A highly efficient sorbitol dehydrogenase from Gluconobacter oxydans G624 and improvement of its stability through immobilization

A sorbitol dehydrogenase (GoSLDH) from Gluconobacter oxydans G624 (G. oxydans G624) was expressed in Escherichia coli BL21(DE3)-CodonPlus RIL. The complete 1455-bp codon-optimized gene was amplified, expressed, and thoroughly characterized for the first time. GoSLDH exhibited Km and kcat values of 38.9 mM and 3820 s−1 toward L-sorbitol, respectively. The enzyme exhibited high preference for NADP+ (vs. only 2.5% relative activity with NAD+). GoSLDH sequencing, structure analyses, and biochemical studies, suggested that it belongs to the NADP+-dependent polyol-specific long-chain sorbitol dehydrogenase family. GoSLDH is the first fully characterized SLDH to date, and it is distinguished from other L-sorbose-producing enzymes by its high activity and substrate specificity. Isothermal titration calorimetry showed that the protein binds more strongly to D-sorbitol than other L-sorbose-producing enzymes, and substrate docking analysis confirmed a higher turnover rate. The high oxidation potential of GoSLDH for D-sorbitol was confirmed by cyclovoltametric analysis. Further, stability of GoSLDH significantly improved (up to 13.6-fold) after cross-linking of immobilized enzyme on silica nanoparticles and retained 62.8% residual activity after 10 cycles of reuse. Therefore, immobilized GoSLDH may be useful for L-sorbose production from D-sorbitol.


Results and Discussion
L-Sorbose is recognized as a suitable intermediate in the industrial manufacturing of value-added chemicals such as vitamin C, 1-deoxygalactonojirimycin, and L-tagatose [1][2][3][4] . Bacterial fermentation has become the sole method for L-sorbose production 31 . Therefore, a catalytically efficient and stable SLDH is required to improve the bioconversion of D-sorbitol to L-sorbose [25][26][27][28][29][30] . Based on the conserved catalytic motif (KXXXXHXXH) in polyol-specific long-chain dehydrogenase, GoSLDH can be grouped into the subfamily of polyol-specific long-chain dehydrogenases 32 . A previous study reported the fermentative production of L-sorbose from D-sorbitol by G. oxydans G624 31 . Here, G. oxydans G624 showed SLDH activity (43.2 U/mL) and a 20% conversion yield from D-sorbitol to L-sorbose at 18 h of fermentation 31 . However, the conversion rate was low, and this process remained suboptimal for high yields. In addition, the concentration of D-sorbitol was inhibitory at > 10% (w/v). Thus, the results showed its limited potential for industrial application. To overcome these problems, we successfully cloned, expressed, and characterized the sldh gene from G. oxydans G624. Although the properties of L-sorbose-producing microbial enzymes have been determined, as listed in Table 1, GoSLDH, the first SLDH to be thoroughly characterized, exhibited higher activity than the previously characterized L-sorbose-producing enzymes, including SLDH and mannitol dehydrogenase (MDH) 25-30 . GoSLDH characterization. The cosmid sequence of G. oxydans G624 sldh was utilized to study the BAA99414.1 protein as described previously 8 . The identified GoSLDH had 32-82% sequence identity with polyol-specific long-chain dehydrogenase family enzymes. A comparative alignment of 17 sequences is presented in Fig. S1. Further, the catalytic region of Pseudomonas fluorescens MDH (PfMDH) was identified as the previously suggested