CO₂ emission were most significant. At this stage, the r(C/E) values increased to 1.74 for S. cerevisiae-tsfdh-ca and 2.20 for the control strain. This shift underscores the enhanced CO₂ reduction capability of S. cerevisiae-tsfdh-ca. With elevated FDH and CA expression, the strain effectively converted the CO₂ into formate, significantly reducing carbon loss. When fermentation neared completion, S. cerevisiae-tsfdh-ca finished earlier than the control strain, which continued to release CO₂. Considering the entire fermentation phase, as shown in Fig. 5b, the r(C/E) of S. cerevisiae-tsfdh-ca (co-expressing TsFDH and CA), S. cerevisiae-tsfdh (expressing only TsFDH), S. cerevisiae-ca (expressing only CA), and the control (NC) were 1.39, 1.50, 1.70, and 1.89, respectively. Statistical analysis also showed a significant difference between the recombinant strains and the control strain. The lower r(C/E) values indicate that the recombinant strains lost less carbon in the form of CO₂ during fermentation. These r(C/E) data further support the conclusions previously. Overexpression of TsFDH increased FDH concentration, thereby enhancing carbon fixation. In contrast, overexpression of CA alone significantly improved CO₂ recovery efficiency mediated by endogenous FDH in S. cerevisiae. Furthermore, the synergistic effect of TsFDH and CA further improved carbon recovery efficiency beyond the individual contributions of each enzyme.

Due to its carbon-reducing properties, the recovery of CO₂ from ethanol production has been spotlighted. The pioneering research on CO₂ fixation in ethanol fermentation was realized by introducing heterologous phosphoribulokinase (PRK) and ribulose-1,5-bisphosphate carboxylase (RuBisCO) in S. cerevisiae to reconstruct the pentose phosphate pathway (PPP).⁴⁰ The primary role of CO₂ in this pathway was as an electron acceptor and CO₂ was incorporated as a co-substrate in the fermentation with a sparging gas (10% CO₂ and 90% N₂). The ethanol production was increased by 10%, but the efficiency of CO₂ fixation was poor. Cell growth and fermentation efficiency were negatively affected, and the fermentation time was delayed from nearly 40 h to nearly 50 h. Xia et al. achieved a phase coupling of PPP and xylose metabolic pathway in S. cerevisiae with a 3.7% increase in ethanol yield.⁴¹ The primary role of CO₂ in this pathway was also as an electron acceptor, meanwhile, the imbalance of cofactor utilization of NADPH and NAD⁺ by xylose reductase (XR)/xylitol dehydrogenase (XDH) in xylose metabolism affected the CO₂ fixation efficiency. Finally, there was a severe inhibition of glucose on xylose metabolism. This limitation allowed the PPP pathway to fix CO₂ only at low glucose concentrations (20 g L⁻¹), making industrial-scale ethanol production infeasible. The above studies achieved CO₂ fixation during ethanol fermentation by constructing a heterologous PPP, but apparently the performance was not satisfactory. In contrast, the co-expression system of FDH and CA in this study overcame the problems containing energy imbalance, low enzyme stability and glucose inhibition. The r(C/E) value significantly decreased from 1.89 to 1.39, indicating the effective reduction of CO₂. It achieved a win–win result of CO₂ fixation and ethanol production simultaneously under a wider range of fermentation conditions, which showed better practicality.

3.6 Mechanism of enhanced fermentation efficiency by FDH and CA

To reveal the mechanism by which the CO₂ reduction process catalyzed by FDH and CA enhanced fermentation efficiency, RNA-seq analysis was performed. Differential expressed genes (DEG) analysis of S. cerevisiae-tsfdh-ca and the control strain was shown in Fig. 6a. DEGs involved in the glycolytic pathway were down-regulated, such as those encoding glyceraldehyde-3-phosphate dehydrogenase (GAP), phosphoglycerate mutase (GPM), and enolase (ENO). This indicated that S. cerevisiae had already consumed most of the glucose, gradually shifting its metabolic flux from primary metabolism (glucose catabolism for energy production) to secondary metabolism. The downregulation of GPP1 and GPP2 encoding glycerol-3-phosphatase (GPP), suggested the decrease of glycerol production. In S. cerevisiae, glycerol synthesis typically served as a secondary metabolic pathway to help the cell balance redox states under anaerobic conditions or when excess NADH needed to be re-oxidized, which was consistent with the conjecture in Section 3.3 that producing more formate rather than glycerol was related to the competitive production of NADH. Previous reports proved that introducing the PRK-RuBisCO pathway or overexpressed A-ALD can re-oxidize NADH to reduce glycerol production, consisting of oxidizing excess NADH by FDHs for CO₂ fixation can effectively decrease glycerol production and improve ethanol yield in this study.^40–42 The up-regulation of ADH4, encoding alcohol dehydrogenase, supported the increased ethanol yield as well.

