Simulation-guided enzyme discovery: A new microbial source of cellobiose 2-epimerase

Abstract

Cellobiose 2-epimerase (CE) is a promising industrial enzyme that can be utilized in the dairy industry. More thermostable CEs from different microorganisms are still needed for a higher lactulose productivity. This study demonstrated the feasibility to use molecular dynamics (MD) simulation as the preliminary computational filter for thermostable enzymes screening. Sequence information of eleven uncharacterized CEs were chosen to be analyzed by MD simulations. The CE from Dictyoglomus thermophilum (Dith-CE) was determined experimentally to be one of the most thermostable CEs with the highest epimerization (160 ± 6.5 U mg⁻¹) and isomerization activities (3.52 ± 0.23 U mg⁻¹) among all the reported CEs. This enzyme shows the highest isomerization activity at 85 °C and pH 7.0. The kinetic parameters (k_cat and K_m) of isomerization activity of this CE are 3.98 ± 0.3 s⁻¹ and 235.2 ± 11.2 mM, respectively. These results suggest that the CE from Dith-CE is a promising lactulose-producing enzyme.

1. Introduction

Current food industrial processes are conducted with an increased awareness for environmental impact and energy saving. Enzymes are environmentally friendly biocatalysts, some of which can be used to accelerate food production process with great efficiency. Most industrially relevant enzymes originate from microorganisms [1]. Generally, the enzymes from different species show different catalytic properties. However, only a small fraction of them satisfy the needs of industrial process. Due to advances in biotechnology, rational design and directed evolution are currently popular strategies to improve enzymatic properties. However, these experimental approaches usually require detailed knowledge of specific enzymes and could be laborious. Generally, we cannot expect to find mutants with highly improved properties in the sequence vicinity of a parent enzyme.

Cellobiose 2-epimerase (EC 5.1.3.11) is a deeply mined industrial enzyme which can catalyze the interconversions of β-l,4-linked disaccharides at the C2 position of the reducing ends. CEs from nineteen bacteria have been extensively characterized in search for higher catalytic activities under different industrial conditions [2], CE is more profitable economically when using lactose as the substrate. Different CEs can produce both epilactose and lactulose with different preference as their epimeric and isomeric products, respectively. Both these two derivatives have beneficial biological functions. Lactulose is more widely used in food and pharmaceutical industries. It is thermodynamically more favorable and can deliver higher yields using thermostable CE than epilactose. That is, the product selectivity controlled by CE is temperature-dependent. Therefore, the key to the discovery of a lactulose-producing CE with a high yield is to discover the CEs with higher thermostabilities.

To date, the most efficient lactulose-producing CE originated from Caldicellulosiruptor saccharolyticus (Casa-CE) [3]. Casa-CE has been studied as a model enzyme to analyze lactulose productivity catalyzed by CE [4–9]. And several mutations of Casa-CE were designed to improve the isomerization activity of Casa-CE [10–12]. We also discovered a lactulose-producing CE from Caldicellulosiruptor obsidiansis (Caob-CE). However, the improvement of its catalytic properties was not significant compared to Casa-CE, and the activities of these two CEs is both not satisfactory [13]. Therefore, new microbial resources of CE are still needed.

The enzyme diversity caused by microbial diversity is a great source to search new enzymes that meet industrial demands. A common strategy for seeking novel enzymes with desired properties is to discover new enzymes based on the growth environments of the organisms. The enzymes from extremophiles are adapted to their growth environments such as extremes of temperature, salinity, pH, pressure, and solvent conditions. Thermostable enzymes generally originate from organisms thrive in extreme hot niches. However, the correlation between thermal stability and microbial growth temperature is weak, it remains a challenging task to compare the thermal stability of homologous enzymes from numerous thermophiles. Thus, vast characterization work is required to discover desirable enzymes in previous researches.

In this study, we used a in silico approach to explore and predict the thermostabilities of 11 uncharacterized CEs orthologs from different bacterial species. Conventional molecular dynamic (MD) simulations were carried out to measure the flexibilities of CE orthologs. The back-bone root-mean-square deviation (RMSD) values of the backbone atom positions and hydrogen bonds of the uncharacterized CEs calculated by MD simulation were used as indicators of structural flexibilities. The MD analysis provides a powerful method of examining the flexibilities of the uncharacterized orthologous proteins. The thermostabilities of CE orthologs were then predicted by comparing the indicators from MD simulation. The predicted most thermostable CE sequence from Dictyoglomus thermophilum (Dith-CE) was cloned and recombinantly produced in Escherichia coli. We then biochemically characterized Dith-CE in comparison with other CEs.

2. Materials and methods

2.1. Materials and chemicals

Unless otherwise stated, all materials and reagents were of analytical grade or higher purchased from Sinopharm Chemical Reagent (Shanghai, China). Epilactose and lactulose were of the highest grade and acquired from Sigma (St Louis, MO, USA). Isopropyl β-D-l-thiogalactopyranoside (IPTG), ampicillin sodium and other chemicals used for enzyme assays and characterizations were purchased from Sangon Biotech (Shanghai, China). Electrophoresis reagents were supplied by Bio-Rad. The competent cells of bacterial strain Escherichia coli BL21 (DE3) pLysS was purchased from Promega (Madison, WI). Chelating Sepharose Fast Flow resin was from GE Healthcare (Uppsala, Sweden).

2.2. Gene cloning

The gene encoding CE originated from Dith was subcloned into pET-22b (+) plasmid which was purchased from Sangon Biotech (Shanghai, China), and two restriction sites of NdeI and XhoI were inserted atthe 5′-terminal and 3′-terminal of the CE gene, respectively. An in-frame 6× His-tag sequence was fused at the C-terminal sequence of the open reading frame.

2.3. Cultivation and purification of recombinant Dith-CE

The recombinant pET-22b (+) plasmid was transformed into the competent cells of E. coli BL21 (DE3). The recombinant E. coli cells were cultivated in 0.4 L of LB medium containing 10 g L⁻¹ of tryptone, 5 g L⁻¹ of yeast extract, 10 g L⁻¹ of NaCl, and 100 mg L⁻¹ of ampicillin with shaking (200 rpm) at 37 °C. IPTG was then added into the culture with a final concentration of 0.5 mM when the OD₆₀₀ reached 0.6–0.8. After 6 h of fermentation at 28 °C by shaking (200 rpm), the cells were harvested by centrifugation at 8000 ×g for 5 min.

The cell pellets were disrupted by sonication in 100 mM NaCl, 50 mM Na₂HPO₄-NaH₂PO₄ buffer (pH 7.0) and then centrifuged at 8000 ×g for 15 min at 4 °C. The supernatant was filtered through a 0.45 μm filter to collect crude enzyme. The crude enzyme was purified by one-step nickel-affinity chromatography (Novagen). The purified enzyme solution was dialyzed overnight against 50 mM Na₂HPO₄-NaH₂PO₄ buffer (pH 7.0). The molecular mass and purity of the purified enzyme was checked by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and the concentration was measured by using a Nanodrop 2000C spectrophotometer.

2.4. Enzyme assay

One unit (U) of CE epimerization activity was defined as the amount of enzyme required to produce 1 μmol epilactose per min at pH 7.0 and 75 °C. The reaction mixture of 1 mL volume included 200 mM lactose, 50 mM Na₂HPO₄-NaH₂PO₄ buffer (pH 7.0) and 20 μg purified enzyme. The reaction was performed at 75 °C for 15 min and was terminated by addition of HC1 to the reaction mixture at a final concentration of 300 mM. One unit (U) of CE isomerization activity was defined as the amount of enzyme required to produce 1 mol lactulose per min at pH 7.0 and 85 °C. The reaction mixture of 0.5 mL volume included 200 mM lactose, 50 mM Na₂HPO₄-NaH₂PO₄ buffer (pH 7.0), and 250 μg enzyme. The reactions were performed at 85 °C for 1 h and were terminated by addition of HC1 to the reaction mixture at a final concentration of 300 mM. The reaction mixture was diluted ten times and analyzed by an HPLC (Agilent 1200 system, Agilent technologies, CA, USA) equipped with a refractive index detector and an Asahipak NH₂P-50–4E column (4.6 mm id × 250 mm, Shodex, Tokyo, Japan). The column was eluted with acetonitrile/methanol/water (75:15:10, v/v/v) at a flow rate of 1 mL min⁻¹ and 40 °C.

2.5. Biochemical characterization

Following the procedures described above, the effect of pH on CE activity was investigated at 75 °C in 50 mM NaOAc-HOAc buffer (pH 4.0–6.0), 50 mM Na₂HPO₄-NaH₂PO₄ buffer (pH 6.0–7.5) and 50 mM Britton-Robinson (pH 7.5–8.5). The effect of temperature on epimerization and isomerization activity of CE was investigated at 50–80 °C and 65–90 °C in 50 mM Na₂HPO₄-NaH₂PO₄ buffer (pH 7.0), respectively. The thermostability of the purified enzyme was obtained by detecting the incubation time versus residual activity at different temperatures, ranging from 70 to 85 °C.

The kinetic parameters of CE epimerization activity were determined using 100–800 mM of lactose and 100–800 mM of cellobiose by measuring the epilactose, respectively. Moreover, the kinetic parameters of CE isomerization activity were determined using 20–400 mM of lactose by measuring the lactulose. The parameters were measured by Lineweaver-Burk plots from the Michaelis-Menten equation included the Michaelis-Menten constant (K_m) and turnover number (k_cat).

2.6. Thermostability measurement

The thermostability of the purified enzyme was obtained by detecting the incubation time versus residual activity at different temperatures, ranging from 70 to 85 °C. The melting temperature of CE was determined using a Q2000 DSC (TA Instruments Nano DSC) and the heating temperature was from 35 to 100 °C at a scan rate of 1 °C min⁻¹. The dialyzed buffer was collected to serve as the reference.

2.7. Sequence and 3D structures of CE

Eleven putative CE sequences were randomly selected from The Universal Protein Resource Protein knowledgebase (UniProtKB) [14]. The 3D structures of uncharacterized CEs and Dictyoglomus turgidum (Ditu-CE) were homologically modeled in the SWISSMODEL online server [15]. The known crystal structures CEs (Casa-CE: PDB ID 4Z4J and CE from Ruminococcus albus (Rual-CE): PDB ID 3VW5) were obtained from RCSB PDB database [16]. These structures were used as the starting models for MD simulation.

2.8. MD simulation

MD simulations were performed by using GROMACS (Version 2018.1) [17] with GROMOS96 54a7 force field [18]. Each system was solvated with SPC/E water and 150 mM NaCl, then was relaxed through energy minimization until a potential energy gradient ∆E less than 1000 kJ mol⁻¹ nm⁻¹ was achieved in order to avoid bad molecular contacts. We conducted a 1 -ns NVT ensemble equilibration followed by a 1 -ns NPT ensemble equilibration at 353 K and 1 bar pressure after energy minimization. Periodic boundary condition was used in each simulation. Parrinello-Rahman and velocity rescaling thermostat were used for pressure and temperature controls, respectively. All bonds were constrained using the LINCS algorithm. A position restraining force with the constant of 1000 kJ mol⁻¹ nm⁻² was applied on the heavy atoms of the protein. The PME calculations were calculated by graphical processing units (GPU), which speeded up the simulation substantially. And the nonbonded interactions were calculated by the L-J model with a cutoff distance of 14 Ǻ. The final outputs of the NPT simulations were then subjected to 50 ns run of equilibrium MD simulations without position restraints. The parameters of the final simulation were as same as that of NPT simulations except running time and position restraints. Three replications were done for each simulation. The hydrogen bonds and RMSD of backbone atom positions of each trajectory (10–50 ns) were calculated from the starting structure using GROMACS analysis tools.

3. Results

3.1. Amino acid sequence alignment

These uncharacterized CE sequences and their entry numbers were listed in Table 1. All these CE sequences originate from different genera. The amino acid sequences of the characterized and uncharacterized CEs were compared. Most of the amino acid sequence identities range from 30% to 50% (Table S1). The crystal structures study of CE from Rhodothermus marinus (Rhma-CE) proposed that His-200, His-259, His-390 were as reversible general acid/base catalysts in epimerization [19]. Seventeen residues in all the CEs we selected for amino acid sequence alignment (Fig. S1), including His-259, His-390, and Arg-66 of Rhma-CE, are completely conserved. However, the histidine corresponding to His-200 of Rhma-CE was not found in CE from Dysgonomonas gadei. This CE exhibited the lowest amino acid sequence identities with the other selected CEs (Table S1). Dith-CE has a higher identity with Ditu-CE [20], in accordance with the fact that they belong to the same genera.

Table 1.

Uncharacterized CE sequences from various microorganisms.

Microorganisms	Uniprot entry no.	AA sequence length
Alistipes sp.	A0A143XWV2	397
Bacteroides uniformis	I9U202	411
Dictyoglomus thermophilum	B5YCW2	389
Lacunisphaera limnophila	A0A1D8AV08	404
Paenibacillus polymyxa	E3EBX8	391
Roseburia hominis	A0A174HAT2	410
Sphingobacterium sp.	F4CC56	387
Thermoanaerobacterium thermosaccharolyticum	L0IL31	394
Treponema brennaborense	F4LJ26	418
Verrucomicrobia bacterium	A0A1V5L3Y5	430
Vibrio fumissii	F0LXX0	406

3.2. Thermostability prediction of CE by MD simulation

The RMSD value can quantify backbone atom movements of the protein. Fig. 1 is the RMSD of backbone atom positions of Dith-CE with respect to time. We chose the starting structures of simulations as reference rather than homologically modeled structures, because they have been well-equilibrated with solvent at the desired temperature and pressure. As shown in Fig. 1, the RMSD of the backbone reaches equilibrium within 5 ns of simulation time and remains stable over the remainder of the MD simulation. The global flexibility of CE structure can be evaluated by the average RMSD values calculated from MD simulation. The average RMSD values of CEs after equilibration (10–50 ns) were selected to determine the flexibilities. The formed hydrogen bonds divided by the length of the corresponding sequence of CEs (hbond/aa) were calculated from each frame in MD simulation. As shown in Fig. 2. Dith-CE exhibit the lowest average RMSD value and the highest average hbond/aa value among those of all the CEs calculated in this study, which means that the Dith-CE with the most rigid structure might have the best thermostability. The enzymatic properties of protein encoded by Dith-CE were experimentally measured in the subsequent work. Thth-CE with a predicted lower thermostability was also characterized in this study.

Fig. 1.

RMSD curve of Dith-CE obtained from MD simulations at 297 K and 353 K.

Fig. 2.

Average RMSD values of different CEs calculated from MD simulations

3.3. Expression and activity confirmation of Dith-CE and Thth-CE

The recombinant plasmids with Dith-CE and Thth-CE sequences were expressed by E. coli, respectively. The purified proteins both exhibited strong bands of approximately 46 kDa on SDS-PAGE (Fig. 3), which were consistent with the predicted molecular mass of 46,421 Da and 46,373 Da, respectively. The native proteins were determined as a 46 kDa monomer by both SDS-PAGE and gel filtration analysis (data not shown). Then the purified proteins were used for confirmation of CE activity. Quantitative HPLC analyses showed that the two proteins were identified to have CE feature that can convert lactose to epilactose and lactulose. No further substrate investigations of these two enzymes were conducted, because such investigations are beyond the scope of this paper. Moreover, reaction catalyzed by Thth-CE synthesized lactulose in small amounts using lactose as substrate. Thus, we concentrated on analysis of Dith-CE for lactulose production in this study.

Fig. 3.

SDS-PAGE analysis of the CEs: Thth-CE, Dith-CE and Case-CE.

3.4. Effects of pH and temperature on the activity

Dith-CE showed different optimal temperatures for epimerization and isomerization reactions. The highest epimerization and isomerization activities were exhibited by Dith-CE at 75 and 85 °C, respectively (Fig. 4A). The pH effect of Dith-CE was investigated by plotting its epimerization activities at various pH values (Fig. 4B). Dith-CE showed a relatively high tolerance to acid conditions. The pH profile of Dith-CE increased steadily from 4 to 7, and reached its optimal activity at 7.0. In the pH range from 7.5 to 9, the activity dropped quickly.

Fig. 4.

(A) Effect of temperature on activity of Dith-CE. Values are the means of three replications ± standard deviation. (B) Effect of pH on epimerization activity of Dith-CE. Values are the means of three replications ± standard deviation.

3.5. Specific activity and kinetic parameters of Dith-CE

The measured specific activities of Dith-CE converting lactose to epilactose and lactulose were 160 ± 6.5 U mg⁻¹ and 3.52 ± 0.23 U mg⁻¹, respectively (Table 2), which were 77 and 48 percentage points higher than those of Casa-CE. Both epimerization and isomerization activities of Dith-CE were of the highest level. In previous studies, most data of the kinetic parameters of CEs were investigated by lactose and cellobiose epimerization reactions. Here we list some reported kinetic parameters of different CEs for comparison in Table 2. Like most the other CEs, Dith-CE showed a higher catalytic efficiency using cellobiose as substrate, because its k_cat/K_m value for cellobiose is higher than that of lactose. Dith-CE exhibited the highest k_cat value (441.4 ± 15.4 s⁻¹ and 320 ± 14.3 s⁻¹, respectively) for both lactose and cellobiose among all the reported CEs. Dith-CE also had considerable high K_m values, which meant Dith-CE had a bad performance at negligible substrate concentration. However, in practical use, industrial processed are usually conducted at high substrate concentrations. Dith-CE will play to its strengths at nearly saturating substrate concentration. Additionally, we measured the kinetic parameters of Dith-CE and Casa-CE for lactose isomerization by Compared to Casa-CE, Dith-CE exhibited higher k_cat (3.98 ± 0.3 s⁻¹) and K_m (235.2 ±11.2 mM) values than those of Casa-CE (1.66 ± 0.4 s⁻¹ and 144.6 ± 7.2 mM, respectively).

Table 2.

Enzymatic properties of CE orthologs from different species.

CE	Optimum pH	Optimum temperature (°C)	Specific activity (epimerization) (U mg−1)	Specific activity (isomerization) (U mg−1)	Ratio (E/I)	Tm	Lactose			Cellobiose			Reference
							kcat (s−1)	Km (mM)	kcat/Km (mM−1 s−1)	kcat (s−1)	Km (mM)	kcat/Km (mM−1 s−1)
Dith-CE	7	75	160.1 ± 6.5	3.52 ± 0.23	45	92.15	441.4	5232	0.844	320.0	354.9	0.902	This study
Thth-CE	7	60	82.1 ± 3.1	Less than 0.08		70.3	200	420.5	0.478	247.0	186.0	1.328	This study
Casa-CEa	7	65	90.4 ± 4.8	2.38 ±0.18	38	84.14	229.8	491.1	0.468	227.2	350.7	0.648	This study
Caob-CE	7.5	70	118	2.81	42	86.7	233.4	301.7	0.773	233.7	164.5	1.421	25
Rual-CE	7.5	30	38.8b	0.001b	38,800	-	52.1	33	1.6	63.8	13.8	4.6	26
Bafr-CE	7.5	45	76.8b	0.0005b	153,600	-	79.5	6.56	12.1	67.6	3.75	18	27
Fljo-CE	8.4	35	48.5b	0.0005b	97,000	-	17.5	34.9	0.501	39.9	53.2	0.75	28

Bafr, Bacteroides fragilis (Q5LH66); Fljo, Flavobacterium johnsoniae (A5FA14); Rual, Ruminococcus albus (P0DKY4).