conserved sequence (KXXXXNXXH) of polyol-specific long-chain dehydrogenase (Fig. S1). This GoSLDH alignment (Q9KWR5) revealed the presence of three highly conserved residues, Lys294, Asn299, and His302, that have been found in other known polyol-specific long-chain dehydrogenases 32 . In addition, GoSLDH possessed the active site residues of Asp190, Val228, Lys294, and Asp299. Based on these studies, we considered GoSLDH a polyol-specific long-chain dehydrogenase in G. oxydans G624. Based on the conserved catalytic motif (KXXXXHXXH) for polyol-specific long-chain dehydrogenase, GoSLDH can be grouped as a subfamily of polyol-specific long-chain dehydrogenases 30,32 . Expression and characterization of the GoSLDH enzyme. Codon optimization and expression of the GoSLDH gene. A codon-optimized Gosldh sequence of 1,455 bp containing 6× His tag was synthesized. The synthesized Gosldh gene was cloned into pET28a, and the recombinant plasmids containing the Gosldh gene were transformed into the E. coli BL21 (DE3)-CodonPlus RIL host strain. SDS-PAGE analysis was performed to evaluate recombinant protein expression as shown in Fig. 1A. The obtained GoSLDH had a molecular mass of 53.64 kDa. Further, an activity assay confirmed that GoSLDH was specifically produced as a target protein. The optimum pH and temperature for the purified GoSLDH were determined (Fig. 1B,C). Maximum activity was observed at pH 10 (100 mM glycine-NaOH buffer) and 70 °C. GoSLDH was subjected to dialysis with 10 mM EDTA after purification in order to analyze the effects of metal ions. This EDTA treatment did not affect the initial enzyme activity. K + , Ba 2+ , Ca 2+ , Cu 2+ , and Zn 2+ metal ions resulted in up to 30% inhibition of GoSLDH (Table S1). Mg 2+ , Co 2+ , and Mn 2+ ions showed no stimulatory influence on GoSLDH activity under similar conditions.
Substrate and coenzyme specificity. Substrate specificity was evaluated using D-sorbitol, mannitol, xylitol, ribitol (adonitol), myo-inositol, glycerol, and D/L-arabinitol (200 mM) as substrates. The observed results suggested that GoSLDH is highly specific towards D-sorbitol, mannitol, and D-arabinitol (Table S2). No activity was detected with L-arabinitol, xylitol, ribitol, myo-inositol, and glycerol. On the other hand, GoSLDH exhibited a preference for NADP + over NAD + as a coenzyme. Measurement of activity with D-sorbitol as a substrate suggested that GoSLDH is exclusively an NADP + -dependent enzyme, resulting in only 2.5% relative activity with NAD + . The recombinant GoSLDH exhibited high activity with D-sorbitol as a substrate, in addition to mannitol and arabinitol among various substrates. These results suggest that recombinant GoSLDH is highly specific and selective towards these substrates at an alkaline optimum pH.
Kinetic parameters. Purified GoSLDH showed the highest specific activity, 3570 U/mg, with D-sorbitol using NADP + as a coenzyme. Initial velocities were determined in glycine-NaOH buffer (pH 10) with D-sorbitol concentrations increasing from 5 to 400 mM under standard assay conditions (Fig. 1D). The K m values of 88.2 μ M and 38.9 mM for NADP + and D-sorbitol, respectively, and turnover rate (k cat ) of 3820 s −1 were obtained via a nonlinear regression analysis. The k cat /K m value for GoSLDH against D-sorbitol was 98.1 mM −1 s −1 . GoSLDH showed the highest catalytic efficiency with D-sorbitol among a number of L-sorbose-producing enzymes (Table 1).