Fig. 6 (a) Differentially expressed gene (DEG) analysis of central carbon metabolism; (b) oxidative phosphorylation up-regulated gene; and (c) oxidative phosphorylation gene set enrichment analysis (GSEA).

The tricarboxylic acid (TCA) cycle also showed a notable upregulation trend, including IDP2 and SDH3 encoding isocitrate dehydrogenase and succinate dehydrogenase. The TCA cycle is one of the important energy-producing pathways. The NADH and FADH2 generated by the TCA cycle are important substrates in the process of oxidative phosphorylation. In the case of S. cerevisiae-tsfdh-ca, more formate is produced through CO₂ fixation. Formate has been reported to be harmful to the electron transport chain (ETC).⁴³ However, SDH3, which participated in FADH₂ oxidation and linked the TCA cycle to the ETC, was significantly upregulated in Fig. 6b. This suggested that the enhanced CO₂ fixation process by FDH and CA not only increased energy production but also modulated the TCA cycle and oxidative phosphorylation pathways. The upregulation of SDH3 may play a role in maintaining the integrity and efficiency of the ETC, thereby counteracting the potential negative effects of formate on the ETC. This regulatory mechanism could be crucial for optimizing glucose fermentation and ethanol yield in the presence of increased formate production. Additionally, PMA2 encoded a proton pump that was upregulated, which helps maintain the proton gradient required for ATP synthesis across the mitochondrial membrane. NDE2 encoding the mitochondrial NADH dehydrogenase (NDE), which catalyzed the oxidation of cytosolic NADH was upregulated. Theoretically, NADH is primarily generated from NAD⁺ through the catabolism of glucose, serving as the reducing equivalent for various redox reactions.^44,45 This excess NADH could be oxidized by external NDE, which was localized on the outer surface of the inner mitochondrial membrane, directly oxidized cytosolic NADH and shuttled the electrons to the quinone pool of the respiratory chain.⁴⁶ This pathway operated independently of the TCA cycle but relied on the capacity of the ETC. It has been proven in various microorganisms that enhancing NADH oxidation can effectively improve glucose metabolism rate and target metabolite production intensity.^47–49 Here, the upregulation of multiple genes in the TCA pathway and oxidative phosphorylation pathway indicates that the enhanced CO₂ fixation process by FDH and CA has increased energy production. Moreover, FDH-mediated NADH oxidation has a similar effect to NDE, thereby improving glucose fermentation and ethanol yield.

The enrichment analysis of the gene set (GSEA) involved in the oxidative phosphorylation process was then performed (Fig. 6c). The oxidative phosphorylation gene set exhibited an overall upregulation trend, releasing more energy to enhance glucose fermentation and ethanol yield in S. cerevisiae-tsfdh-ca. In summary, the upregulated DEGs involved in oxidative phosphorylation suggested an increase in ATP production. Enhanced activity in the TCA cycle (especially through SDH3) fed more electron donors (NADH and FADH₂) into the oxidative phosphorylation pathway, maximizing energy output and supporting cellular metabolic demands. This coordination between the TCA cycle and oxidative phosphorylation enabled efficient energy production, crucial for high-energy-demand processes. This work explained the mechanism of CO₂ fixation contributing to the glucose consumption rate in S. cerevisiae-tsfdh-ca. This regulatory mechanism could be crucial for optimizing biomanufacturing in the presence of increased formate derived from CO₂ reduction.

4 Conclusions

Since the co-expression of FDH and CA addressed the issues of low solubility and low reduction efficiency of CO₂, the dilemma of CO₂ emissions during ethanol production is alleviated. As a win–win strategy, glucose fermentation and formate production of S. cerevisiae-tsfdh-ca increased by 69% and 108%, respectively, and reduced CO₂ emissions by approximately 3.12%. Furthermore, the mechanism of the advantageous effect on fermentation performance by harnessing FDH and CA was revealed. The in situ CO₂ fixation mediated by them enhanced the TCA cycle and oxidative phosphorylation processes, thereby counteracting the potential negative effects of formate. The FDH-mediated oxidation of NADH for CO₂ reduction has a function similar to NDE, leading to an improved ethanol yield. This approach offers a dual benefit, which includes reducing CO₂ emissions and supplying carbon-based fuels and chemicals. These dual benefits have a substantial impact on ethanol production and other renewable fuels and chemicals.

Author contributions

Ying He: methodology, data curation, and writing. Yimin Li: conceptualization and writing – review. Liming Su: validation and methodology. Jiaxin Liu: investigation and methodology. Cong Du: conceptualization and writing – review. Wenjie Yuan: conceptualization, data curation, and writing – review.

Data availability

The data supporting this article have been included as part of the ESI.†

Conflicts of 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.

Acknowledgements

This study was supported by the National Key Research and Development Program of China (2021YFC2101305) and the National Natural Science Foundation of China (22408041). The authors acknowledge the assistance of DUT Core Facilities of the School of Bioengineering.

References

1. M. Kumar, A. Gupta and I. S. Thakur, Bioresour. Technol., 2016, 213, 249–256.
2. A. S. Alothaim, J. King Saud Univ., Sci., 2023, 35, 102939.
3. D. Tilman, R. Socolow, J. A. Foley, J. Hill, E. Larson, L. Lynd, S. Pacala, J. Reilly, T. Searchinger, C. Somerville and R. Williams, Science, 2009, 325, 270–271.
4. IEA – International Energy Agency, https://www.iea.org, (accessed March 22, 2023).
5. A. Nisar, S. Khan, M. Hameed, A. Nisar, H. Ahmad and S. A. Mehmood, Microbiol. Res., 2021, 251, 126813.
6. M. Orlić, C. Hochenauer, R. Nagpal and V. Subotić, Energy Convers. Manage., 2024, 314, 118601.
7. S. Jin, Z. Hao, K. Zhang, Z. Yan and J. Chen, Angew. Chem., Int. Ed., 2021, 60, 20627–20648.
8. J. Tian, W. Deng, Z. Zhang, J. Xu, G. Yang, G. Zhao, S. Yang, W. Jiang and Y. Gu, Nat. Commun., 2023, 14, 7758.
9. W. Jiang, D. Hernández Villamor, H. Peng, J. Chen, L. Liu, V. Haritos and R. Ledesma-Amaro, Nat. Chem. Biol., 2021, 17, 845–855.
10. Y. Chen, P. Li, H. Noh, C.-W. Kung, C. T. Buru, X. Wang, X. Zhang and O. K. Farha, Angew. Chem., Int. Ed., 2019, 58, 7682–7686.
11. Y. Lu, Z.-Y. Jiang, S.-W. Xu and H. Wu, Catal. Today, 2006, 115, 263–268.
12. J. Luo, A. S. Meyer, R. V. Mateiu and M. Pinelo, New Biotechnol., 2015, 32, 319–327.
13. A. Ünlü, Z. E. Duman-Özdamar, B. Çaloğlu and B. Binay, Protein J., 2021, 40, 489–503.
14. Y. J. Li, M. M. Wang, Y. W. Chen, M. Wang, L. H. Fan and T. W. Tan, Sci. Rep., 2017, 7, 43875.
15. C. Du, Y. Li, Y. He, L. Su, H. Wang, W. Yuan and F. Bai, Green Chem., 2022, 24, 6989–6999.
16. H. Rasouli, I. Iliuta, F. Bougie, A. Garnier and M. C. Iliuta, Chem. Eng. J., 2022, 432, 134029.
17. M. S. Mesbahuddin, A. Ganesan and S. Kalyaanamoorthy, Protein Eng., Des. Sel., 2021, 34, gzab021.
18. T. Sharma, S. Sharma, H. Kamyab and A. Kumar, J. Cleaner Prod., 2020, 247, 119138.
19. X. Ji, Z. Su, P. Wang, G. Ma and S. Zhang, ACS Nano, 2015, 9, 4600–4610.
20. S. Zaidi, N. Srivastava and S. Kumar Khare, Bioresour. Technol., 2022, 365, 128174.
21. X. Zhang, W. Shao, B. Chen and M. Wang, Enzyme Microb. Technol., 2021, 146, 109763.
22. K. Min, Y.-S. Park, G. W. Park, J.-P. Lee, M. Moon, C. H. Ko and J.-S. Lee, Bioresour. Technol., 2020, 305, 123155.
23. U. K. Laemmli, Nature, 1970, 227, 680–685.
24. D. Gietz and R. A. Woods, Yeast Gene Anal., 1998, 26, 53–66.
25. M. Moon, G. W. Park, J.-P. Lee, J.-S. Lee and K. Min, J. CO₂ Util., 2020, 42, 101353.
26. C. F. Nielsen, L. Lange and A. S. Meyer, Biotechnol. Adv., 2019, 37, 107408.
27. L. Calzadiaz-Ramirez and A. S. Meyer, Curr. Opin. Biotechnol., 2022, 73, 95–100.
28. V. O. Popov and V. S. Lamzin, Biochem. J., 1994, 301, 625–643.
29. R. Sato and Y. Amao, New J. Chem., 2020, 44, 11922–11926.
30. M. Meneghello, A. R. Oliveira, A. Jacq-Bailly, I. A. C. Pereira, C. Léger and V. Fourmond, Angew. Chem., Int. Ed., 2021, 60, 9964–9967.
31. G. Ÿzgün, N. G. Karagüler, O. Turunen, N. J. Turner and B. Binay, J. Mol. Catal. B: Enzym., 2015, 122, 212–217.
32. M. Persson, M. Galbe and O. Wallberg, Biotechnol. Biofuels, 2020, 13, 156.
33. J.-W. Kim, S.-O. Seo, G.-C. Zhang, Y.-S. Jin and J.-H. Seo, Bioresour. Technol., 2015, 191, 512–519.
34. A. Alissandratos, H.-K. Kim and C. J. Easton, Bioresour. Technol., 2014, 164, 7–11.
35. F. M. Schwarz and V. Müller, Biotechnol. Biofuels, 2020, 13, 32.
36. S. K. You, Y. J. Ko, S. K. Shin, D. H. Hwang, D. H. Kang, H. M. Park and S. O. Han, Bioresour. Technol., 2020, 318, 124072.
37. R. Sato and Y. Amao, React. Chem. Eng., 2022, 7, 181–191.
38. G. D. Price, Photosynth. Res., 2011, 109, 47–57.
39. Q. Liao, W. Liu and Z. Meng, Biotechnol. Adv., 2022, 60, 108024.
40. V. Guadalupe-Medina, H. W. Wisselink, M. A. H. Luttik, E. de Hulster, J.-M. Daran, J. T. Pronk and A. J. A. van Maris, Biotechnol. Biofuels, 2013, 6, 125.
41. P.-F. Xia, G.-C. Zhang, B. Walker, S.-O. Seo, S. Kwak, J.-J. Liu, H. Kim, D. R. Ort, S.-G. Wang and Y.-S. Jin, ACS Synth. Biol., 2017, 6, 276–283.
42. A. C. A. van Aalst, I. S. van der Meulen, M. L. A. Jansen, R. Mans and J. T. Pronk, Metab. Eng., 2023, 80, 151–162.
43. Z. Zhu, P. Geng, M. Li, H. Yin, C. Li, X. Cao and J. Tian, Chem. Eng. J., 2024, 488, 151182.
44. B. M. Bakker, K. M. Overkamp, A. J. A. van Maris, P. Kötter, M. A. H. Luttik, J. P. van Dijken and J. T. Pronk, FEMS Microbiol. Rev., 2001, 25, 15–37.
45. V. E. Balderas-Hernández, K. Correia and R. Mahadevan, J. Ind. Microbiol. Biotechnol., 2018, 45, 735–751.
46. P. J. Gommers, B. J. van Schie, J. P. van Dijken and J. G. Kuenen, Biotechnol. Bioeng., 1988, 32, 86–94.
47. Z. Dong, X. Li, L. Liu, G. Du and J. Chen, Acta Microbiol. Sin. (Engl. Transl.), 2008, 48, 1061–1066.
48. S. J. Bae, S. Kim, H. J. Park, J. Kim, H. Jin, B. G. Kim and J. S. Hahn, Metab. Eng., 2021, 66, 68–78.
49. Y. G. Lee, Y. S. Jin, Y. L. Cha and J. H. Seo, Bioresour. Technol., 2017, 228, 355–361.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5gc00705d

---