Crystal structure of a novel homodimeric l‐ribulose 3‐epimerase from Methylomonus sp.

d‐Allulose has potential as a low‐calorie sweetener which can suppress fat accumulation. Several enzymes capable of d‐allulose production have been isolated, including d‐tagatose 3‐epimerases. Here, we report the isolation of a novel protein from Methylomonas sp. expected to be a putative enzyme based on sequence similarity to ketose 3‐epimerase. The synthesized gene encoding the deduced ketose 3‐epimerase was expressed as a recombinant enzyme in Escherichia coli, and it exhibited the highest enzymatic activity toward l‐ribulose, followed by d‐ribulose and d‐allulose. The X‐ray structure analysis of l‐ribulose 3‐epimerase from Methylomonas sp. (MetLRE) revealed a homodimeric enzyme, the first reported structure of dimeric l‐ribulose 3‐epimerase. The monomeric structure of MetLRE is similar to that of homotetrameric l‐ribulose 3‐epimerases, but the short C‐terminal α‐helix of MetLRE is unique and different from those of known l‐ribulose 3 epimerases. The length of the C‐terminal α‐helix was thought to be involved in tetramerization and increasing stability; however, the addition of residues to MetLRE at the C terminus did not lead to tetramer formation. MetLRE is the first dimeric l‐ribulose 3‐epimerase identified to exhibit high relative activity toward d‐allulose.

Edited by Alex Wlodawer D-Allulose has potential as a low-calorie sweetener which can suppress fat accumulation. Several enzymes capable of D-allulose production have been isolated, including D-tagatose 3-epimerases. Here, we report the isolation of a novel protein from Methylomonas sp. expected to be a putative enzyme based on sequence similarity to ketose 3-epimerase. The synthesized gene encoding the deduced ketose 3-epimerase was expressed as a recombinant enzyme in Escherichia coli, and it exhibited the highest enzymatic activity toward L-ribulose, followed by D-ribulose and D-allulose. The X-ray structure analysis of L-ribulose 3-epimerase from Methylomonas sp. (MetLRE) revealed a homodimeric enzyme, the first reported structure of dimeric Lribulose 3-epimerase. The monomeric structure of MetLRE is similar to that of homotetrameric L-ribulose 3-epimerases, but the short C-terminal a-helix of MetLRE is unique and different from those of known L-ribulose 3 epimerases. The length of the C-terminal a-helix was thought to be involved in tetramerization and increasing stability; however, the addition of residues to MetLRE at the C terminus did not lead to tetramer formation. MetLRE is the first dimeric L-ribulose 3-epimerase identified to exhibit high relative activity toward D-allulose. D-Allulose (alternative name D-psicose) is a rare sugar, and its physiological functions, such as altering the blood glucose level, suppressing fat accumulation, and use as a low-calorie sweetener [1][2][3][4][5][6][7][8][9], have been focused on as a promising functional food ingredient. In Japan, a low-calorie syrup containing D-allulose, Rare Sugar Sweet Ò (Matsutani Chemical Industry Co., Ltd., Hyogo, Japan) [10], is commercially available and its labeling as a functional food was recently approved. In addition, the U.S. Food and Drug Administration (FDA) stated that 'it allows the low-calorie sweetener allulose to be excluded from total and added sugar counts on Nutrition and Supplement Facts labels when used as an ingredient' (FDA In Brief, April, 2019).
Based on the sequence similarity with these potential enzymes by BLAST search, we selected a putative protein from Methylomonas sp. DH-1 (NCBI, WP_ 064020855.1, 286 a.a.) as a putative sugar isomerase/ epimerase.
In this study, using a synthesized gene encoding the putative sugar isomerase/epimerase from Methylomonas sp., the expression, characterization, crystallization, and structure determination of the putative protein were performed. In the first screening, enzymatic activity toward D-allulose was confirmed and the recombinant enzyme from Methylomonas sp. was suspected to be D-allulose 3-epimerase. In further investigation of its substrate specificity, the enzyme exhibited the highest enzymatic activity toward L-ribulose, followed by D-ribulose and D-allulose (Fig. 1). Therefore, it was considered to be L-ribulose 3-epimerase (LRE), but it also had relatively high enzymatic activity toward D-allulose. X-ray structure analysis of the LRE from Methylomonas sp. DH-1 (MetLRE) revealed a homodimer different from previously reported homotetrameric LREs, including AgDAE, and the most structurally similar protein to MetLRE is AgDAE, which has a long C-terminal a-helix. The crystal structure of MetLRE is the first known homodimeric L-ribulose 3epimerase with a short C-terminal a-helix. The effects of the length of the C-terminal a-helix on forming a homotetramer were discussed.

