The Izumoring strategy is a promising approach for the bioproduction of any kind of hexose sugar and involves D-tagatose 3-epimerases (DTEases), polyol dehydrogenases, and aldose isomerases (Izumori, 2006). In the past few decades, the Izumoring strategy has proven effective for biosynthesizing rare sugars. DTEase family enzymes are the key enzymes for the biosynthesis of D-allulose from D-fructose (Figure 2) and include DTEase and D-allulose 3-epimerase (DAEase). DAEase exhibits higher specificity for D-allulose than D-tagatose (Kim et al., 2006). Since the first DTEase from Pseudomonas cichorii ST-24 was identified in 1993, other DTEase enzymes were isolated from a variety of species, such as Agrobacterium tumefaciens (Kim et al., 2006), Clostridium Bolteae (Jia et al., 2014), Dorea sp. CAG317 (Zhang et al., 2015), Ruminococcus sp. 5_1_39BFAA (Chen et al., 2016), Treponema primitia ZAS-1 (Zhang et al., 2016c), Rhodobacter sphaeroides (Qi et al., 2017), Ruminococcus sp. (Li et al., 2018), Arthrobacter globiformis M30 (Yoshihara et al., 2017), Clostridium cellulolyticum H10 (Su et al., 2018), Sinorhizobium sp. (Zhu et al., 2019b), and Rhodopirellula baltica SH 1 (Zhang et al., 2020). Details of the catalytic properties of DTEase-family enzymes from different species are presented in Table 1. Itoh et al. (1995) first reported the immobilization of DTEase from P. cichorii on chitopearl beads and 90 g D-allulose from 500 g D-fructose was generated. Since then, the methods for DTEase immobilization have attracted extensive interest (Lim et al., 2009; Tseng et al., 2014; Narayan Patel et al., 2018), and the immobilized enzymes exhibited improved thermal stability and storage stability.

Currently, agricultural byproducts, such as fruit and vegetable residue, are generating major agricultural problems (Lai et al., 2017). Agricultural residues are usually buried in landfills or incinerated (Park and Yoon, 2015). However, these residues contain large amounts of dietary fibers and sugars, including sucrose, D-glucose, and D-fructose. The conversion of dietary fiber and sugars to high value-added products represents a significant step towards alleviating agricultural problems.

Cascade catalysis is considered as a very attractive approach compared with traditional step-by-step synthesis. This strategy is often used to produce rare sugars from inexpensive materials, such as sucrose, Jerusalem artichoke, inulin, and fruit/vegetable residues (Figure 3; Wagner et al., 2015; Song et al., 2016, 2017; Zhang et al., 2017; Yang et al., 2019; Li et al., 2020a). D-Allulose has been efficiently synthesized from sucrose using purified recombinant invertase, D-xylose isomerase, and DTEase. Moreover, practical integration of a cascade solution with simulated moving bed (SMB) chromatography has been used to produce pure D-allulose (99.9%) with very high yields (89%; Wagner et al., 2015). D-Allulose can also be synthesized in a two-step cascade reaction involving Jerusalem artichoke hydrolysis (Song et al., 2017), cruciferous vegetable residue (Song et al., 2016) and inulin (Li et al., 2020a). To decrease production costs and avoid enzyme purification, Zhu et al. developed a one-pot two-enzyme reaction system with a novel exo-inulinase from Bacillus velezensis (BvInu) and DAEase from Ruminococcus sp. for the production of D-allulose from Jerusalem artichoke. BvInu and DAEase were expressed in Bacillus subtilis and secreted into supernatant without purification. Under the optimal ratio of BvInu/RDAE (80:40 U/g inulin) at 50°C for 2 h, 10.3 g/L of D-allulose was obtained from 50 g/L inulin (Zhu et al., 2020).

Compared with one-pot cascade reactions, whole-cell biocatalyst reactions have several advantages (Figure 4): (1) Cells containing enzymes are easily obtained without tedious purification of enzymes; (2) the cellular context provides a suitable microenvironment and cofactor regeneration (ATP, NAD⁺/NADH); (3) the cell walls and membranes protect the enzymes against harsh reaction conditions; and (4) the colocalization of multiple enzymes within the cell enhances the local concentrations of enzymes and decreases the diffusion of intermediates in cascade reactions (Wu and Li, 2018).

Engineered Escherichia coli is one of the most commonly used organisms for the production of D-allulose because of its clear background, fast growth rates, simple culture, and stable genetics. DAEase and D-glucose isomerase (GIase) from Acidothermus cellulolyticus were coexpressed to produce D-allulose from D-glucose (Zhang et al., 2017). Similarly, DAEase and xylose isomerase (XI) were coexpressed to produce D-allulose with D-glucose as the substrate (Chen et al., 2017). In the above two approaches, D-glucose was first converted to D-fructose by GIase or XI, and then, D-fructose was immediately isomerized to D-allulose by DAEase.