Substrate docking. The reference PDB ID of the GoSLDH crystal structure is 5ITG. The structure of GoSLDH and PfMDH bound with D-sorbitol lacked sufficient detail to reveal precisely how it bound. To understand the binding conformation of D-sorbitol along with GoSLDH and PfMDH, molecular docking was performed. Molecular docking assesses the interactions of the protein with the ligand using scoring and energy parameters. The binding orientation of D-sorbitol with GoSLDH and PfMDH was similar, but the scoring and interactions showed a major difference. D-Sorbitol showed a − 8.68 kcal/mol docking score and − 56.30 kcal/mol docking energy with GoSLDH (Table S3; Fig. S2). In the case of PfMDH with D-sorbitol, the docking score was − 6.519 kcal/mol and the docking energy was − 41.17 Kcal/mol. A hydrogen-bonding network was formed between the hydrogen bond donors and acceptors in both GoSLDH and PfMDH. D-Sorbitol formed strong interactions of five hydrogen bonds with GoSLDH, with Asn190, Lys294, and Leu191 playing the roles of interacting amino acids (Fig. S3A). In comparison with GoSLDH, PfMDH shows weak bonding of only three hydrogen bonds with Asn191 and Lys381. In both, the Asn and Lys residues showed similar interactions, but Leu191 gave additional support to GoSLDH (Fig. S3A,B). In addition to the hydrogen bonding network, the amino acids Met227, Val287, Ile295, and Phe384 showed hydrophobic bonding with the D-sorbitol in GoSLDH. In PfMDH, Leu192, Phe385, and Tyr136 showed hydrophobic bonding between the protein and ligand (Fig. S2). The reaction mechanism for the conversion of mannitol to fructose by PfMDH was reported previously 33 . In this process, the hydroxyl group at C2 is oxidized via sequential abstractions of protons from the sugar O 2 and transfer of this proton to bulk solvent, followed by the hydride transfer from the sugar (C2) to nicotinamide (C4). Finally, the Lys residue at position 295 acts as a proton acceptor. Therefore, the distance between PfMDH Lys295 and O 2 hydrogen of the substrate (mannitol or D-sorbitol) is mechanistically important for the abstraction of protons for oxidation 33 . Our docking results were cross-checked with the reaction mechanism of GoSLDH and PfMDH. The distances between the protein and ligand atoms are tabulated in Table S3. The mean distance value of D-sorbitol with GoSLDH was 1.9 Å, whereas the average distance of D-sorbitol with PfMDH was 2.08 Å. In GoSLDH, the distance between the N atom (K294) and H atom was analyzed. The distance between Lys294 and the O 2 hydrogen atom of D-sorbitol was 2.35 Å (Fig. 2), which favors close contact. This was also seen in PfMDH, which showed a distance of 3.41 Å between the Lys295 N-atom and H-atoms (Fig. 2). This result suggests that the GoSLDH distance is shorter than that of PfMDH, which may allow easier abstraction of protons from D-sorbitol O 2 hydrogens. Overall, the interaction pattern, distance matrix, and scoring parameters suggest that GoSLDH shows stronger binding with D-sorbitol than does PfMDH. This may be the reason for the higher turnover rate of D-sorbitol by purified GoSLDH (3820 S −1 ) than by PfMDH (12.8 S −1 ).
Binding energy calculation. Molecular docking with a related post-scoring approach (MM-GBSA) was performed for D-sorbitol-bound GoSLDH and PfMDH. The MM-GBSA results for the free energy of binding prediction are presented in Table S3. The calculated free energy of D-sorbitol bound to the GoSLDH complex was − 37.46 kcal/mol, and the value with PfMDH was − 32.61 kcal/mol. The MM-GBSA scoring results generally significantly correlate with the activity determined experimentally, although the process is more computationally demanding. Based on the components of the binding-free energy, the major favorable contributors to ligand binding are nonpolar solvation (Δ Gsolv SA) and van der Waals terms. However, polar solvation (Δ Gsolv GB) opposes binding as suggested previously 34 . Thus, GoSLDH appears to have a strong binding affinity with D-sorbitol and can be used for L-sorbose production. In comparison, energy intake was more efficient for GoSLDH than for PfMDH, indicating that the docking interactions favor the distance between the C2-OH (D-sorbitol) and nitrogen (Lys294), which utilizes the high energy yield better than PfMDH. These results are consistent with the turnover rates of GoSLDH (k cat = 3820 S −1 ) and PfMDH (k cat = 12.8 S −1 ). Overall, the interaction pattern, distance matrix, and scoring parameters suggest that GoSLDH binds more strongly to D-sorbitol than does PfMDH. CV measurements. Given the shorter distance between Lys294 and the O 2 hydrogen of the substrate, it seems that GoSLDH could abstract a proton more efficiently than PfMDH (Fig. 2). After proton abstraction, the hydride is transferred from D-sorbitol to NADP + with the electron flowing to the positively charged nitrogen of NADP + to serve as an electron sink (Fig. 3A). As a result, D-sorbitol is oxidized. The D-sorbitol oxidation activities of GoSLDH and PfMDH were compared based on electron transfer capability using CV. The oxidation current peak using D-sorbitol was found to be 0.35 μ A at a potential of 0.76 V (Fig. 3B). Under the same conditions, PfMDH exhibited negligible oxidation of D-sorbitol with a peak current of 0.01 μ A. Based on these results, GoSLDH is adequate to efficiently electrocatalytically oxidize D-sorbitol, leading to a higher operating potential than PfMDH. This result suggests that GoSLDH can more efficiently oxidize D-sorbitol than PfMDH.