Results and Discussion
In this study, the crystal structures of C-terminal histagged MetLRE alone and in complexes with D-fructose, D-allulose, and L-tagatose were determined at resolutions of 2.08, 1.80, 1.80, and 1.70 A, respectively. In addition, the crystal structures of N-terminal histagged MetLRE in complex with D-allulose and modified MetLRE with additional C-terminal residues in complex with D-fructose were determined at resolutions of 2.05 and 1.58 A, respectively.
Overall structure of MetLRE_his in complex with

D-fructose
MetLRE_his in complex with D-fructose was crystallized in the orthorhombic space group P2 1 2 1 2 1 with two molecules in an asymmetric unit. MetLRE forms a homodimer (Mol-A and Mol-B) and the catalytic sites of the molecules are located on the same side ( Fig. 2A, top view). There are four bound metal ions in the structure, a metal ion at the catalytic site in each molecule, and the others on the surface of the protein molecules. The former metals were found in all X-ray structures and considered to be manganese that was added in the medium for protein expression in Escherichia coli cells (1 mM MnCl 2 ), whereas the bound metal ions on the surface of the protein molecules are not found in all Xray structures and were considered to originate from a reservoir solution containing magnesium (0.2 M MgCl 2 ) used for crystallization. The metal ions were checked by CheckMyMetal web server (https://cmm.minorlab. org/) [35]. The catalytic metals were acceptable as Mn. The bound metals on the surface of the protein molecules were suggested to be Ca as an alternative metal to Mg; however, the B-factor and geometry of Mg were acceptable or borderline, and the bound metals were refined as Mg considering the concentration of MgCl 2 in the reservoir solution.
In Mol-A, the C terminus was refined to His293 including a linker and part of the His-tag, and the C terminus of Mol-B was refined to Gly287, which is the beginning of a linker before the His-tag. As bound b-D-fructopyranose was found around the C-terminal region of Mol-A, the His-tag region was able to be refined in Mol-A. The carbonyl oxygen of Ser288 in the linker forms a hydrogen bond with O2 of b-D-  The monomeric structure has twelve a-helices and eight b-strands, forming a (b/a) 8 barrel fold with four additional short a-helices (a0, a3 0 , a6 0 , and a8 0 ) (Fig. 2B). There is a metal ion at the center of the barrel, which is coordinated by Glu152, Asp185, His211, and Glu246 to form the catalytic site and bind the substrate. The bound metal ion at the catalytic site was refined as Mn 2+ because MetLRE was expressed in cells in medium containing 1 mM MnCl 2 , and strong electron density maps for a metal ion at the catalytic site were also observed from the crystal grown in the metal-free reservoir solution.

