Produce D-allulose from non-food biomass by integrating corn stalk hydrolysis with whole-cell catalysis

D-allulose is a high-value rare sugar with many health benefits. D-allulose market demand increased dramatically after approved as generally recognized as safe (GRAS). The current studies are predominantly focusing on producing D-allulose from either D-glucose or D-fructose, which may compete foods against human. The corn stalk (CS) is one of the main agricultural waste biomass in the worldwide. Bioconversion is one of the promising approach to CS valorization, which is of significance for both food safety and reducing carbon emission. In this study, we tried to explore a non-food based route by integrating CS hydrolysis with D-allulose production. Firstly we developed an efficient Escherichia coli whole-cell catalyst to produce D-allulose from D-glucose. Next we hydrolyzed CS and achieved D-allulose production from the CS hydrolysate. Finally we immobilized the whole-cell catalyst by designing a microfluidic device. Process optimization improved D-allulose titer by 8.61 times, reaching 8.78 g/L from CS hydrolysate. With this method, 1 kg CS was finally converted to 48.87 g D-allulose. This study validated the feasibility of valorizing corn stalk by converting it to D-allulose.


Introduction
The excessive intake of high energy sugars has caused many health problems, such as diabetes, hypertension, hyperlipidemia, and other diseases (Xia et al., 2021). D-allulose (D-psicose or D-ribo-2-hexulose) is a good sugar substitute, because it is an low calorie sweeter (70% sweetness of sucrose), and inert in energy metabolism (Iida et al., 2010). Besides, D-allulose also has many other health benefits, such as anti-oxidative, anti-diabetic (Hossain et al., 2015), anti-obesity (Nagata et al., 2015), and neuroprotective effects, and others (Juneja et al., 2019). The approval of D-allulose as GRAS by FDA has stimulated its market demand as food ingredient and dietary supplement (Zhang W. L. et al., 2017). However, as a rare sugar, D-allulose is hardly found in nature (Mu et al., 2011). Very small quantities of D-allulose can be found in the stems or leaves of Itea and wheat and some bacteria (Zhang et al., 2016b).
To co-overexpress glucose isomerase and D-psicose epimerase in single cell, AcceGI and CcDPEase were assembled into monocistronic form by using isocaudomers (AvrII, NheI, and SalI) and subsequent T4 ligation (Xu et al., 2012). The gene copy number optimization was also carried out by using this method.

Protein expression and enzymatic activity analysis
The recombinant protein expression was carried out following previous methods with moderate modification (Lv et al., 2017). A E. coli BL21 (DE3) colony containing corresponding recombinant plasmid was inoculated into a sterile incubating tube containing 2 mL LB medium. After overnight incubation at 37°C and 200 rpm, 1 mL culture was inoculated into 250 mL shaking flask containing 25 mL LB medium. The strain was cultured at 37°C and 200 rpm until OD 600 reach 0.6-0.8. Cool down the culture to 25°C, and add 0.5 mM isopropyl-β-D-thiogalactopyranoside (IPTG) for induction. After another 8 h incubation at 25°C and 200 rpm, centrifuge at 4°C and 8,000 rpm to collect cell pellet. After wash with pure water, the cells were resuspended in water, and used for SDS-PAGE assay, enzymatic activity analysis, process optimization, or immobilization.
The biomass was measured by recording the optical density at 600 nm (OD 600 ) with a Tecan Infinite ® M Plex microplate reader (Tecan, Männedorf, Switzerland). SDS-PAGE (Sangon Biotech, Shanghai, China) was used to validate the recombinant protein expression. The enzymatic activities of AcceGI and CcDPEase were measured following previous methods (Mu et al., 2011;Mu et al., 2012).
Frontiers in Bioengineering and Biotechnology frontiersin.org