ITC analysis.
Binding affinities between D-sorbitol and GoSLDH or PfMDH were determined by ITC. The K d (dissociation constant) values were determined for inactive GoSLDH (Lys294Q) and PfMDH (Lys295Q), which were lacking the catalytic residues for substrate conversion during the titration experiments. The binding curves indicated that the heat emitted per mole of titrant as a function of the molar ratio of total ligand to total enzyme was similar to a previous analysis (Fig. 4) 35 . The GoSLDH affinity (Lys294Gln, K d = 96 μ M) towards D-sorbitol was 2.9-fold higher than that of PfMDH (Lys295Gln, K d = 278 μ M). These bindings were driven by the energy contribution Δ G (Gibbs free energy) values of − 22.88 and − 20.29 kJ mol for GoSLDH (Lys294Gln) and PfMDH (Lys295Gln), respectively. In addition, PfMDH (Lys295Gln) binding to D-sorbitol showed an increase in Δ G of 2.59 kJ mol. The lower Δ G value of GoSLDH (Lys294Gln) was likely due to a high binding affinity between GoSLDH and D-sorbitol. This clearly shows that GoSLDH exhibits better binding with D-sorbitol than does PfMDH. These results are consistent with the kinetic parameters (K m ) of GoSLDH and PfMDH.
Immobilization of GoSLDH. The purified recombinant GoSLDH was very unstable, with a t 1/2 of 70 min at 25 °C, although it was more stable than the GoSLDHs reported previously 8,27 . Thus, the covalent immobilization of GoSLDH onto SiO 2 nanoparticles was performed to improve its stability. Here, the amino groups of amino acids such as lysine present on the surface of GoSLDH react with the glutaraldehyde activated SiO 2 nanoparticles to form covalent bonds during immobilization at pH 7 21 . The IY and IE of GoSLDH immobilized on different silica nanoparticles were in the ranges of 40.4-71.2% and 53.5-76.7%, respectively (Table 2). These variations in  IY and IE were primarily associated with the properties of the silica particles 16,21 . Immobilization of GoSLDH onto 4830HT SiO 2 particles resulted in the highest IY and IE values of 71.2% and 76.7%, respectively. This efficient enzyme loading and IE on SiO 2 (80 nm) was mainly due to the large surface area of particles (440 m 2 /g). Further, modified SiO 2 particles with carbodiimide and cyanogen groups were evaluated for the immobilization of GoSLDH 20 . These modified particles resulted in lower IY and IE values, in the ranges of 53.2-57.7 and 54.7-67.4%, than the particles activated by glutaraldehyde (Fig. S4A). These results suggested that both IY and IE are significantly dependent on the methods of immobilization. The maximum loading of GoSLDH was 207 mg/g of support on glutaraldehyde functionalized SiO 2 particles (Fig. S4B). The presence of additional observed peaks at 1,600-1,800 cm −1 in the FTIR spectra of GoSLDH on particles, corresponding to the amide bonds and -N= C= O and C= C stretches as described previously 21 , confirmed that the enzyme was successfully immobilized (Fig. S5). Additional peaks at 3,444, 1,630, and 1,083 cm −1 were correlated with -OH stretching, C-O bending, and Si-O-Si stretching, respectively 21 . The residual activity of purified GoSLDH followed similar trends over a pH range of 7-10, and the stability slightly increased with an increase in enzyme concentration from 0.1 to 1.0 U/mL (Fig. S6). The thermal dissociation constants (k d ) for free and immobilized GoSLDH were 0.59 and 0.10 per h at 25 °C, respectively (Fig. 5A). These values indicate that both free and immobilized GoSLDH activities were decreased with an increase in incubation time. Here, the significantly lower k d value represents the high stability of the immobilized GoSLDH (Table S4). The t 1/2 value of 70 min for free GoSLDH was significantly increased to 420 min after immobilization at 25 °C, and a 6-fold increase in stability was observed after immobilization. To improve the stability of the immobilized enzyme, we then performed the cross-linking of immobilized GoSLDH by glutaraldehyde (0.5%) as described previously 21,23 . After cross-linking, the stability improved 13.6-fold compared to the free enzyme (Table S4). Here, the significant stabilization of the enzyme after immobilization might be primarily associated with favorable interactions between the enzyme and support. Reusability is a key parameter of an immobilized enzyme for its efficient application. After 10 cycles of reuse, the immobilized GoSLDH retained about 62.8% of residual activity (Fig. 5B). Here, the high reusability of immobilized GoSLDH might be primarily associated with its higher stability 21 . Similarly, after 10 days of storage at 4 °C, immobilized GoSLDH retained approximately 74.6% of residual activity, whereas the free enzyme retained less than 5% of activity under the same conditions (Fig. S7). In contrast, a previous report suggested that recombinant GoSLDH expressed in P. putida was very unstable, even at − 20 °C 8 .

L-Sorbose production by GoSLDH. D-Sorbitol, a naturally abundant and commercially available polyol
(1.8 €/kg; chemical market report 2009) was used as the starting material to produce L-sorbose. Purified GoSLDH was employed to oxidize D-sorbitol into L-sorbose. Further, the rate of either D-sorbitol utilization or L-sorbose formation was analyzed. The maximum bioconversion to L-sorbose reached 47% after 3 h of reaction (Fig. S8). The product was confirmed as L-sorbose by the HPLC retention time of pure L-sorbose of 9.2 min. Absence of additional peaks suggested that there was not formation of by-products during the course of the reaction (Fig. S8). The D-sorbitol oxidation product was recovered and polarimetrically analyzed. It was confirmed with an [α ] D of − 43.4°, matching that of a L-sorbose standard (Table S5).

Conclusions
This study reports the detailed characterization of recombinant GoSLDH, showing the highest activity towards D-sorbitol ever reported. Further, the high activity of GoSLDH was analyzed in a structural analysis together with CV and ITC studies. These observations suggested that recombinant GoSLDH is more catalytically efficient than other SLDH or MDH due to the close proximity of D-sorbitol to the catalytic residue, along with high binding affinity. The stability of GoSLDH was significantly improved by its immobilization on the SiO 2 nanoparticles. This recombinant GoSLDH with high catalytic activity can be used for efficient L-sorbose production.

Materials and Methods
Plasmids and reagents. The Ex-Taq DNA polymerase, a genomic DNA extraction kit, and PCR reagents were purchased from Promega (Madison, USA). Restriction enzymes were purchased from New England Biolabs (MA, USA) 36 . The expression vector, plasmid isolation kit, and NiNTA superflow column were obtained from Qiagen (Hilden, Germany) 36 . Oligonucleotide primers were obtained from Bioneer (Daejeon, South Korea). Chemicals used in assays were purchased from Sigma-Aldrich (St. Louis, MO, USA). Nanoparticles were obtained from Nanostructured and Amorphous Materials (Houston, TX, USA) as described previously 21 .
Bacterial strains and culture conditions. The genomic sequence and SLDH protein sequence of G. oxydans G624 were accessed in the NCBI database (www.ncbi.nlm.nih.gov). E. coli DH-5α and E. coli BL21 (DE3)-CodonPlus RIL strains were used as hosts for transformation of plasmids and expression, respectively, as described previously 36 . These strains were grown in LB medium supplemented with 50 μ g mL −1 of kanamycin at 37 °C and were maintained at 4 °C. , and incubation with shaking was continued for 20 h. Cells were centrifuged at 4,000 rpm for 12 min and 4 °C as described previously 21 . The resulting pellets were resuspended in pH 10.0 glycine-NaOH buffer (100 mM). Further, cells were sonicated for 10 min at 4 °C, and lysates were centrifuged at 14,000× g for 20 min at 4 °C to remove cell debris as described previously 21 . The cell-free extract of crude GoSLDH was purified on a Ni-NTA Superflow Column as described previously [36][37][38] . Enzyme samples were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and stained with SimplyBlue ™ SafeStain 39,40 . Protein concentrations were determined by the Bradford method as described previously [41][42][43] .
Determination of molecular mass. GoSLDH molecular mass was calculated using a protein molecular weight standard and a Bio-Sil SEC-250 column (300 × 7.8 mm, Bio-Rad) on an HPLC system (Shimadzu, Kyoto, Japan). Here, the mobile phase (pH 6.8) consisted of Na 2 HPO 4 (50 mM), NaH 2 PO 4 (50 mM), NaCl (50 mM), and NaN 3 (10 mM), and the flow rate was maintained at 1.0 mL/min. Further, the quaternary structures of the protein were analyzed through the comparison of molecular weights observed by HPLC and SDS-PAGE (Fig. 1A).
Enzyme assays. Activity was determined spectrophotometrically through the measurements of rate formation of NADH (ε = 5.62 mM −1 cm −1 ) or NADPH (ε = 5.12 mM −1 cm −1 ) at 340 nm at room temperature. A 1 mL standard reaction mixture was used in the assay containing cofactor NAD(P) + (0.5 mM) and substrate (20-200 mM) along with enzyme in pH 10 glycine-NaOH buffer (100 mM). One unit of GoSLDH activity was defined as the enzyme amount needed to catalyze the formation of 1 μ mol NAD(P)H per minute.

Determining the optimum temperature and pH values.
To determine the optimum pH, the enzyme activity was assessed in the range of 7.0-10.5 in the following buffers (100 mM): Tris-HCl (pH 7.0-9.0) and glycine-NaOH (pH 9.0-10.5). Similarly, the optimal temperature vale was determined by analyzing the enzyme activity at 20-85 °C.
Effect of metal ions. Initially, purified GoSLDH was dialyzed against glycine-NaOH buffer (100 mM, pH 10) with EDTA (10 mM) for 24 h 4 °C and thereafter with pure glycine-NaOH. GoSLDH was concentrated by ultrafiltration using an Amicon- described previously 20 . The kinetic data are presented as the averages of statistically relevant triplicate measurements with standard deviations 16 .

Grid generation and molecular docking. The crystal structures of GoSLDH (PDB ID: 5ITG) and
PfMDH (PDB ID: 1LJ8) were used for docking analysis. Both optimized structures were minimized until the average Root-Mean-Square-Deviation (RMSD) of the non-hydrogen atoms reached 0.3 Å as described previously 44 . D-sorbitol (3-D structure) was retrieved from the PubChem database, and the ligand structure was prepared using LigPrep v3.5. The bond orders, length, and angle were fixed and minimized with the Optimized Potentials for Liquid Simulations (OPLS-2005) force field as described previously 45 . After the protein and ligand structure was minimized and setup for the molecular modeling calculations through GLIDE grid generation, the grid was generated for both proteins 46 . Sitemap v3.6 was used to identify the five highest ranked potential binding sites for the prepared structures of GoSLDH and PfMDH using high accuracy and a restrictive hydrophobicity definition 47 . The XYZ axes of the grid box were set around the predicted sites, and to soften the receptor potential for non-polar interactions, we scaled the van der Waals radii of receptor atoms by 2.00 Å with a partial charge cut-off of 0.25. Successful grids were docked with D-sorbitol using the Extra Precision (XP) mode in Glide, and the results were analyzed as described previously 34,48 . Glide Emodel combines the Glidescore, coulombic, van der Waals, and strain energy of the ligand as described previously 49 .
Binding energy calculation. The free energy of binding was calculated for both docked complexes with D-sorbitol using the Prime/MM-GBSA method as described previously 34 . It was performed using the OPLS-2005 force field and GBSA continuum model in Prime version 3.0 (Schrodinger) as described previously 50,51 .