Protein analysis and enzymatic activity
The molecular size of MetLRE was estimated to be 53 057 Da by size exclusion chromatography. As the molecular weight of the monomeric MetLRE was approximately 32 kDa by SDS/PAGE (calculated MW is 31.3 kDa based on the amino acid residues), MetLRE likely forms a rigid homodimer in solution considering its crystal structure. After we confirmed enzymatic activity of the recombinant enzyme toward D-allulose, the optimum temperature of the enzyme was found to be 70°C, as shown in Fig. 4A. Compared with the optimum temperatures of AgDAE (70°C, [29]), MlLRE (60°C, [34]), PcDTE (60°C, [12]), AtDAE (50°C, [17]), and CcDAE (60°C, [22]), this homodimeric enzyme belongs to the group having a higher optimum temperature.
In As additional thermal stability evaluation of MetLRE, the residual activity was measured in the presence and absence of MnCl 2 at 70°C (Fig. 4C). The residual activity of MetLRE with Mn was slightly higher than that of MetLRE alone for initial 15 min, but each residual activity was less than 40% after 10 min incubation. The half-life time of MetLRE in the absence of MnCl 2 was calculated as 208 AE 2.0 min at 60°C (Fig. 4C).

Ligand-free and D-fructose
The structure of the catalytic site in ligand-free MetLRE is shown in Fig. 5A. The r.m.s.d. for the Ca atoms between Mol-A and Mol-B is 0.20 A (the refined regions: Ala2-Gly287 in both molecules). A metal ion is coordinated by two catalytic glutamate residues (Glu152 and Glu246), Asp185, His211 and two water molecules (W1 and W2). In the structure of Mol-B with bound D-fructose of MetLRE/D-fructose ( Fig. 5B), O1 of D-fructose forms hydrogen bonds with Glu158, His188, and Arg211, and O2 and O3 coordinate with a metal ion instead of W1 and W2 in the ligand-free structure. O2 of D-fructose also forms a hydrogen bond with His188 and Arg211. O3 is recognized by catalytic Glu152. O5 and O6 of D-fructose are recognized by His12 and Ser69 through hydrogen bonds, with a distance of 2.8 and 2.7 A, respectively. In the structure of Mol-A (Fig. 5C), O4, O5, and O6 of D-fructose were slightly shifted rotating around the C3-C4 axis. O4 forms a hydrogen bond with Glu152, with a distance of 2. 6 A, and O6 interacts with His12 and Ser69, with a distance of 2.8 and 2.9 A, respectively. The two catalytic glutamates (Glu152 and Glu246) face the substrate in the structure of MetLRE/D-fructose, as observed in the complex structure of PcDTE  [30]. While one catalytic Glu152 forms a hydrogen bond with O3, with a distance of 2.6 A, the other catalytic Glu246 is directed to the hydrogen atom of C3 with a distance of 2. 8 A, in Mol-B. Glu246 can remove a proton at C3 of the substrate in the epimerization reaction of MetLRE, as suggested in PcDTE [30,38]. Thus, MetLRE can produce D-allulose from D-fructose through the C3-O3 proton exchange mechanism via a cis-enediolate intermediate. In Mol-A of MetLRE/D-fructose, Glu152 can interact with O3, with a distance of 3.1 A, but the distance between Glu246 and C3 of D-fructose is 3.3 A. As Glu152 prefers to form a hydrogen bond with O4 (2.6 A) in Mol-A, MetLRE was unable to catalyze the epimerization of the substrate in this situation. As the electron density maps of D-fructose in Mol-A were slightly weaker than those in Mol-B, a bound form of MetLRE may exist.  Fig. 5E. O2 and O3 of D-allulose coordinate with the metal ion, forming hydrogen bonds with His188 and Glu246, with a distance of 2.9 and 2.3 A, respectively. Glu152 is directed to the hydrogen atom attached to C3 with a distance of 2.8 A. These distances are suitable for the C3-O3 proton exchange mechanism, suggesting that D-allulose is also recognized as a substrate of MetLRE. Although O4 of D-allulose faces the hydrophobic pocket (formed by Leu45, Val111, Leu116, and Leu257) and cannot form a hydrogen bond with any residue, O5 forms a hydrogen bond with Ser69 (2.9 A), and O6 interacts with Ser69 with a distance of 2.7 A.