Although good productivity was obtained by engineering E. coli, which was not applicable in the food industry because of the endotoxins and their non-food grade classification. To date, DAEase has been successfully expressed in several food-safe strains, such as B. subtilis, Saccharomyces cerevisiae, and Corynebacterium glutamicum (Li et al., 2015b; He et al., 2016). He et al. (2016) displayed DTEase from Clostridium scindens ATCC 35704 on the surface of B. subtilis spores for the production of D-allulose from D-fructose. DAEase was fused at the C-terminus of CotZ and exhibited high thermostability. After five cycles of utilization, 60% activity was maintained (He et al., 2016). In our previous study, we innovated a spore surface display technique to produce D-allulose from D-glucose. In this approach, the key enzymes XI from Thermus thermophilus and DAEase from A. tumefaciens were immobilized on S. cerevisiae spores using biological and chemical methods, respectively (Li et al., 2015b). In addition, multiple DAEases and invertase (INV) were overexpressed in C. glutamicum, and the engineered cells immobilized with alginate were subjected to a cascade reaction in a one-pot, two-step reaction system to generate D-allulose from cane molasses. After reaction for 8 h, 61.2 g/L D-allulose was obtained, which represented 17.4% of the total monosaccharides (Yang et al., 2019). Currently, the Izumoring strategy is the main method for industrial production of D-allulose, which could be achieved by using SMB. Additionally, the separated D-fructose can be reused for D-allulose production. In brief, Izumoring strategy is the most simple and common way to biosynthesize D-allulose. However, the limitation of thermodynamic balance is a bottleneck which restricts the large-scale application of D-allulose in the food industry.

A main disadvantage of DHAP-dependent aldolase strategies is that the donor substrate DHAP is very expensive and unstable for large-scale synthesis (Schümperli et al., 2007). However, there are several routes for DHAP synthesis via enzymatic strategies. For instance, DHAP can be obtained via dihydroxyacetone (DHA), glycerol or glycerol 3-phosphate or through metabolic pathways from an inexpensive raw material, such as glucose or glycerol (Figure 6; Sánchez-Moreno et al., 2004; Li et al., 2012, 2015a; Wei et al., 2015; Yang et al., 2015, 2016). In engineered E. coli, catalysis of the aldol reaction by RhaD to synthesize rare sugars (D-allulose and D-sorbose) has been achieved using glucose as the carbon source and a continuous supply of D-glyceraldehyde into the medium. Following optimization of fermentative conditions, the isolated yield of D-allulose and D-sorbose was 0.21 mol/mol D-glyceraldehyde (Wei et al., 2015). D-Allulose and D-sorbose have been produced in a GRAS C. glutamicum strain with glucose and D-glyceraldehyde as feedstocks. The recombinant C. glutamicum strains harboring RhaD and fructose-1-phosphatase (YqaB) have been found to accumulate 19.5 g/L of D-sorbose and 13.4 g/L of D-allulose in a fed batch fermentation (Yang et al., 2015). In addition, D-allulose and D-sorbose have been produced in a recombinant E. coli strain overexpressing aldolase RhaD and YqaB from glycerol by fermentation. After 15 h of fermentation, the concentrations of D-sorbose (1.6 g/L) and D-allulose (1.23 g/L) were determined in the supernatant (Li et al., 2015a). Recently, our group constructed an efficient system for whole-cell cascade synthesis of D-sorbose and D-allulose from glycerol and D-glyceraldehyde, which yielded 15.3 g/L of D-sorbose and 6.4 g/L of D-allulose in a batch biotransformation (Chen et al., 2020a).

A microorganism platform for D-allulose synthesis based on aldolase RhaD would hold much promise for large-scale production. The main challenge is the high cost of the donor substrate DHAP and acceptor molecule D-glyceraldehyde. Fortunately, the problem of how to accumulate DHAP using the inexpensive industrial byproduct glycerol as a “green” carbon source has been solved. Thus, the next issue is to identify a novel enzyme to convert low-value glycerol to D-glyceraldehyde. A glycerol dehydrogenase or glycerol oxidase that can efficiently catalyze glycerol to D-glyceraldehyde is needed.

To address this issue, our group constructed a platform for the whole-cell cascade synthesis of D-sorbose and D-allulose from glycerol as the sole substrate. In this system (Figure 7), the donor substrate DHAP is generated by the glycerol assimilation pathway, and the endogenous DHAP is produced via overexpression of glycerol kinase (GK) and glycerol phosphate oxidase (GPO). The acceptor D-glyceraldehyde is directly generated from glycerol by alditol oxidase from Streptomyces coelicolor A3 (AldO_S.coe). Then, RhaD catalyzes the aldol reaction between DHAP and D-glyceraldehyde to form the corresponding ketose-1-phosphate. Finally, D-sorbose and D-allulose are obtained by removing the phosphate group by YqaB phosphatase. Using this system, the production yields of D-sorbose and D-allulose were enhanced approximately 1.7-fold and 1.2-fold after the overexpression of peroxidase (Prx02 or KatE) to eliminate the harmful effects of H₂O₂. A total of 7.9 g/L of D-sorbose and D-allulose was obtained from glycerol, with a total conversion rate of 17.7% (Chen et al., 2020b).

Our group also constructed a one-pot multienzyme system for the synthesis of D-sorbose and D-allulose from glycerol as the sole carbon source (Figure 8). Here, acid phosphatase PhoN from Shigella flexneri (PhoN-Sf) was introduced to the system instead of GK and YqaB, which catalyzed the phosphorylation reaction of glycerol in the first step and helped recycle the phosphate of ketose-1-phosphate in the last step to provide free rare-sugar molecules. AldO_S.coe was introduced to the above multienzyme cascade to synthesize D-sorbose and D-allulose exclusively from the readily available glycerol. Finally, 14.8 g/L of D-sorbose and D-allulose was obtained from glycerol (Li et al., 2020b). The above strategy also represents a very useful and low-cost approach for producing various other rare sugars. In a word, biosynthesis of D-allulose by fermentation based on DHAP-dependent aldolases is very promising. It would be more profitable to construct a cell factory based on DHAP-dependent aldolases strategy for the synthesis of D-allulose.