Corn stalk pretreatment
The CS was collected from local farm (Yuanyang county, Henan province, China, 113.9 o E, 35.1 o N), and then totally dried, ground and screened with a 60 mesh sieve. The pulping was carried out by cooking with 2% (w/v) NaOH solution at 80°C and atmospheric pressure for 2 h. The ratio of CS to NaOH was 1:20 (w/w). Wash the pulp with pure water until pH maintain stable, then totally dry the pulp in oven at 105°C. The hydrolysis was carried out with cellulases (Qingdao Vland Biotech Inc., Qingdao, China) in deionized water (adjust to pH5.0 with acetic acid). It should be noted that the hydrolysis buffer should not contain any sodium ion, which will cause the calcium alginate fiber instable in the next step. A final concentration of 10% (w/v) substrate and (10 FPU cellulase)/(g substrate) was added. Incubate the mixture at 50°C and 200 rpm for 3 days for sufficient hydrolysis.
To develop the one-step D-allulose producing whole-cell catalyst, we co-overexpressed AcceGI and CcDPEase in monocistronic form (G-P) in 1 cell by using the ePathBrick method (Xu et al., 2012). When incubating the whole-cell catalyst (G-P) with 50 g/L D-glucose, 3.94 g/L D-allulose was produced with a yield of 7.88% ( Figures 2B, C).

Improving the activity of whole-cell catalyst by balancing enzyme expression level
Spatially confining enzymes is a commonly used approach to preventing intermediates diffusion and thus improving enzymatic cascade efficiency (Zhong et al., 2022). Consequently, we designed a panel of fusion proteins, which linked AcceGI and CcDPEase directly or through flexible or rigid linkers. The expression of fusion proteins was validated by SDS-PAGE analysis (Figure 2A). The enzymatic activity analysis showed that all these fusion enzymes maintained their native activities. Moreover, longer linkers conferred enzymes with higher activities, which should be a

Plasmid
Genetic characteristics Reference or source pET28a (PB)-AcceGI ×7 -CcDPEase pET-28a (PB) carrying 7-copy AcceGI and CcDPEase in monocistronic form This study a These genes were annotated as both D-psicose 3-epimerase and xylose isomerase in the local blast analysis.
Frontiers in Bioengineering and Biotechnology frontiersin.org result of less steric hindrance (Supplementary Figure S1). Consequently, we used the two best performing fusion proteins (GS3P and GE3P) to validate the enzymatic cascade. HPIC results showed that the fusion enzymes (GS3P and GE3P) directly converted D-glucose to D-allulose ( Figure 2B). However, neither of the fusion enzymes performed better than that of the free enzyme cascade ( Figure 2C). We deem this to be resulted from the steric hindrance effect, because the enzymes with longer linkers showed higher activities and the rigid linker performed better than the flexible one (Supplementary Figure S1) (Chen et al., 2013). These results indicated that the intermediate diffusion should not to be the predominant rate limiting factor in this whole-cell catalyst. We next tried to improve the activity of the whole-cell catalyst by balancing the expression levels of the enzymes. This was achieved by optimizing gene copy numbers by using the ePathBrick method (Xu et al., 2012). The results showed that increasing AcceGI copy number improved the whole-cell activity by 24.38%, while increasing CcDPEase copy number had no effect on the wholecell catalyst ( Figure 2D). This result indicated that AcceGI is the rate limiting step of the enzymatic cascade. Subsequently we used this straightforward method to further stepwise increase the gene copy number to 7. The results showed that the activities of the while-cell catalyst improved along with the increasing of the AcceGI copy number until 5, and thereafter activities decreased ( Figure 2E). The improvement of AcceGI expression level was validated with SDS-PAGE analysis ( Figure 2F). Moreover, improving AcceGI copy number did not have obvious affect on the biomass (Supplementary Figure S2). Consequently, we used the optimal whole-cell catalyst containing 5 AcceGI (AcceGI ×5 -CcDPEase) in the subsequent research.

Producing D-allulose from corn stalk
The purpose of this study is to produce D-allulose from CS. So we next developed a process to convert CS into monosaccharide by subsequential grind, alkaline pretreatment, and digestion. The resulting hydrolysate was a light yellow solution, containing 70.82 g/L D-glucose, 19.13 g/L D-xylose, 1.69 g/L L-arabinose, and other minor components (D-mannose, D-galactose, lignans, and furfural) (Figures 3A, B). The D-glucose yield from dry CS was 27.83%.
When using the best performing whole-cell catalyst (AcceGI ×5 -CcDPEase) to convert this hydrolysate, 2.82 g/L D-allulose was