^aRemeasured in this study.

^bRemeasured by Kuschel et al. [29].

3.6. Thermostability of Dith-CE

The DSC was used to quantify the thermostabilities of Dith-CE and Thth-CE. The thermostability of estimated CEs was ranked based on T_m, which is a good predictor of protein stability. As shown in Table 2, Dith-CE exhibited the highest T_m value (92.2 °C) among all the estimated CEs, which was considerably higher than the second highest one of Caob-CE (86.7 °C). The RMSD value measured by MD simulation also provided an index of protein flexibility.

The functional stability of Dith-CE was then studied at 70–85 °C The decay constant (k_d) of recombinant Dith-CE were calculated to be 0.00688,0.0250,0.346, and 1.763 h⁻¹ at 70,75, 80 and 85 °C, respectively. And the half-lives (t_1/2) of the enzyme at respective temperatures were 99.0 h, 27.7 h, 2.0 h, 0.4 h. Dith-CE showed the best thermostability at 70 °C among all the reported CEs. It retained 93% and 74% of its initial activity after denaturation at 70 °C and 75 °C for 12 h (Fig. 5). However, the stabilities of Dith-CE became worse at 80 °C and 85 °C, which were comparatively lower than those of Casa-CE, Caob-CE, and Ditu-CE.

Fig. 5.

Effect of temperature on the stability of Dith-CE. (A) The initial activity without preincubation was taken as 100%. Values are means of three replications ± standard deviation. (B) The decay constant (k_d) of Dith-CE at different temperatures, ranging from 70 to 85 °C. The vertical axis is In (A_t/A_o). A_t represents the incubation time versus residual activity of Dith-CE. A_o is the initial activity of Dith-CE without preincubation.

3.7. Lactulose production by Dith-CE

In order to obtain a higher isomerization reaction rate, the purified Dith-CE in high concentration (0.5 mg mL⁻¹) was used. Because high enzyme concentration is essential for efficient lactose isomerization into lactulose [21]. Dith-CE had a significantly higher initial epimerization rate than the isomerization rate. A high amount of epilactose was produced rapidly but a small amount of lactulose was synthesized in the first 10 min of production. The epilactose yield catalyzed by Dith-CE slowly decreased but lactulose was steadily produced. After 4-h productions at 80 °C, 50.7 ± 2.3% and 46.74% of lactulose were produced from 200 g L⁻¹ and 400 g L⁻¹ lactose solutions, respectively. A better efficiency level was achieved by using a higher concentration of lactose because of the high saturating substrate concentration of Dith-CE.

4. Discussion

It is well established that reaction balance of CE between epimerization and isomerization is temperature dependent. Generally, a higher reaction temperature leads to a higher isomerization rate. Therefore, thermostable CEs are useful catalysts for industrial processes not only because they can avoid contaminating microbial growth and increase solubility of reactants, but also because their higher activities at relatively high temperatures, especially isomerization activities. Although mutagenesis of Casa-CE has been made to enhance its thermostability successfully, the improvement was not enough for Casa-CE to work at everlasting higher temperature [12]. It is still a challenge to efficiently screen new thermostable CEs as better-quality parental enzymes.

This study provides an in-silico approach to screen uncharacterized CE from the protein sequence database. The information of CE structures from homology modelling was used to set up the simulation systems. Thus, all we need to predict variation thermostability by MD analysis is the sequence of the proteins provided that thermostability of some orthologs is known from experiment. The simulations at only one temperature were carried out to save the computational cost. Furthermore, the simulations at different temperatures (297 K and 373 K) showed the nearly same pattern but quantitatively different by RMSD analyses. It is also supported by Khan et al. that the RMSD of lipase estimated at lower temperatures were smaller and raised gradually at higher temperatures [22]. The reason why we chose 353 K (80 °C) as a simulation temperature is that 353 K is close to the average thermal denaturation temperature of thermostable CEs, at which the differences in backbone atom positions might be more significant.

The average RMSD value of backbone atom positions of Dith-CE during equilibrium simulation was the lowest among all those of the calculated CEs in this study, which indicated that the overall flexibility of Dith-CE was theoretically the smallest The relationship between flexibility and stability is still debated. It is well accepted that thermophilic enzymes are more rigid than their mesophilic homologues, which helps them to. The RMSD values were only weakly correlated with experimentally determined melting temperatures values in our study. Some possible reasons may be rooted in the poor relationship of flexibility and thermostability, or the relatively large simulation error of this method. Nevertheless, the T_m of Dith-CE is the highest among all CEs we estimated. T_m, the melting temperature, is a metric for the thermostability of a protein. Using MD analysis, we predicted that Dith-CE is one of the most, if not the most, thermostable CEs. RMSD and hydrogen bonds are two of the most fundamental properties to predict whether the protein is stable. Mohamadnezhadi et al. found that the average values of RMSD in equilibrium state for thermophilic enzyme are universally lower compared to mesophilic and psychrophilic enzymes [23]. Besides RMSD value, the generalized order parameter, S, calculated from MD simulation is also correlated with T_m values and can be used to predict thermostability [24].

This article provides a strategy to discover novel thermostable enzymes and this strategy can be used in other enzymes. To our knowledge, this strategy hasn’t been used in other studies, the mainly reason might be the big computational cost of this strategy. RMSD analysis is best used in comparison of homologous enzyme, because they usually have the same structural scaffold and similar amino acid lengths. Lower structural similarity and variation of length of the proteins may lead to expected differences in MD simulations, which can significantly impact precision of predictions.

Dith-CE has a much better catalytic activity compared to the reported result of Ditu-CE from Kim et al. [20]. However, they both have better thermostabilities. And Dith-CE has the highest similarity with Ditu-CE. We presume that Ditu-CE exhibited a low activity may because Kim et al. used a low substrate concentration of 10 mM. Dith-CE with a concentration of 0.5 mg mL⁻¹ can produce 64.14 g mL⁻¹ lactulose from 200 g mL⁻¹ of lactose at 80 °C after 1 h, while 1.02 mg mL⁻¹ Caob-CE produced a similar level of lactulose (61.4 g mL⁻¹) at 70 °C with the same substrate concentration and reaction time [21]. This indicates that Dith-CE can produce lactulose more efficiently than Caob-CE because of the higher catalytic activity and higher thermostability of Dith-CE. However, the final yield of lactulose produced by Dith-CE after 4 h was approximately 51% of the initial substrate, which was slightly lower than that of Caob-CE. The ratio of epimerization rate to isomerization rate of CE correlates to the final content of products. The lower the ratio is, the more lactulose will be obtained after the reaction reaches equilibrium. The ratio of Caob-CE and Casa-CE is 42 and 38, respectively. Thus, the final yield of lactulose catalyzed by Caob-CE and Casa-CE [9] are higher than that by Dith-CE.

Abbreviations

CE

cellobiose 2-epimerase

HPLC

high-performance liquid chromatography

IPTG

isopropyl-(β-D-l -thiogalactopyranoside

SDS-PAGE

sodium dodecyl sulfate polyacrylamide gel electrophoresis

LB medium

Luria-Bertani medium

DSC

differential scanning calorimetry

MD

molecular dynamics