L-Tagatose
The structures of Mol-A and Mol-B in MetLRE with bound L-tagatose are also almost identical (0.26 A r.m.s.d., Ala2-Gly287 in both molecules). The bound L-tagatose was found in the catalytic site of Mol-B, and a moderately bound a-L-tagatopyranose was found on the surface of Mol-B (O1, O2, and O5 of a-L-tagatopyranose interact with the carbonyl oxygen of Ile161, Arg128, and Glu167, respectively).
The structure of the catalytic site with bound L-tagatose in Mol-B of MetLRE/L-tagatose is shown in Fig. 5F. O2 and O3 of L-tagatose coordinate with the metal ion, forming hydrogen bonds with His188 and Glu246, with a distance of 3.1 and 2.7 A, respectively.
Glu152 is directed to the hydrogen atom attached to C3 with a distance of 3.0 A. O4 and O5 are directed to the hydrophobic pocket and do not form a hydrogen bond with the enzyme. Instead, O6 is recognized by His12 and Ser69, with a distance of 2.9 and 2.5 A, respectively. In Mol-A of MetLRE/L-tagatose, L-sorbose (C3 epimer of L-tagatose) was bound in the catalytic site based on the electron density maps. Glu152 forms a hydrogen bond with O3 of L-sorbose (2.8 A), and the distance between Glu246 and C3 of L-sorbose is 2.9 A. The bound L-tagatose was epimerized at the position of C3 and converted to L-sorbose. O6 also forms hydrogen bonds with His12 and Ser69, with a distance of 2.8 and 2.6 A, respectively. In the previously reported structure of PcDTE/Dfructose (Fig. 6A), PcDTE formed strong interactions with the 1, 2, and 3 positions of the substrate with strong electron density, and a relatively weak interaction occurred at the 4, 5, and 6 positions of the substrate with poor electron density. The structures of MetLRE in the complexes with the above substrates with strong electron density maps suggested stronger interaction with the substrate at the 4, 5, and 6 positions. Recognition of the substrates in the catalytic structure of MetLRE is consistent with the enzymatic activities toward ketohexoses.
In this study, we were unable to determine the complex structure with L-ribulose and D-ribulose and cannot explain their mode of binding; however, MetLRE preferably binds ketopentoses with the same orientation of the hydroxyl groups at C3 and C4 positions in the Fischer projection (cis-configuration) (Fig. 1, Table 2). The orientation of the hydroxyl group at C5 of ketohexoses together with the hydroxyl group at C6 affects the enzymatic activity toward ketohexoses by forming hydrogen bonds with MetLRE. On comparison between D-allulose and L-tagatose (or D-tagatose and L-allulose), the orientation of the hydroxyl group at C5 influenced the enzymatic activity. Therefore, the hydroxyl group at C5 of ketopentoses may also affect the activity, and MetLRE prefers to bind the L-form of ketopentoses according to its enzymatic activity. As an example of the structure in complex with ketohexose harboring a different orientation of the hydroxyl groups at C3 and C4 positions (trans-configuration), two binding forms were observed in the complex structure of MetLRE/D-fructose (Fig. 5B,C). In one form (Fig. 5C), O4 of D-fructose can form a hydrogen bond with catalytic residue Glu152 and likely influences the catalytic activity. This may be a reason for why MetLRE prefers substrates with hydroxyl groups at C3 and C4 in a cis-configuration and exhibits relatively high activity toward D-allulose. However, as the recognition mechanism of substrates with a ring-form and the ring-opening mechanism of MetLRE are unknown, further studies are needed to elucidate the substrate specificity in detail.
Moreover, we found that some residues approach the substrate to stabilize the active site, as shown in Fig. 5D, and the Tm shifted in the presence of D-allulose (Fig. 4B). MetLRE changes to fit the substrate, which was not observed in our previous study on the homodimer enzyme PcDTE. Therefore, some residues, including His12, Leu45, and Phe248, may approach closer to ketopentoses such as L-ribulose and D-ribulose by forming hydrogen bonds with the substrate and increasing the hydrophobicity at the opposite face.
Of note, this enzyme exhibited the highest activity toward L-ribulose, followed by D-ribulose, at the optimum temperature (70°C). At 70°C, the residues recognizing substrates, such as His12, Ser69, and Phe248, may be more flexible in the active site and recognize Lribulose as a preferable substrate.