FIGURE 1
Expression and activity analysis of predicted glucose isomerase, xylose isomerase, and D-psicose 3-epimerase. (A) Expression of predicted glucose isomerase and glucose isomerase (B) Expression of predicted D-psicose 3-epimerase. (C) Enzyme activity analysis of predicted glucose isomerase and glucose isomerase. The analysis was carried out by converting D-glucose to D-fructose (D) Enzyme activity analysis of predicted D-psicose 3-epimerase. The analysis was carried out by converting D-fructose to D-allulose. In all panels, control refers to E. coli BL21 (DE3) containing empty pET-28a (PB) plasmid. In SDS-PAGE results, red arrows indicate the recombinant proteins. In HPIC results, standards refer to mixture of D-glucose, D-fructose, and D-allulose standards.
Frontiers in Bioengineering and Biotechnology frontiersin.org produced from 50 g/L D-glucose (from CS hydrolysate) with a yield of 5.64% ( Figure 3C). This yield equals to 61.82% of that from pure D-glucose ( Figure 3C). We deem the loss of catalytic activity to two reasons, the inhibition effect of inhibitory factors (such as furfural) and the competition effect of other potential substrates. For instance, the glucose isomerase has been shown being capable of utilizing both D-glucose and D-xylose (Patra and Bera, 2014). To validate our hypothesis, we mimicked the hydrolysate by adding D-xylose to D-glucose to the same final concentrations. The activities of free AcceGI and CcDPEase to pure D-glucose and D-fructose were defined as 100%, respectively. The results showed that adding D-xylose decreased AcceGI activity by 24.14%, which is similar to that of the CS hydrolysate ( Figure 3D). Moreover, the HPIC results showed that free AcceGI converted D-xylose into D-xylulose, which indicated the competitive effect between D-xylose and D-glucose to AcceGI ( Figure 3E). On the other hand, potential substrates did not have effect on CcDPEase activity ( Figure 3F). These results indicated that the activity decrease when using CS hydrolysate as substate is predominantly resulted from competition between D-glucose and D-xylose.

Enhancing D-allulose production by developing and optimizing a recyclable catalytic fiber
To improve the whole-cell catalyst reusability and process economical efficiency, we designed a simple microfluidic system to immobilize the whole-cell catalyst in calcium alginate fiber. As shown in Figure 4A, a mixture of whole-cell catalyst and sodium alginate solution was in the syringe, which was driven by a motor. Calcium chloride solution in the beaker reacts with the fiber from the syringe, resulting in stable calcium alginate fibers ( Figure 4B). The fiber's diameter was controlled by the needle, and the flow rate was controlled by the controller. To validate the cell immobilization in the fiber, we substituted the enzymes (AcceGI and CcDPEase) with an enhanced green fluorescent protein (EGFP) to track the recombinant cells. Fluorescent microscope results showed that these cells were immobilized in the fibers, whose diameter was 500 μm ( Figure 4C).
We next tried to improve the fiber's activity by process optimization. The results showed that for the recombinant protein induction, the optimal induction time was 12 h, and optimal IPTG concentration was 0.05 mM (Figures 4D, E). For the immobilization process, the optimal concentrations of calcium chloride and sodium alginate were 2.5% (w/w) and 0.5% (w/w) respectively ( Figures 4F, G), and the optimal immobilization time was 80 min ( Figure 4I). Although higher D-allulose production was obtained along with higher cell dosage (OD 600 = 10-35), however it increased very slow after OD 600 reaching 15 ( Figure 4H). As a result, we used OD 600 = 15 in the subsequent optimization. For the D-allulose production process, the optimal substrate (D-glucose in CS hydrolysate) concentration and temperature were 50 g/L and 65°C ( Figures 4J, K) respectively. Although higher D-allulose production was obtained along with longer reaction time during 2-12 h, however it increased very slow after 8 h ( Figure 4L). As a Taking together, the process optimization improved D-allulose production by 8.61 times from 1.02 g/L to 8.78 g/L. To validate the fiber's reusability, we reused one single fiber in 10 catalytic reactions. The results showed that the fiber maintained 84.53% relative activity after 10-cycle reuse ( Figure 4M). It should be noted that calcium alginate fiber is unstable in the presence of some ions like sodium. Consequently, we did not optimize the reaction pH (avoiding the involvement of sodium buffer) and carried out all the bioconversions in pure water at natural pH (around 7.5). The results showed that the calcium alginate fiber was stable in pure water ( Figure 4M). This not only stabilized the fiber, but also could be helpful for reducing the cost of future large scale applications (eliminating the involvement of any buffer).

