Thermostable D-Allulose 3-Epimerase for Long-Term Food-Compatible Continuous Production Systems

Abstract

D-allulose is a rare sugar with promising applications in food and health industries, owing to its low caloric value and multiple health benefits. In this study, we systematically investigated a thermostable D-allulose 3-epimerase (TcDAEase) from Thermogemmatispora carboxidivorans for food-compatible continuous production. The enzyme exhibited remarkable thermostability, with over 70% activity retained at 80 °C, and showed broad pH tolerance across the range of 8.0 to 13.0. Notably, TcDAEase exhibited high catalytic activity toward D-allulose and D-fructose even without the addition of metal ions. Moreover, food-grade Mg²⁺ was identified as enhancing enzyme activity by 14.3%, thus ensuring compliance with Generally Recognized as Safe (GRAS) standards for food applications. To improve industrial applicability, the enzyme was immobilized using a chitosan-diatomaceous earth (DE) matrix via three-step adsorption–crosslinking–embedding strategy. The immobilized TcDAEase achieved 48.7% ± 2.4% activity recovery and retained 90.3% ± 1.5% activity over seven reaction cycles. Furthermore, continuous production of D-allulose was realized in a packed-bed reactor, operating stably at 60 °C, pH 8.0 and 5 mM Mg²⁺ for 150 days, producing 756 kg of D-allulose with a conversion yield exceeding 89.7% of the theoretical maximum. Overall, this study provides a feasible strategy for the continuous and efficient production of high-value-added D-allulose in the food industry.

1. Introduction

In recent decades, the global prevalence of metabolic disorders such as obesity, diabetes, and cardiovascular diseases has surged, primarily due to high-fat and high-sugar dietary patterns [1,2]. In response, consumer preferences have shifted toward healthier sweetening alternatives, including erythritol, acesulfame and stevia, which offer low- or zero-calorie profiles [3,4]. Among them, D-allulose (also known as D-psicose), a rare monosaccharide and the C-3 epimer of D-fructose, has garnered increasing attention owing to its favorable nutritional and physiological properties. D-allulose exhibits approximately 70% of the sweetness and only 10% of the caloric value of sucrose (0.4 kcal/g vs. 4 kcal/g). In addition to its low caloric contribution, D-allulose has been shown to offer a range of health benefits, such as anti-obesity effects [5], modulation of cholesterol metabolism [6] and support for gut microbiota homeostasis [7]. These attributes position D-allulose as a highly promising functional sweetener for food and nutraceutical applications [8].

Despite its advantages, D-allulose is scarce in nature and its chemical synthesis typically requires harsh conditions and generates undesirable byproducts, limiting scalability [9,10]. As a result, biocatalytic methods offer a cleaner, more sustainable, and cost-effective route [11]. Among these, enzymatic conversion using ketose 3-epimerases, particularly D-tagatose 3-epimerase (DTEase, EC 5.1.3.31) and D-allulose 3-epimerase (DAEase, EC 5.1.3.30), has become the most viable industrial-scale production method for D-allulose [12].

To date, more than 20 microbial DAEases have been identified from various genera such as Arthrobacter globiformis M30 [13], Clostridium bolteae [14], Ruminococcus sp. [15] and Bacillus subtilis [16]. However, most of these enzymes display limited thermostability, with optimal activity typically below 50 °C. This low-temperature requirement constrains their application in high-temperature processes, which are desirable in the industry for improving substrate solubility, reducing viscosity and suppressing microbial contamination [17]. Furthermore, many reported DAEases require divalent metal cofactors such as Co²⁺ or Mn²⁺ for activation. These ions, despite their catalytic roles, are either toxic or disallowed in food processing, thereby limiting the practical use of such enzymes in food-related manufacturing processes [18]. Consequently, there remains a pressing need for the discovery and engineering of novel DAEases that combine high thermostability with activation by food-grade metal ions such as Mg²⁺ or Ca²⁺.

In recent years, various immobilization strategies have been explored to improve the performance of DAEases [19], yet challenges remain. Li et al. [20] developed a DNA-origami-based co-immobilization platform integrating DAEase and glucose isomerase, enabling one-pot cascade conversion of glucose to D-allulose with minimal byproduct formation (reduced by approximately 80%). Tang et al. [21] reported a ZnCo-MOF-based immobilized DAEase with enhanced thermostability (80% activity retention at 70 °C) and reusability (over 95% after eight cycles), attributed to Zn²⁺/Co²⁺ coordination-induced structural stabilization. However, the high cost and complexity of such nanomaterials limit their industrial applicability. Alternatively, several studies have explored traditional immobilization platforms, such as chitopearl beads and ion-exchange resins [13,22], enabling medium-term continuous operation for 60–120 days. Nevertheless, few systems have demonstrated stable catalytic performance beyond 100 days under food-compatible conditions, underscoring the need for robust and durable immobilized biocatalysts.