The binding free energy, Δ G bind, was obtained using the following equations as described previously 52 : where E ligand , E protein , and E complex are the minimized energies of the inhibitor, protein, and protein-inhibitor complex, respectively. where G solv (ligand), G solv ( protein), and G solv (complex) are the solvation free energies of the inhibitor, protein, and complex, respectively. Electrochemical measurements. CV analysis was performed to compare the oxidation potential of GoSLDH and PfMDH using an SP-150 potentiostat (BioLogic, USA). A three-electrode setup of Ag/AgCl, glassy carbon electrode (GCE), and platinum wire as a reference, counter, and working electrode were used for analysis, respectively, as described previously 23 . Enzyme samples (5 μ l) at a ratio of 9:1 with nafion (5%) were fixed on the GCE and dried at 4 °C for 12 h. The prepared electrodes were used for the CV determination of D-sorbitol oxidation in buffer (pH 10) containing NADP + and NAD + as cofactors (2 mM) for GoSLDH and PfMDH, respectively.
Isothermal titration calorimetry. ITC measurements were carried out using a Nano ITC (TA Instruments, DE, USA). In a typical ITC experiment, D-sorbitol (1 mM) was titrated with the GoSLDH Lys294Gln (3.7 mg/mL) and PfMDH Lys295Gln (3.7 mg/mL) mutants as described previously 35 . Titrations were carried out at 25 °C with 20 injections (2.4 μ L) of inactive mutant enzymes or their variant solutions. All solutions were filtered, degassed to avoid bubble formation, and equilibrated to the appropriate temperature before starting the experiment. The observed data were analyzed using Nano Analyze (TA 100 Instruments) and Origin software as described previously 35 . Enthalpy of binding was determined on the basis of three titrations each for GoSLDH (Lys294Gln) and PfMDH mutants (Lys295Gln). The intrinsic molar enthalpy change (Δ Ha), binding stoichiometry (n), binding constant (K a ), Gibbs free energy (Δ G b ), and Δ S were obtained as described previously 35 .
Immobilization of the purified GoSLDH. SiO 2 particles were functionalized by glutaraldehyde, 1-cyclohexyl-3-(2-morpholinoethyl) carbodiimide metho-p-toluenesulfonate, and cyanogen bromide to produce aldehyde, carbodiimide, and cyanogen groups on the particle surfaces for the covalent immobilization of GoSLDH as described previously 16,20 . The various modified SiO 2 particles (10 mg) were mixed with 1 mL of purified GoSLDH (1.0 mg protein), and incubation was performed for 24 h (4 °C) under shaking (150 rpm) conditions. Further, immobilized enzymes on the particles were recovered through centrifugation and subsequently washed three times with buffer (100 mM potassium phosphate, pH 7.0). The Bradford method was used for protein concentration measurements in the washed solution as described previously 16 . Immobilized enzyme residual activity was determined under standard assay conditions. The IE and IY were calculated as described previously 20 . The loading of enzyme on the particles was examined at different enzyme concentrations (50-800 mg/g of support). The stability at 25 °C and storage stability at 4 °C of immobilized GoSLDH were studied in glycine-NaOH buffer at the optimum pH. The reusability of GoSLDH immobilized on SiO 2 nanoparticles was examined at 25 °C over 10 cycles under assay conditions. For each cycle, the reusability was assessed by monitoring the rate of NADPH formation under standard assay conditions until stable absorbance at 340 nm was reached. The initial activity of the immobilized enzyme was considered as 100%. After each cycle, the immobilized enzyme was removed by centrifugation as described previously 21 and then washed with buffer and used for the second cycle. Each assay reaction was carried out three times.
Sorbose production and analysis. Reaction mixture (1 mL) containing sorbitol (50 mM) and NADP (50 mM) in pH 10 glycine-NaOH buffer (100 mM) was used for to obtain the product. The prepared reaction mixture was incubated for 2 h at 30 °C. Samples were withdrawn for the product analysis at various intervals using the cysteine carbazole sulfuric method at an absorbance of 560 nm 54 . Further, products were confirmed by HPLC (Ultimate 3000, USA) as described earlier 21 . Retention times of 11 min and 9.8 min were observed for D-sorbitol and L-sorbose, respectively. In addition, the D-or L-configuration of the obtained products (D-or L-sugars) was monitored by their rotation angles in a quartz cell at 25 °C with deuterium lamp emissions at 589 nm using pure standards (D-and L-sorbose) 55 .