Effects of the length of the C-terminal a-helix of MetLRE
MetLRE was found from the genome database by amino acid similarity search with AgDAE and was thought to form a homotetramer such as AgDAE from a similar enzyme family. However, MetLRE formed a homodimer and has a shorter C-terminal helix (a8) than AgDAE, and its dimer form is similar to that of PcDTE with a relatively short a8. To eliminate the influence of the C-terminal His-tag, a new expression vector for MetLRE using the plasmid pET15b was constructed. A recombinant N-terminal His-tagged MetLRE (his_MetLRE) was prepared, and its crystal structure was determined. His_MetLRE in complex with D-allulose was obtained in the triclinic space group P1 (a = 51.02, b = 81.05, c = 106.13, a = 99.30, b = 101.50, c = 87.24) with six molecules in an asymmetric unit. His_MetLRE also forms a homodimer similar to C-terminal his-tagged MetLRE, but does not form a tetramer by symmetry operation (Fig. 7A,B). Although the electron density maps of the His-tagged N terminus were not fully visible, a part of the additional N-terminal region did not form a secondary structure and was not affected by oligomer formation. We next tried to make MetLRE form a homotetramer by elongating the C-terminal a8 with additional residues. Based on the tetramer form of AgDAE, N-terminal his_tagged Metlong2 (his_MetL-RElong2) was prepared with nine residues (TAIK-TIELH) added at the C terminus of MetLRE and three residues (Y9I, Y37F, and Y286L) were replaced to increase intramolecular interactions and eliminate the influence of intermolecular steric hindrance in the expected tetramer formation. Using the recombinant protein of his_MetLRElong2, the crystal structure of The visible additional C-terminal region did not elongate the a-helix, and his_MetLRElong2 did not form a tetramer by symmetry operation or PISA analysis [39]. However, the additional C-terminal tail affected the crystal packing because his_MetLRE and his_MetLRElong2 were crystallized under the same conditions using the same reservoir solution but their crystal packing differed. The dimer form of his_-MetLRElong2 aligns face to face with the side of the C-terminal region, although they did not contact each other to form a tetramer (Fig. 7C,D). In addition, we investigated the effects of the additional C-terminal residues on the thermal stability of his_MetLRElong2 and compared his_MetLRE and the newly constructed his_MetLRE3 that contains triple mutations (Y9I, Y37F, and Y286L) ( Table 3). The optimum temperature of all was 80°C in the presence of MnCl 2 (his_MetLRElong2: 68.1 AE 1.0 UÁmg À1 ; his_MetLRE3: 119.7 AE 2.3 UÁmg À1 ; his_MetLRE: 71.5 AE 1.6 UÁmg À1 ). The relative activity of his_MetL-RElong2 at 70°C (87.1%) was slightly higher than that of his_MetLRE (81.8%), and the relative activities of his_MetLRElong2 at 50°C (40.6%) and 60°C (55.0%) were almost similar to those of his_MetLRE. As the relative activities of his_MetLRE3 were slightly higher than those of his_MetLRE, the triple mutations (Y9I, Y37F, and Y286L) affected the thermal stability by increasing the intramolecular interactions, but the additional C-terminal residues in the loop region did not affect the thermal stability, because the region was unable to form a secondary structure and the length of the a-helix was not extended, precluding tetramer formation. In a homotetramer, a smaller angle is related to a shorter C-terminal a-helix, and a larger angle is related to a longer C-terminal a-helix (Table 1). They  (the angle of metal-a8N-a8c) = 1.8673 9 (the number of residues) + 3.6939 with R 2 = 0.9992. In this case, MetLRE requires 27 residues for the C-terminal a-helix to form a tetramer. Thus, nine additional residues were insufficient, and an entire a-helix consisting of 27 residues may be needed.