In this study, we characterized and immobilized a novel thermostable D-allulose 3-epimerase (TcDAEase) from Thermogemmatispora carboxidivorans. The enzyme exhibited high catalytic efficiency, remarkable thermostability, and activation by GRAS-approved Mg²⁺, thereby satisfying key requirements for industrial application. A cost-effective immobilization strategy based on chitosan and diatomaceous earth was further developed, enabling long-term continuous production of D-allulose in a packed-bed reactor over 150 days. This work provides a scalable and sustainable solution to the enzyme and process-level limitations in the production of D-allulose.

2. Materials and Methods

2.1. Materials and Strains

TcDAEase gene (NCBI accession: WP_052889376.1) was synthesized by General Biologicals Corporation (Chuzhou, Anhui, China). Escherichia coli BL21 (DE3) was utilized as the host strain for recombinant expression and was obtained from Vazyme Biotech Co., Ltd. (Nanjing, China). Substrates including D-allulose, D-tagatose, D-sorbose, and D-fructose were sourced from Sigma-Aldrich (St. Louis, MO, USA). The diatomaceous earth carrier was supplied by Yuantong Mineral Products Co., Ltd. (Jilin City, Jilin, China), and chitosan was purchased from Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). All other analytical-grade reagents were purchased from Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan), Acros Organics (Geel, Belgium), and Macklin Biochemical Co., Ltd. (Shanghai, China).

2.2. Gene Cloning, Expression, and Sequence Analysis of TcDAEase

The gene sequence encoding TcDAEase (NCBI ID: WP_052889376.1) was cloned into the pET22b (+) expression vector to construct the recombinant plasmid pET22b-TcDAEase [23]. The plasmid was transformed into E. coli BL21(DE3) for heterologous expression. Pre-cultures were grown in 100 mL Luria–Bertani (LB) medium supplemented with 50 μg/mL ampicillin at 37 °C 200 rpm overnight. The seed culture was then inoculated (2% v/v) into 100 mL auto-induction Terrific Broth (15 g/L tryptone, 25 g/L yeast extract, 10 g/L NaCl, 2 g/L glucose, 3 g/L lactose) [24]. Cultures were incubated at 37 °C until OD₆₀₀ reached 0.6, then shifted to 25 °C for 24 h to enhance protein yield. Cells were harvested via centrifugation (8000 rpm, 4 °C, 20 min). Multiple sequence alignment of TcDAEase with other reported D-allulose 3-epimerases was conducted using Clustal Omega (version 1.2.4, EMBL-EBI, UK) [25] to identify conserved catalytic residues and structural stability. A phylogenetic tree was constructed using MEGA version 11.0.13 (Molecular Evolutionary Genetics Analysis, Pennsylvania State University, University Park, PA, USA) via the maximum likelihood method with 1000 bootstrap replications to assess evolutionary relationships and statistical confidence.

2.3. Gene Cloning, Expression and Purification of TcDAEase

Harvested cells were lysed by ultrasonic disruption (60% amplitude, 1 s on/2 s off, 10 min on ice), and the lysate was clarified by centrifugation and 0.22 μm filtration. The supernatant was purified using a 5 mL nickel-affinity chromatography column equilibrated with 50 mM phosphate buffer (500 mM NaCl, pH 8.0). The enzyme was eluted using a linear imidazole gradient (0–500 mM), concentrated via centrifugal ultrafiltration for 20 min with 8000 rpm at 4 °C conditions (22 kDa MWCO). And dialyzed against 50 mM phosphate buffer (pH 8.0). SDS-PAGE with Coomassie Brilliant Blue staining was employed to confirm protein size and purity. SDS–PAGE was conducted under denaturing conditions using a 12% polyacrylamide resolving gel, following the method of Laemmli [26]. Protein samples were mixed with sample loading buffer and denatured at 95 °C for 5 min. Electrophoresis was carried out in Tris-glycine-SDS running buffer (pH 8.3) at a constant voltage of 120 V. A pre-stained low-range molecular weight marker (14.3–97.2 kDa) was included for size estimation. After separation, proteins were visualized using Coomassie Brilliant Blue R-250 staining (0.1% in 40% methanol and 10% acetic acid). For protein quantification, bovine serum albumin (BSA) standards (0–2 mg/mL) were used in the Bradford [27] assay.

2.4. Enzyme Assay

Enzymatic activity was measured in 1.0 mL reactions containing 50 mM sodium phosphate buffer (pH 8.0) and 50 g/L D-fructose. Reactions were initiated by adding purified enzyme and incubated at 60 °C for 5 min. The reaction was terminated by the addition of 100 μL of 1 M HCl. Prior to HPLC analysis, samples were neutralized with an equal volume of 1 M NaOH to avoid potential acid-induced retention time shifts. Termination was achieved by adding 100 μL of 1 M HCl and immediately freezing at −80 °C. Sugar concentrations were analyzed via HPLC equipped with a refractive index (RI) detector, using a calcium-form carbohydrate column (Benson Polymeric, Reno, NV, USA) at 80 °C with deionized water as the mobile phase (flow rate: 0.6 mL/min). The sample injection volume was 10 μL, and the total run time was 20 min. A standard calibration curve was generated using D-allulose dissolved in 50 mM sodium phosphate buffer (pH 8.0) supplemented with NaCl to match the ionic strength of the neutralized samples. One unit (U) of enzymatic activity was defined as the amount of enzyme required to produce 1 μmol of D-allulose per minute under the specified conditions. All measurements were performed in triplicate and reported as mean ± SD.

2.5. Biological Characteristics of TcDAEase

To simulate native conditions relevant to food-compatible applications, no metal chelation pretreatment was employed. The optimal temperature was determined by performing assays at 50–90 °C in 10 °C increments, while thermostability was evaluated by pre-incubating enzyme samples (without substrate) at 60, 70, and 80 °C in phosphate buffer, followed by activity measurement at intervals up to 8 h.

The pH dependence was assessed using five buffer systems (50 mM): citrate-phosphate (pH 4.0–5.0), phosphate (pH 6.0–8.0), glycine-NaOH (pH 9.0–10.0), carbonate (pH 11.0), and KCl-NaOH (pH 12.0–13.0). Enzyme pH stability was tested by 6 h incubation at 60 °C across these buffers.

To evaluate metal ion effects, divalent cations (1 mM final concentration; Fe²⁺, Co²⁺, Ni²⁺, Cu²⁺, Zn²⁺, Ca²⁺, Mg²⁺, Mn²⁺) were added individually. Additionally, the Mg²⁺ dose dependence was assessed at 0–25 mM concentrations to determine its catalytic enhancement profile.

2.6. Substrate Specificity and Kinetic Parameters

The substrate specificity of TcDAEase was evaluated against four ketohexose substrates: D-fructose, D-allulose, D-sorbose and D-tagatose. Each reaction was conducted under the standard assay conditions (60 °C, pH 8.0), and the specific activities for each substrate were determined. The relative activity was calculated by setting the activity toward D-fructose as 100%. For kinetic analysis, initial reaction rates were measured at varying concentrations of D-fructose ranging from 5 to 300 mM. The Michaelis–Menten kinetic parameters [28], including the Michaelis constant (K_m), turnover number (k_cat), and catalytic efficiency (k_cat/K_m), were determined by nonlinear regression fitting of the rate data to the Michaelis–Menten equation using OriginPro 9.0 [29]. All assays were performed in triplicate, and results were expressed as mean ± SD.

2.7. TcDAEase Immobilization Procedure

The bacteria used for enzyme immobilization were cultivated as described in Section 2.2. A three-step adsorption–crosslinking–embedding approach was employed using a chitosan–diatomaceous earth (DE) composite as the support matrix. Briefly, chitosan (10 g/L) was dissolved in 1% (v/v) acetic acid with constant stirring. Diatomaceous earth (100 g/L) and glutaraldehyde (0.5 g/L) were added, and the pH was adjusted to 4.5 using 1 M NaOH to activate the support. Wet biomass was then added at the desired concentration, followed by gentle stirring at 25 °C for 1 h to facilitate adsorption. Chitosan precipitation was induced by gradually adjusting the pH using either 1 M NaOH or 1.4 M Na₂SO₄ to reach one of three target pH ranges: acidic (5.0–6.0), neutral (6.5–7.0) or alkaline (10.0–11.0). To enhance mechanical strength, 1.4 M Na₂SO₄ was used for osmotic dehydration, after which the immobilized material was filtered, washed thoroughly with deionized water, and dried at 60 °C for 2 h until a constant weight was achieved. The final immobilized enzyme preparation was stored at 4 °C until use. To optimize immobilization efficiency and operational stability, three key parameters were systematically investigated. The effects of wet cell loading (5–30 g/L), chitosan precipitation pH (5.0–6.0, 6.5–7.0, and 10.0–11.0), and chitosan concentration (5–40 g/L) were investigated. For each condition, protein loading capacity and activity recovery were quantified. The enzyme activity of the immobilized preparations was determined under standard assay conditions. The activity recovery (%) of the immobilized TcDAEase was calculated as the ratio between the specific activity (U·g⁻¹) of the immobilized preparation and that of the corresponding amount of free wet cells. Breakpoint analysis and first-order kinetic fitting were applied to model saturation and decay trends for protein loading and catalytic activity, respectively.

2.8. Characterization of Immobilized TcDAEase

The biochemical performance of the immobilized enzyme was evaluated following the same methodology as used for the free enzyme. Specifically, temperature profiles were measured at 50–90 °C, while pH activity profiles were assessed at 60 °C using the buffer systems described in Section 2.5. Operational stability was evaluated through repeated batch conversions. In each cycle, immobilized TcDAEase was incubated in 50 g/L D-fructose solution (pH 8.0, containing 5 mM MgCl₂) at 60 °C with shaking at 100 rpm for 2 h. After each cycle, the catalyst was recovered by centrifugation, washed with phosphate buffer (pH 8.0) and reused. This procedure was repeated for seven consecutive cycles. The residual activity was calculated based on the initial activity measured in the first batch, which was set as 100%.