Discussion
D-allulose is a GRAS sugar substitute, and characterized with many health benefits (Iida et al., 2010). As a rare sugar, D-allulose is hardly found in nature. The currently predominant production method is based on the Izumoring strategy, which use either D-glucose or D-fructose as substrate (Izumori, 2006;Juneja et al., 2019;Zhang et al., 2020). With the increasing concerns about insufficient food supply, exploring non-food based D-allulose producing method is becoming necessary (Hobbs, 2020). CS is one of the main agricultural wastes continuously produced worldwide. Dry CS contains more than 30% cellulose, which is a polymer of D-glucose . Consequently, CS biomass could be a promising alternative feedstock to D-glucose or D-fructose.
The Izumoring strategy is limited by thermodynamic equilibrium. When the reaction reach equilibrium in the D-allulose production from D-glucose, the D-allulose yield was 18.18% (the ratio of D-glucose:D-fructose:D-allulose was 6.5:7:3) . In the present study, 8.78 g/L D-allulose was produced from 50 g/L D-glucose (obtained by CS hydrolysis). The D-glucose yield from dry CS was 27.83%. Taking together, 1 kg CS was finally converted to 48.87 g D-allulose. This yield seemed very low. However, the D-allulose yield from D-glucose was 17.56% (8.78 g/L D-allulose from 50 g/L D-glucose in CS hydrolysate), which is close to the equilibrium state (18.18%).
The dry CS is mainly composed of cellulose, hemicellulose, and lignin. In this study, the hydrolysis yield three main monosaccharides (70.82 g/L D-glucose, 19.13 g/L D-xylose, and 1.69 g/L L-arabinose). Both D-glucose and D-xylose can be utilized by D-glucose isomerase, and the products are D-fructose and D-xylulose respectively (Mu et al., 2012). The substrate competition between D-xylose and D-glucose decreased D-fructose production by 24.14% ( Figure 3D). This issue can be resolved by removing D-xylose from the solution or improving D-glucose isomerase's substrate specificity. However, neither of

FIGURE 3
Producing D-allulose from CS. (A) CS hydrolysate by subsequential grind, alkaline pretreatment, and digestion (B) Major monosaccharide components of the CS hydrolysate. (C) D-allulose titers from pure D-glucose and CS hydrolysate (D) Relative activities of free AcceGI to D-glucose and D-xylose mixture and CS hydrolysate. The activity of free AcceGI to pure D-glucose was defined as 100%. (E) D-xylose can be used as the substrate of AcceGI, and converted to D-xylulose (F) Relative activities of free CcDPEase to mixtures of monosaccharides. The activity of free CcDPEase to pure D-fructose was defined as 100%. D-Glu refers to D-glucose. D-Xyl refers to D-xylose. L-Ara refers to L-arabinose. Hydro refers to CS hydrolysate. When using CS hydrolysate as substrate, the performance conditions were identical to those using D-glucose as substrate. The detailed performance conditions were described in the materials and methods section. Red stars indicate the effects of CS hydrolysate to enzymes.
Frontiers in Bioengineering and Biotechnology frontiersin.org them is easy to achieve. In the following research, we are going to simultaneously produce D-allulose and D-ribose, which is a important pentose with many valuable physiological functions (Mahoney et al., 2018;Li et al., 2021). D-ribose is currently produced from D-glucose through the pentose phosphate pathway Park et al., 2017). However, using this method 1 carbon is lost by releasing 1 CO 2 for each D-ribose production, which is neither economical efficient or environmental friendly (Srivastava et al., 2012). By converting D-xylose to D-ribose (both of them are pentose) will not only explore novel D-ribose producing method, but also further improve the full utilization of CS.

Conclusion
In the present study, we produced D-allulose from CS by integrating the CS hydrolysis with a D-allulose producing whole- Frontiers in Bioengineering and Biotechnology frontiersin.org cell catalyst. By gene screening, co-overexpressing, copy number optimization, we obtained a high-efficient whole-cell catalyst, and used it to produce D-allulose from CS hydrolysate. We next designed a microfluidic system and used it to immobilize the whole-cell catalyst. After process optimization, the D-allulose titer improved by 8.61 times, reaching 8.78 g/L from CS hydrolysate. In this process, 1 kg CS was finally converted to 48.87 g D-allulose. The bioconversions were carried out in pure water. This study validated the feasibility of producing D-allulose from CS, a non-food feedstock.

Data availability statement
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors.