Due to the depletion of fossil resources, ever-growing energy demands, and the growing concern over carbon dioxide emissions, attempts to convert petroleum-based chemical manufacturing into carbon-neutral production from renewable resources are now in the limelight ^[1][2][3][1,2,3]. Biorefining with biomass as feedstocks has been actively pursued worldwide as a sustainable strategy not only to produce fuels but also to obtain various valuable chemicals [4] which. Most chemical compounds, which previously could only be produced by a chemical approach, can now be generated by biological processes from renewable resources by using enzymes or organisms.

2,3-butanediol (2,3-BDO) is a representative C4 platform chemical which can be produced from various biomass resources by biological routes. 2,3-BDO is also known as dimethylene glycol, 2,3-butylene glycol, and butane-2,3-diol (IUPAC name). At room temperature, 2,3-BDO is an odorless, colorless, transparent liquid. Three stereoisomers exist: a D-(−)- isomer and an L-(+)-isomer, which are optically active, and meso-2,3-BDO. The boiling points of 2,3-BDO range from 177 °C to 182 °C, being slightly different among the three stereoisomers [5].

Currently, 2,3-BDO is manufactured in very limited quantities via petrochemical routes (Figure 1). The petrochemical route for BDO synthesis is primarily the catalytic hydrolysis of butene oxide at a high temperature (160–220 °C) under high pressure (50 bar) and with a high energy consumption [6]. Furthermore, upon chemical production, the selective production of the pure stereoisomer is difficult to control due to the complicated reactions and extremely expensive separation and purification process.

The chemical production of 2,3-BDO has some drawbacks, and thus the biological process using microorganisms has been extensively investigated to obtain a high yield under mild conditions. The microbial production of 2,3-BDO has been investigated for over 100 years from the time Harden et al. first reported 2,3-BDO production by using Klebsiella pneumoniae in 1906 ^[7][11]. After that, Fulmer et al. reported the first industrial-scale production in 1933. Thereafter, numerous studies have been conducted to produce 2,3-BDO from various feedstock such as glucose, fructose, mannose, and galactose by several types of bacteria, yeasts or even algae ^[8][12]. However, the chemical production of 2,3-BDO using petroleum resources has gained more attention due to its easy availability and low manufacturing costs. Nonetheless, the biological production of 2,3-BDO from biomass resources has obvious environmental benefits and sustainability. Rehman et al. analyzed the footprint of bio-based 2,3-BDO production and found that CO₂ emissions were 36% lower than that of petroleum-based processes ^[9][13]. Thus, instead of a conventional fossil-based 2,3-BDO product, large industries have tried to produce 2,3-BDO by utilizing sugar or a direct biomass, which is a future solution to the depletion of fossil fuels.

Biosynthesis of 2,3-BDO is considered ea mixed-acid fermentation pathway, which is accompanied by the formation of various end-products depending on the fermentation strategies and microorganisms ^[10][15]. Until now, the physiological function of the metabolic regulation of 2,3-BDO biosynthesis has not been completely elucidated yet. However, several physiological significances have been elucidated, including intracellular acidification prevention, energy and carbon reservation, and regulation of the NAD⁺/NADH ratio ^[11][12][13][8,16,17]. Since the production of 2,3-BDO in bacteria is accompanied by the formation of acids as co-products ^[14][18], the biosynthesis of 2,3-BDO would be triggered to prevent this acid formation as the pH decreases ^[15][19]. It has been proven that an extra supplementation of acid promotes the metabolic pathway of 2,3-BDO, suggesting that 2,3-BDO is a neutral metabolite to counteract too high intracellular acidification ^[16][20]. Moreover, many investigations have been conducted by using acetic acid, which is already known to be involved in the 2,3-BDO metabolic pathway ^[17][21] as acid supplementation to analyze the influence on 2,3-BDO biosynthesis ^[16][20]. The glycolysis during 2,3-BDO production is accompanied by the oxidation of NADH, which can contribute to the maintenance of the NAD⁺/NADH ratio in the cell ^[18][22]. Finally, 2,3-BDO can be reutilized when carbon and energy sources are depleted; hence, the metabolism of 2,3-BDO can be treated as a strategy for carbon and energy storage ^[19][23].

In bacterial metabolism, assorted monosaccharides (glucose, hexoses, pentoses) need to be converted to pyruvate in the first step. From glucose, pyruvate can be formed by the glycolysis pathway, while pentose needs an extra step involving pentose phosphate. Three key enzymes are required for the 2,3-BDO production from pyruvate, namely α-acetolactate synthase (ALS), α-acetolactate decarboxylase (α-ALD), and 2,3-BDO dehydrogenase or acetoin reductase (BDH, AR) ^[20][10]. Pyruvate is the key intermediate, which undergoes reduction or oxidation to produce several organic acids (lactate, formate, acetate, succinate), ethanol, and acetoin, depending on the pH value and oxygen content. To enhance the accumulation of 2,3-BDO and eliminate by-product formation during fermentation, genetic engineering and optimization of the fermentation parameters have been investigated ^[21][24]. Most microorganisms produce 2,3-BDO by three consecutive metabolic conversion steps: from pyruvate to α-acetolactate, from α-acetolactate to acetoin, and from acetoin to 2,3-BDO. First, pyruvate can be converted either into lactate, by employing L-/D-lactate dehydrogenase (LDH) and NADH, or into α-acetolactate via decarboxylation by ALS. Furthermore, α-acetolactate can spontaneously transform into diacetyl by decarboxylation with the existence of oxygen. The diacetyl can subsequently be converted into acetoin by employing the reducing power of NADH and diacetyl reductase (DAR). Otherwise, α-acetolactate can be converted into acetoin by α-ALD without the presence of oxygen. Finally, acetoin can be converted into 2,3-BDO by BDH. It is noteworthy that, as an intermediate compound prior to the formation of 2,3-BDO, acetoin would be obtained by the reversible metabolic conversion from 2,3-BDO. It has been reported that low pH is an effective inducer for all the vital enzymes, α-ALD, ALS, and BDH, resulting in the enhanced biosynthesis of 2,3-BDO ^[17][22][21,25].

Butanediol dehydrogenase (BDH) is a tetrameric enzyme that needs NADH as a coenzyme. Three stereoisomers of 2,3-BDO can be formed from two acetoin isomers due to the activity of BDHs with variable stereospecificities, or by a cyclic pathway. The exact mechanism of the stereoisomer formation of 2,3-BDO has not been clearly understood. Over the past few years, several investigations have been conducted to understand the formation of 2,3-BDO stereoisomers; in particular, various mechanisms of stereoisomers have been elaborated in K. pneumonia. In 1960, the first series of preliminary experiments were conducted based on the assumption that acetoin racemase, L-(+)-BDH and D-(−)-BDH were involved in the production of stereoisomers of 2,3-BDO. A possible hypothesis mechanism was deduced by Taylor and Juni, but the existence of racemase was not proven at that time ^[23][26]. Then, Voloch et al. developed a method that used liquid chromatography to determine the racemic acetoin, meso-2,3-BDO, (2R,3R)-2,3-BD, and (2S,3S)-2,3-BD and proposed a model that assumed that D-(−)-BDH, L-(+)-BDH, and acetoin racemase had an effect on the formation of stereoisomers. It was proved that only R-acetoin was produced from pyruvate via α-acetolactate in K. pneumoniae; thus, the formation of (2S,3S)-2,3-BDO was dependent on acetoin racemase. Only a slight amount of acetoin racemase was detected ^[24][27]. After that, Ui et al. established a method to measure all the types of stereoisomers of 2,3-BDO and acetoin ^[25][28]. Based on this finding, Ui et al. proposed a new model involving NADPH-link diacetyl reductase (S-acetoin forming) and D-(−)-BDH for the formation mechanism of 2,3-BDO isomers in P. polymyxa. Blue Sepharose CL-6B was used to separate and identify the two enzymes with standard pure isomers of 2,3-BDO and acetoin, respectively. In addition, acetoin and butanediol racemases were not detected. Therefore, the author concluded that R-isomers of acetoin were formed in the absence of NADH, while the racemate form of acetoin was formed in the presence of NADH. The D-isomer of 2,3-BDO was predominantly produced, together with a small amount of meso-form in P. polymyxa ^[26][29]. In the meantime, Ui et al. also proposed a model in K. pneumoniae considering the existing of three BDHs: meso-BDH, L-(+)-BDH, and D-(−)-BDH. The three types of BDH were separated individually, and their stereo specificities were verified by the utilization of pure isomers, including (2R,3R)-2,3-BDO, (2S,3S)-2,3-BDO, meso-2,3-BDO, R-acetoin, and S-acetoin. Also, the acetoin or 2,3-BDO racemase was not detected. Therefore, the author concluded that the R-acetoin was formed in the absence of NADH, while S-acetoin was formed in the presence of NADH ^[27][30].

In nature, there are numerous microbial species which can produce 2,3-BDO from the pyruvate metabolic pathway. Besides bacteria, several yeasts and even algae have been reported to produce 2,3-BDO. However, considering the industrialization of 2,3-BDO production, bacteria appear to be the only alternative at present because of the very low productivity of other species. The representative species are Klebsiella pneumoniae, Klebsiella oxytoca, Enterobacter aerogenes, Paenibacillus polymyxa and Serratia marcescens, which are considered promising producers for scaled-up 2,3-BDO production. Species of Klebsiella were firstly investigated for 2,3-BDO production due to their ability to be cultivated rapidly in a simple medium on various carbon resources; in particular, K. pneumoniae can use almost all carbon resources, such as lignocellulosic biomass, agriculture waste, and glycerol. However, ethanol, lactate, and acetate are also produced as by-products, which increases the costs of downstream processing. Therefore, many research studies have been conducted on how to modify the metabolic pathways. When genes related to the formation of co-products, such as lactic acid, ethanol, and acetic acid were knocked out from K. pneumoniae, a higher titer of 2,3-BDO in batch fermentation could be achieved ^[28][35].

Similarly, ethanol was one of the major by-products during 2,3-BDO production when K. Oxytoca ME-UD-3 was used. A homologous recombination technique was employed to inactivate ADH, which exhibited an obvious augmentation of 2,3-BDO production with a yield of 0.48 g/g, and no ethanol was detected ^[29][38]. After that, Cho et al. reported an enhancement of 2,3-BDO production by K. oxytoca M1 through the overexpression of budC. 2,3-BDO production was remarkably increased to 142.5 g/L, while the production of acetoin decreased. As a result of overexpression of budC, acetoin reductase presented an evidently high catalytic activity ^[30][39]. Jantama et al. eliminated alcohol dehydrogenase E, acetate kinase A-phosphotransacetylase, and lactate dehydrogenase A in K. oxytoca KMS005. The production of 117.4 g/L of 2,3-BDO was obtained, with a yield of 0.49 g/g and productivity of 1.20 g/L/h. Only trace amounts of acetate and ethanol were detected as by-products ^[31][40].

E. aerogenes is a competitive candidate for 2,3-BDO production, but its by-products, such as lactic acid, impede the target production efficiency. Kim et al. eliminated the lactate dehydrogenase gene to improve 2,3-BDO production. The ldhA deletion mutant resulted in a 97% decrease in lactate production (0.34 g/L), while the productivity and yield of 2,3-BDO increased by 16.3% and 11.1%, respectively.

Engineered Saccharomyces cerevisiae is a potential host strain to produce 2,3-BDO, as substantial amounts of pyruvate could be shunted to 2,3-BDO production rather than to ethanol synthesis. By optimizing the expression levels of Pdc in S. cerevisiae, the maximum 2,3-BDO titer was 154.3 g/L which was the highest among the reported microbial production studies ^[32][41].

Bacillus amyloliquefaciens B10-127 was also considered a possible candidate for the scaled-up production of 2,3-BDO, but its industrialization was impeded by the high yield of acetoin, lactate, and succinate as by-products. During 2,3-BDO production, BDH catalyzes the conversion from acetoin to 2,3-BDO with the concomitant oxidation of NADH to NAD⁺. Yang et al. overexpressed bdh and gapA and showed that this could effectively enhance 2,3-BDO production and inhibit the formation of by-products ^[33][42].

Also, S. marcescens can be used to produce 2,3-BDO; however, few studies on this are available. Bai et al. identified the slaC gene for meso-2,3-BD and (2S,3S)-2,3-BDO production. The bdhA gene from Bacillus subtilis 168 was introduced into S. marcescens MG1 to replace the slaC gene, leading to (2R,3R)-2,3-BDO accumulation. The excess (2R,3R)-2,3-BDO dehydrogenase expression could enhance the production of (2R,3R)-2,3-BDO to 89.81 g/L, with only 2.11 g/L meso-2,3-BDO formation ^[34][43].

Moreover, pure meso-2,3-BDO or (2R,3R)-2,3-BDO can be synthesized by Bacillus licheniformis MW3 (ΔbudC). Ge et al. developed a process that included the screening, analysis, mutation, and evaluation of chiral 2,3-BDO production ^[35][44]. Li et al. attempted to synthesize pure (2R,3R)-2,3-BDO by E. cloacae via deletion of the bdh, ldh, and frdA genes to block by-production formation. The engineered strain produced 152 g/L (2R,3R)-2,3-BDO with 3.5 g/L/h productivity ^[36][45]. Obliteration of the dudA gene in P. polymyxa ZJ-9 resulted in the production of (2R,3R)-2,3-BDO at 25.88 g/L with a 99% purity in fed-batch cultivation ^[37][46]. Then, the enhancement of (2R,3R)-2,3-BDO production to 111 g/L by P. polymyxa DSM 365 was achieved ^[38][47]. Thus, P. polymyxa is an efficient producer for (2R,3R)-2,3-BDO production.

Although massive progress in metabolic engineering has been made to increase the efficiency of 2,3-BDO production, its microbial production from refined sugars has not been economically feasible ^[39][50]. The cost of commercial refined sugars exceeded 30% of the total cost ^[40][51]. Cheaply available and abundant substrates such as hydrolysates of starch, corn cob, Jerusalem artichoke, whey permeate, molasses, and crude glycerol derived from the biodiesel industry have attracted great interest for 2,3-BDO production (Figure 2) ^[41][52]. Generally, these feedstocks can be classified as lignocellulose biomass, food industry side streams, and by-products from the biofuel industry. Normally, the production of 2,3-BDO from lignocellulosic biomass includes three essential processes: (1) pretreatment of the biomass to break up interactions between the polymeric components and cellulose to make them more accessible for the upcoming step, commonly known as enzymatic hydrolysis or saccharification; (2) employing hydrolases like cellulases, xylanases, and other carbohydrases to hydrolyze the pre-treated biomass; and (3) fermentation of monosaccharides to produce 2,3-BDO. It is generally agreed that the effectiveness of pretreatment and enzymatic hydrolysis are the decisive factors determining the efficiency of 2,3-BDO production.