2.9. Production of D-Allulose by a Continuous Packed-Bed Reactor

A jacketed packed-bed column reactor (2.0 cm inner diameter × 30 cm height) was filled with 30 g of immobilized TcDAEase (particle size 0.5–1.0 mm; bulk density ~0.7 g/mL). The system was maintained at a constant temperature of 60 °C using a circulating water-ethanol bath (70:30 v/v). The substrate solution, comprising 500 g/L D-fructose and 5 mM MgCl₂ in 50 mM sodium phosphate buffer (pH 8.0), was pumped through the reactor at a flow rate of 30 mL/min, resulting in an empty-bed residence time (EBRT) of 1.3 min. The total volume of substrate solution delivered to the reactor was approximately 43.2 L per day. Effluent samples were collected every 24 h, filtered through 0.22 μm membranes, and analyzed by HPLC as described in Section 2.4 to determine substrate conversion and residual enzyme activity.

3. Results and Discussion

3.1. Amino Acid Sequence Alignment of TcDAEase

TcDAEase exhibited considerable sequence divergence from previously characterized DAEases. As shown in Figure 1, TcDAEase shared amino acid sequence identities from 26.12% to 48.08% with other known DAEases, with the highest similarity observed for the enzyme derived from Arthrobacter globiformis. Domain architecture analysis classified TcDAEase as a member of the xylose isomerase-like family (IPR036237). Multiple sequence alignment (Figure 2) revealed several conserved residues that are functionally important for the isomerization of D-fructose [30]. These include Glu149, Asp182, His208, and Glu243 [31], which participate in metal ion coordination, and residues Glu156, His185, and Arg214, which are involved in anchoring the O1, O2, and O3 positions of D-fructose [32]. These findings suggest that TcDAEase preserves the core structural features essential for efficient ketose epimerization.

3.2. Expression and Purification of Recombinant TcDAEase

The TcDAEase gene was successfully overexpressed in E. coli BL21 (DE3) and highly purified via Ni²⁺-NTA chromatography. As shown in Figure 3 and Figure S1, the SDS-PAGE analysis revealed a distinct protein band at approximately 33 kDa; this result is consistent with the calculated molecular mass of 33.7 kDa. Although minor faint bands below 30 kDa were observed, the major band indicates high purity and successful expression of the target protein. The enzyme’s molecular size is within the range typically associated with favorable stability and diffusion properties, which may facilitate its integration into immobilized systems or continuous flow reactors for large-scale D-allulose production. Notably, the purified enzyme retained high catalytic activity (7.9 ± 0.3 U·mg⁻¹) in the absence of externally supplemented metal ions, suggesting that essential cofactors may have remained bound during expression and purification. A similar phenomenon has also been observed in DAEases from Arthrobacter globiformis M30 [13] and Clostridium bolteae [14], which exhibited baseline activity under metal-free conditions, although their maximal activity was achieved upon Co²⁺ addition. In comparison with these enzymes, TcDAEase demonstrates a favorable expression yield and robust activity profile, highlighting its practical advantages for industrial application.

3.3. Thermostability, pH Tolerance and Metal Ion Dependence

The enzymatic properties of TcDAEase were evaluated under the conditions described in Section 2.4. As shown in Figure 4a, TcDAEase exhibited optimal activity at 80 °C and retained over 70% of its activity within the 60–90 °C range. Thermal inactivation analysis (Figure 4b) revealed half-lives of 720 min at 60 °C, 300 min at 70 °C, and 40 min at 80 °C, indicating superior thermostability compared to most previously reported DAEases. Specifically, TcDAEase offered a favorable balance between stability and activity (Table 1). The DaeM-derived enzyme [16] showed the highest thermal stability (t_1/2 = 9900 min at 60 °C) but low catalytic activity (1.14 U/mg). The enzyme from Caballeronia fortuita [33] exhibited high catalytic activity (270 U/mg) but poor thermostability (t_1/2 = 63 min at 60 °C). TcDAEase showed a moderate activity of 7.9 U/mg and a t_1/2 of 720 min at 60 °C, making it well-suited for long-term high-temperature applications.

In terms of pH tolerance, TcDAEase displayed a broad activity profile with an optimum at pH 10.0. Notably, more than 70% of its maximal activity was retained across the pH range of 8.0–13.0, whereas activity declined sharply below pH 5.0 (Figure 4c). Long-term pH stability tests (Figure 4d) showed that the enzyme remained stable after 6 h incubation at 60 °C between pH 7.0 and 9.0. Although TcDAEase tolerated strongly alkaline conditions, maintaining the reaction environment below pH 9.0 is advisable to avoid non-enzymatic browning, particularly in food-related processes [34]. Therefore, we choose the pH = 8.0 for all subsequent experiments, which provides the highest enzymatic activity and maintain good enzymatic stability.

The effects of various metal ions on enzymatic activity were assessed by supplementing 1 mM of different divalent cations (Figure 4e). Among them, Ni²⁺, Co²⁺, and Mg²⁺ enhanced catalytic activity by 30.3%, 20.2%, and 14.3%, respectively. While Ni²⁺ exhibited the strongest activation effect, its toxicity and regulatory limitations preclude its application in food-grade systems [18]. In contrast, Mg²⁺ is FDA-GRAS approved and thus suitable for industrial use. Further investigations on Mg²⁺ concentration (Figure 4f) revealed that enzymatic activity increased with Mg²⁺ concentration and plateaued beyond 5 mM. And the enzyme activity increased by approximately 2.1-fold when 5 mM Mg²⁺ was added, compared to without Mg²⁺. Accordingly, 5 mM Mg²⁺ was selected as the optimal cofactor concentration for continuous D-allulose production. Taken together, the combination of high thermostability, broad alkaline tolerance, and Mg²⁺-activation distinguishes TcDAEase from most previously characterized DAEases, reinforcing its potential for robust, food-compatible D-allulose production under industrial conditions.

Table 1. Comparison of biochemical properties of reported DAEases and DTEases.

Strain of Enzyme Source	Optimal Temperature (°C)	Optimal pH	Metal Ion	Half-Life	Catalytic Activity (U/mg)	Reference
Pseudomonas sp.	60	7.0–9.0	NR	NR	NR	[35]
Pseudomonas cichorii	60	7.5	NR	NR	NR	[36]
Agrobacterium tumefaciens	50	8.0	Mn2+	63.5 min/50 °C	8.89	[37]
Sinorhizobium sp.	40	8.5	Mn2+	11,050 min/35 °C, 934 min/40 °C, 251 min/45 °C, 1151 min/50 °C	NR	[38]
Rhodobacter sphaeroides	40	9.0	Mn2+	NR	NR	[39]
Clostridium cellulolyticum	55	8.0	Co2+	408 min/60 °C	NR	[40]
Ruminococcus sp.	60	7.5–8.0	Mn2+	96 min/60 °C	8.95	[15]
Clostridium sp.	65	8.0	Co2+	15 min/60 °C	NR	[41]
Clostridium scindens	60	7.5	Mn2+	108 min/50 °C	NR	[42]
Desmospora sp.	60	7.5	Co2+	NR	NR	[43]
Clostridium bolteae	55	7.0	Co2+	156 min/55 °C	NR	[14]
Dorea sp.	70	6.0	Co2+	NR	803	[44]
Treponema primitia	70	8.0	Co2+	30 min/60 °C	NR	[31]
Flavonifractor plautii	65	7.0	Co2+	40 min/65 °C	20	[32]
Arthrobacter globiformis	70	7.0–8.0	Mg2+	NR	23.6	[13]
Agrobacterium sp.	55–60	7.5–8.0	Co2+	267 min/55 °C, 28.2 min/60 °C, 3.8 min/65 °C	90.5	[45]
Corynebacterium glutamicum	55	8.0	Mn2+	140 min/60 °C	22.7	[46]
DaeM	80	7.0	Co2+	9900 min/60 °C, 3240 min/70 °C, 49 min/80 °C	1.14 ± 0.048	[16]
Caballeronia fortuita	65	7.5	Co2+	427.8 min/50 °C, 307.8 min/55 °C, 63 min/60 °C	270 ± 1.5	[33]
Arthrobacter psychrolactophilus	70	8.5	Mg2+	128.4 min/70 °C	14.3	[47]
Blautia produca	55	8.0	Mn2+	180 min/55 °C	1.76 ± 0.099	[48]
Thermogemmatispora carboxidivorans	80	10.0	Ni2+	720 min/60 °C, 300 min/70 °C, 40 min/80 °C	7.9 ± 0.3	This work

NR, not reported.

Table 1. Comparison of biochemical properties of reported DAEases and DTEases.

Strain of Enzyme Source	Optimal Temperature (°C)	Optimal pH	Metal Ion	Half-Life	Catalytic Activity (U/mg)	Reference
Pseudomonas sp.	60	7.0–9.0	NR	NR	NR	[35]
Pseudomonas cichorii	60	7.5	NR	NR	NR	[36]
Agrobacterium tumefaciens	50	8.0	Mn2+	63.5 min/50 °C	8.89	[37]
Sinorhizobium sp.	40	8.5	Mn2+	11,050 min/35 °C, 934 min/40 °C, 251 min/45 °C, 1151 min/50 °C	NR	[38]
Rhodobacter sphaeroides	40	9.0	Mn2+	NR	NR	[39]
Clostridium cellulolyticum	55	8.0	Co2+	408 min/60 °C	NR	[40]
Ruminococcus sp.	60	7.5–8.0	Mn2+	96 min/60 °C	8.95	[15]
Clostridium sp.	65	8.0	Co2+	15 min/60 °C	NR	[41]
Clostridium scindens	60	7.5	Mn2+	108 min/50 °C	NR	[42]
Desmospora sp.	60	7.5	Co2+	NR	NR	[43]
Clostridium bolteae	55	7.0	Co2+	156 min/55 °C	NR	[14]
Dorea sp.	70	6.0	Co2+	NR	803	[44]
Treponema primitia	70	8.0	Co2+	30 min/60 °C	NR	[31]
Flavonifractor plautii	65	7.0	Co2+	40 min/65 °C	20	[32]
Arthrobacter globiformis	70	7.0–8.0	Mg2+	NR	23.6	[13]
Agrobacterium sp.	55–60	7.5–8.0	Co2+	267 min/55 °C, 28.2 min/60 °C, 3.8 min/65 °C	90.5	[45]
Corynebacterium glutamicum	55	8.0	Mn2+	140 min/60 °C	22.7	[46]
DaeM	80	7.0	Co2+	9900 min/60 °C, 3240 min/70 °C, 49 min/80 °C	1.14 ± 0.048	[16]
Caballeronia fortuita	65	7.5	Co2+	427.8 min/50 °C, 307.8 min/55 °C, 63 min/60 °C	270 ± 1.5	[33]
Arthrobacter psychrolactophilus	70	8.5	Mg2+	128.4 min/70 °C	14.3	[47]
Blautia produca	55	8.0	Mn2+	180 min/55 °C	1.76 ± 0.099	[48]
Thermogemmatispora carboxidivorans	80	10.0	Ni2+	720 min/60 °C, 300 min/70 °C, 40 min/80 °C	7.9 ± 0.3	This work

NR, not reported.