Conclusion
In this study, we expressed a novel homodimeric Lribulose 3-epimerase from Methylomonas sp. and determined its X-ray structure. As the structure contained a unique short C-terminal a-helix compared to the known structures of the structurally similar homotetrameric L-ribulose 3-epimerases and D-allulose 3-epimerases, we attempted to form tetrameric MetLRE by adding C-terminal residues, but we were unable to obtain the tetrameric form of MetLRE. As the dimeric form of MetLRE is relatively stable compared to the optimum temperature of homodimeric PcDTE (60°C, [12]), MetLRE may be an optimized dimeric enzyme and a new type of L-ribulose 3-epimerase with relatively high activity toward D-allulose.

Cloning and expression
An expression vector was constructed using the vector pQE60 (QIAGEN, Hilden, Germany) and a synthesized gene encoding the putative sugar isomerase/epimerase from Methylomonas sp. DH-1 (NCBI, WP_064020855.1). The synthesized gene was amplified using the following primers: 5 0 -GAGAAATTAACCATGGCCTTTCCGAAACAGCCTG-3 0 (forward primer) and 5 0 -GATGGTGATGAGATCCG-TACCGTTGTTTGATGAAT-3 0 (reverse primer) and mixed with the digested pQE60 by NocI and BglII for In-Fusion reaction using the In-Fusion HD Cloning Kit (TaKaRa Bio Inc., Shiga, Japan). Recombinant MetLRE was expressed in E. coli JM109 containing the resultant plasmid pQEMetLRE as a His-tagged protein at the C terminus (MetLRE_his).

Protein preparation
Escherichia coli cells expressing each recombinant MetLRE were grown at 37°C in 29YT medium containing 100 lgÁmL À1 of ampicillin until the culture reached an optical density of 0.4-0.5 at 600 nm and cultivated at 25°C overnight after the addition of 0.3-0.5 mM b-D-isopropylthiogalactopyranoside and 1 mM MnCl 2 . The harvested cells were sonicated in buffer solution (50 mM NaH 2 PO 4 , 300 mM NaCl, pH 8.0), and the produced Histagged MetLRE was purified by affinity chromatography using a HisTrap HP column (GE Healthcare Life Sciences, Uppsala, Sweden) and same sonication buffer including imidazole with a 0-500 mM imidazole gradient and dialyzed against 5 mM Tris/HCl, pH 8.0 overnight. The purified Cterminal His-tagged MetLRE_his protein solution was used for the enzyme assay and concentrated to 5.5 mgÁmL À1 using an Amicon Ultra-15 30 kDa Ultracel (Millipore, Billerica, MA, USA) for crystallization. The purified N-terminal His-tagged his_MetLRE and his_MetLRElong2 were also concentrated to 22.3 and 24.9 mgÁmL À1 , respectively, for crystallization, in the same manner as MetLRE_his.

Enzyme assay
The enzymatic activity of MetLRE_his toward D-allulose was confirmed as previously measured for an enzyme similar to AgDAE [29]. The reaction mixture consisted of the purified enzyme solution in buffer (50 mM Tris/HCl pH 7.5) containing 1 mM MnCl 2 and 25 mM D-allulose. To investigate the optimum temperature of MetLRE_his, the enzyme was incubated at each temperature (30,40,50, 60, 70, or 80°C) for 10 min and its enzymatic activity toward D-allulose was measured.  Allowed region (%)  To investigate the substrate specificity of MetLRE_his, a reaction mixture consisting of the purified enzyme solution in buffer (50 mM Tris/HCl pH 7.5) containing 1 mM MnCl 2 and 100 mM substrate (D-, L-allulose; D-, L-fructose; D-, Ltagatose; D-, L-sorbose; D-, L-xylulose; D-, L-ribulose) was incubated at 70°C for 10 min. After stopping the reaction by boiling for 2 min, the amount of product was measured by HPLC analysis. One unit of enzyme activity was defined as the amount of enzyme that epimerized 1 lmol of substrate per minute [12].