3.4. Substrate Specificity and Kinetic Parameters

The substrate specificity of TcDAEase was evaluated using four ketohexoses: D-allulose, D-fructose, D-tagatose, and D-sorbose. As shown in Figure 5, TcDAEase exhibited the highest specific activity toward D-allulose (15.9 ± 0.2 U·mg⁻¹), followed by D-fructose, D-tagatose, and D-sorbose. This substrate preference supports the hypothesis of a conserved epimerization mechanism among DAEases and is consistent with previous reports on enzymes from Arthrobacter globiformis M30 [13], Clostridium scindens [42], Ruminococcus sp. [15], Clostridium sp. [41], and Corynebacterium glutamicum [46], all of which showed a similar catalytic bias toward D-allulose. These findings indicate a conserved substrate recognition pattern among DAEases that likely underpins a shared catalytic mechanism [49].

The kinetic parameters of TcDAEase toward D-fructose were further determined by Michaelis–Menten analysis. The enzyme exhibited a K_m of 260.9 ± 7.4 mM, a k_cat of 96.6 ± 0.5 s⁻¹, and a catalytic efficiency k_cat/K_m of 0.38 ± 0.02 s⁻¹·mM⁻¹. A comparison with previously reported DAEases (

Table 2. Comparison of the kinetic parameters of reported DAEases and DPEases.

Enzyme Source (Strain)	kcat (s−1)	Km (mM)	kcat/Km (s−1·mM−1)	Conversion Rate (%)	Reference
Agrobacterium tumefaciens	34.46 ± 0.46	24 ± 0.15	1.41 ± 0.02	32.9	[37]
Clostridium cellulolyticum	55.91 ± 0.787	53.5 ± 1.8	1.045 ± 0.025	29	[40]
Ruminococcus sp.	59.4 ± 0.483	216 ± 2	0.267 ± 0.017	28	[15]
Clostridium sp.	272.87 ± 1.37	279 ± 10.8	0.978 ± 0.065	33	[41]
Clostridium scindens	5.83 ± 0.43	40.1 ± 2.5	0.145 ± 0.02	30	[42]
Clostridium bolteae	59 ± 1	59.8 ± 8.2	0.99 ± 0.05	28.8	[14]
Dorea sp.	5507.45	153	3.31	30	[44]
Treponema primitia	292.88 ± 6.48	279 ± 20.7	1.05 ± 0.097	28	[31]
Flavonifractor plautii	225,280	162	156	31	[32]
Arthrobacter globiformis	41.8	37.5	1.12	26	[13]
Agrobacterium sp.	106.7	110	1.06	30	[45]
Corynebacterium glutamicum	40.92 ± 0.175	366.7 ± 10.5	0.11 ± 0.077	30	[46]
Bacillus subtilis	41.99 ± 2.3	141.43 ± 7.5	0.29 ± 0.09	31	[16]
Caballeronia fortuita	157.21 ± 0.53	81.9 ± 4.3	1.30 ± 0.12	29.4	[33]
Arthrobacter psychrolactophilus	2920	738.7	3.953	27	[47]
Blautia produca	65.03 ± 3.56	235.7 ± 9.938	0.28 ± 0.014	30	[48]
Thermogemmatispora carboxidivorans	96.6 ± 0.5	260.9 ± 7.4	0.38 ± 0.02	30	This work

) shows that while the kinetic performance of TcDAEase is comparable to other characterized enzymes, its combination of high activity and exceptional thermostability renders it a promising biocatalyst for large-scale, continuous D-allulose production in industrial settings.

3.5. Immobilization Optimization and Stability Assessment of TcDAease

To achieve large-scale production of D-allulose, we optimized the immobilization process for TcDAEase and used whole-cell immobilization techniques while circumventing the cost of enzyme purification by using the natural permeability of E. coli membranes to small molecule substrates [50,51]. Drawing on immobilization studies of glucose isomerase [52,53], we developed a diatomite–chitosan composite carrier that integrates the high mechanical strength of diatomaceous earth with the strong binding capacity of chitosan. The adsorption-crosslinking-embedding process was finely controlled by adjusting pH conditions. Specifically, we used glutaraldehyde pretreatment at pH 3.0 to ensure uniform dispersion of the carrier materials, adjusted the pH to 4.5 to promote adsorption and cross-linking of the cells, and finally adjusted the pH over 5.0; a gradual alkalinization promoted the precipitation of chitosan. Besides, the composite carrier was optimized for compatibility with packed-bed reactor applications.

We first focus on the optimization of chitosan. Chitosan is a bio-compatible binder that enhances the mechanical robustness of the matrix and prevents the leaching of enzymes during manipulation. To improve the operational stability and reproducibility of the immobilized enzyme, three key parameters were systematically evaluated: wet cell loading, chitosan precipitation pH, and chitosan concentration. As shown in Figure 6a, protein loading increased linearly with cell concentration in the range of 0.5–1.52 g/100 mL (slope = 10.3 ± 0.8 mg/100 mL). Beyond this range, loading efficiency declined sharply (slope = 1.8 ± 0.3 mg/100 mL). Breakpoint analysis identified a critical saturation threshold at 1.52 ± 0.10 g/100 mL, which is consistent with the surface saturation of the carrier material [54]. In parallel, enzyme activity recovery exhibited first-order decay kinetics (k = 0.45 ± 0.05). The half-maximal activity was observed at 1.54 g/100 mL, which was closely aligned with the protein loading saturation point. This strong correlation suggests that excessive cell density hinders the effective utilization of the enzyme by restricting the diffusion of crosslinking agents [55]. Additionally, spatial constraints at high cell densities may induce conformational distortions that impede substrate access and ultimately reduce catalytic efficiency [56]. According to these results, 1.5 g/100 mL was selected as the optimal cell loading, yielding a protein loading of 23 ± 0.8 mg/g and approximately 50% activity recovery. This condition offered a balanced trade-off between carrier utilization and enzyme performance. In addition, the pH for chitosan precipitation was identified as a critical factor influencing immobilization efficiency. We embedded them in three pH intervals, acidic (5.0–6.0), neutral (6.5–7.0) and basic (10.0–11.0). Although chitosan’s theoretical precipitation pKa is ~6.3, partial gelation was observed even at pH 5.0–6.0, which is likely due to localized pH fluctuations during NaOH titration under vigorous stirring. As shown in Figure 6b, immobilized TcDAEase prepared at pH 5.0–6.0, 6.5–7.0 and 10.0–11.0 exhibited activities of 1.4 ± 0.3, 6.7 ± 0.2 and 6.1 ± 0.2 U·g⁻¹, corresponding to activity recoveries of 19 ± 0.5%, 49 ± 1.2% and 41 ± 1.3%, respectively. These results indicate that neutral to mildly alkaline conditions significantly enhanced catalytic performance, likely due to improved chitosan precipitation and network formation. In contrast, excessively alkaline conditions may result in over-crosslinking, which can impair enzyme conformation or hinder substrate diffusion [57]. Overall, a neutral pH range (6.5–7.0) was found to provide optimal precipitation efficiency and enzymatic activity, potentially due to the favorable protonation state of chitosan for controlled gelation and Schiff base (CH=N) formation [58]. The concentration of chitosan is also one of the key factors affecting performance. Too low a concentration can result in an insufficiently intact matrix, which can weaken the particle strength. Conversely, too high a concentration will result in the formation of a dense gel, which will limit matrix diffusion and thus reduce apparent activity [59]. As shown in Figure 6c,d, the highest initial activity (9.7 ± 0.9 U·g⁻¹) was observed at a chitosan concentration of 5 g/L. However, poor operational stability was noted under this condition, with rapid activity decay (k = 0.24 ± 0.02 cycle⁻¹, t_1/2 = 2.9 cycles, R² = 0.98). In contrast, increasing the chitosan concentration to 10 g/L significantly enhanced stability (k = 0.01 ± 0.003 cycle⁻¹, t_1/2 = 69.3 cycles, R² = 0.91). Under these optimized conditions, 90.3 ± 1.5% of the initial activity was retained after seven consecutive cycles. This corresponds to a 24-fold reduction in the rate of inactivation compared to the 5 g/L condition. These improvements align with previous findings suggesting that a critical chitosan concentration (~8 g/L) is required to establish a percolation threshold for network formation on silica-based supports [60].

In summary, under optimized conditions (wet cell loading: 15 g/L; precipitation pH: 6.5–7.0; chitosan concentration: 10 g/L), the immobilized TcDAEase achieved 49 ± 1.2% activity recovery and demonstrated favorable mechanical integrity. As shown in Figure 7, the enzyme prepared via the three-step method exhibited excellent operational stability, with no detectable protein leaching after drying and repeated washing at 60 °C. These findings confirm the successful formation of a robust immobilization matrix, whose particles after filtration are shown in Figure S2.

3.6. Biochemical Properties of the Immobilized TcDAEase

The biochemical properties of immobilized TcDAEase were evaluated under optimized conditions. As shown in Figure 8a, the immobilized enzyme retained over 70% of its relative activity across the temperature range of 60–90 °C, with peak activity (over 90%) between 70 and 90 °C. Compared to the free enzyme which exhibited rapid activity loss above 80 °C, the immobilized form displayed enhanced thermal tolerance. Figure 8b illustrates that TcDAEase also retained high activity under strongly alkaline conditions, with a pH optimum above 10.0. However, from an industrial perspective, extremely high temperatures and pH values pose challenges such as increased energy consumption and the risk of non-enzymatic browning reactions (e.g., Maillard reaction), which can compromise product quality [34]. Combining the biochemical properties of immobilized TcDAEase derived from our experiments and the challenges of real food production, 60 °C and pH 8.0 were chosen as the optimal operating conditions. Under these conditions, a balance between catalytic performance, stability and process feasibility can be achieved.

To assess operational stability, reusability tests were conducted under these conditions (60 °C, pH 8.0, 5 mM Mg²⁺). As shown in Figure 9, immobilized TcDAEase retained 90.3% ± 1.5% of its initial activity after seven consecutive cycles—outperforming commonly used agarose- and alginate-based systems, which typically retain only 40–70% activity under similar conditions [61]. This exceptional stability is likely attributed to the synergistic effects of the composite matrix. The glutaraldehyde-crosslinked chitosan network restricts conformational mobility and protects against thermal denaturation, while the porous diatomaceous earth framework provides a large surface area and abundant binding sites, mitigating enzyme leaching and aggregation [62]. Although structural analyses such as SEM or BET were not performed, the physical properties of the composite carrier were consistent with those of food-grade diatomite materials reported in previous literature [55]. Furthermore, the hydrophilic amino and hydroxyl groups of chitosan may create a favorable microenvironment that shields the enzyme from interfacial stresses during packed-bed operation [63].

Together, these results confirm that the chitosan–diatomaceous earth matrix imparts excellent thermal and operational stability to TcDAEase, supporting its application in continuous, food-compatible, and cost-effective D-allulose biomanufacturing.

3.7. Long-Term Continuous D-Allulose Production in Packed-Bed Reactor

As a result of the above findings, 30 g of immobilized TcDAEase was loaded into a packed-bed reactor (Figure 10a) under fixed reaction conditions (60 °C, pH 8.0, 5 mM Mg²⁺, 500 g/L D-fructose, and a flow rate of 30 mL/min) for long-term production testing. The recombinant TcDAEase exhibited a substrate conversion rate of 30%, comparable to most reported DAEases (26–32.9%; Table 2), indicating sufficient catalytic efficiency for potential industrial application. The total volume of substrate solution delivered to the reactor was approximately 43.2 L per day. Over 150 consecutive days of operation, the system produced a cumulative total of 756 kg of D-allulose, corresponding to 89.7% of the theoretical maximum (842.2 kg) yield. This discrepancy could be attributed to multiple factors, including potential side reactions such as non-specific isomerization [32] of D-fructose into undesired products like D-glucose or D-mannose under suboptimal pH or temperature [19]. In addition, gradual loss of enzyme activity during prolonged operation and minor handling losses throughout the process may also contribute to the observed deviation.

Throughout the trial, the reactor exhibited stable hydraulic performance with no detectable pressure buildup or clogging, indicating efficient mass transfer and the mechanical robustness of the immobilized matrix. As shown in Figure 10b, the specific activity of TcDAEase gradually decreased by about 30% (from 6.5 to 4.5 U·g⁻¹), and the substrate conversion rate declined from 26.0 ± 1.2% to 22.5 ± 1.1%. These results confirm the long-term operational stability and excellent performance of the immobilized TcDAEase system for continuous, food-grade D-allulose production.

This system notably outperformed previously reported immobilized enzyme platforms. For instance, Takeshita et al. [22] achieved a 60-day operation using 200 g of immobilized D-tagatose 3-epimerase (70 U·g⁻¹), yielding only 20 kg of D-allulose at pH 7.0. Yoshihara et al. [13] reported continuous production using DAEase from Arthrobacter globiformis immobilized on an ion-exchange resin, which produced 215 kg of D-allulose per liter of catalyst over four months at 55 °C and pH 7.5.

Collectively, these findings underscore the scalability, durability, and industrial viability of the TcDAEase-based biocatalytic system for large-scale, continuous D-allulose manufacturing under food-compatible conditions.

4. Conclusions

In this study, we systematically investigated the catalytic properties and industrial potential of a thermostable D-allulose 3-epimerase (TcDAEase) from Thermogemmatispora carboxidivorans. The enzyme exhibited outstanding thermostability, broad pH tolerance, and high catalytic activity toward D-allulose and D-fructose, even in the absence of metal ion supplementation. Among the tested cofactors, food-grade Mg²⁺ significantly enhanced enzymatic activity, emphasizing the suitability of TcDAEase for food-grade rare sugar production. To improve practical applicability, TcDAEase was immobilized on a chitosan–diatomaceous earth matrix, resulting in favorable activity recovery and excellent operational stability. The immobilized biocatalyst achieved an activity recovery of 48.7% ± 2.4% and retained 90.3% ± 1.5% of its initial activity after seven consecutive cycles. Remarkably, the immobilized enzyme enabled continuous D-allulose production in a packed-bed reactor for 150 days with minimal decline (approximately 4%) in conversion efficiency, outperforming several previously reported systems. These results demonstrate that the integration of enzyme thermostability, food-compatible activation, and cost-effective immobilization offers a promising strategy for the scalable and sustainable production of D-allulose.