Molecular weight
The molecular weight of the monomer and molecular size of MetLRE were determined by SDS/PAGE analysis and size exclusion chromatography with a Superdex 200 Increase 10/300 GL column (GE Healthcare Life Sciences), respectively. The molecular size of the native protein was estimated using the Gel Filtration Standard (Bio-Rad, Hercules, CA, USA) as the standard protein markers with the range of 1.3-670 kDa.

Thermal stability
A thermal shift assay was carried out using a Viia7 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA) with SYPRO Orange as the fluorescence tag (Molecular Probes, Eugene, OR, USA) to determine the melting temperature (Tm) of the proteins by detecting the changes in the fluorescence spectra [42,43].
Briefly, 25 lL of mixture (0.5 mgÁmL À1 of purified enzyme solution and 59 SYPRO Orange in 40 mM Tris/ HCl, pH 7.6, in the presence or absence of 0.8 mM MnCl 2 and 4 mM D-allulose) was incubated in a 0.1 mL Micro-Amp TM Fast Optical 96-Well Reaction Plate (Applied Biosystems) at temperatures from 20°C to 95°C with a gradient of 0.015°C increment per 1 s. The assay was performed in duplicate.
The residual activity of MetLRE was measured in the buffer (50 mM Tris/HCl pH 7.5) incubated in the presence and absence of 2 mM MnCl 2 at 70°C for different time intervals. The half-life time (t 1/2 ) was calculated by measuring the residual activity of MetLRE in the absence of Mn at 60°C.

X-ray crystallography
The crystals of MetLRE_his were grown in a droplet containing 0.8 lL of protein solution (5.5 mgÁmL À1 in 5 mM Tris/HCl, pH 8.0) and 0.8 lL of reservoir solution [0.2 M MgCl 2 , 0.1 M Tris/HCl pH 8.5, 25% (w/v) PEG3350] against 50 lL of reservoir solution by the sitting-drop vapor-diffusion method at 293 K. Each crystal mounted in a cryoloop was soaked in solution containing 15% (w/v) D-fructose, D-allulose, or L-tagatose and flash-cooled in a stream of nitrogen gas at 100 K.
As for the crystals of his_MetLRE and his_MetLRE-long2, each protein solution (his_MetLRE: 22.3 mgÁmL À1 ; his_MetLRElong2: 24.9 mgÁmL À1 ) in 5 mM Tris/HCl, pH 8.0 and the reservoir solution [0.1 M HEPES pH 7.5, 10% (w/v) polyethylene glycol 8000 and 8% (v/v) ethylene glycol] were used with the above crystallization conditions. X-ray diffraction data were collected using the ADSC Quantum 315r CCD detector, PILATUS3 S6M PAD detector on the PF BL5A and the EIGER X 16M PAD detector on the PF BL17A in the KEK (Tsukuba, Japan). Diffraction data were processed using HKL2000 [44], XDS [45], and CCP4 program suite [46]. The initial phases of MetLRE_his were obtained by molecular replacement using the program MOLREP [47,48] with the structure of AgDAE (PDB code: 5ZFS) as a probe model. Further model building for protein and ligand molecules was performed in the program COOT [49], and the structure was refined using the program REFMAC5 [50,51]. Data collection and refinement statistics are listed in Table 4. Figures 2, 3, 5, 6, and 7 were drawn using the program PYMOL (Schr€ odinger, LLC., New York, NY, USA).