Characterization of the (Engineered) Branching Sucrase GtfZ-CD2 from Apilactobacillus kunkeei for Efficient Glucosylation of Benzenediol Compounds

In order to evaluate the glucosylation capacity of GtfZ-CD2, and possible effects of the relative position of hydroxyl group on the efficiency of glucosylation, we chose catechol, resorcinol and hydroquinone as acceptor substrates for glucosylation reactions. HPLC analysis showed that GtfZ-CD2 was capable of glucosylating catechol, resorcinol, and hydroquinone, generating two glucosylated products in each case (Fig. 1).

Glucosylated products were purified by fractionation on a Bio-gel P2 column for structure elucidation. Mass spectrometry analysis revealed that the two glucosylated products of catechol, resorcinol, and hydroquinone represent their monoglucoside ([M-H]^-, 271.08) and diglucoside products ([M-H]^-, 433.13), respectively (Fig. S1). ¹H NMR analysis showed that the monoglucoside product of catechol, resorcinol. and hydroquinone are α-D-Glcp-1,1-catechol (Cat-G1), α-D-Glcp-1,1-resorcinol (Res-G1), α-D-Glcp-1,1-hydroquinone (Hq-G1, i.e., α-arbutin). Their ¹H NMR spectra (Fig. 2) are identical to those previously reported by Devlamynck et al. (9). The catechol diglucoside product (Cat-G2) was identified as α-D-Glcp-α-(1→3)-Glcp-1,1-catechol, and its ¹H NMR spectrum (Fig. 2) matched that previously reported by Devlamynck et al. (9). The NMR spectra of Hq-G2 showed two α-anomeric signals at δ 5.538 (A1) and 5.431 (B1), fitting a di-glucosylated structure. From 2D ¹H-¹H TOCSY and ¹³C-¹H HSQC spectra all ¹H chemical shifts were assigned (Fig. S2, Table S1); residue B showed an H-4 signal at δ 3.49 ppm, which is a structural reporter signal for a terminal residue (33). All ¹³C chemical shifts for residue B, assigned from the 2D ¹³C-¹H HSCQ spectrum, fit a terminal glucose residue. Residue A showed an H-3:C-3 combination of δ 4.03/81.1 ppm, indicative of a 3-substituted residue (33). The 3-substitution of residue A is further reflected in the downfield shift of H-2. Residue B H-5 is also shifted downfield to δ 4.05, which was previously observed for α-D-Glcp-(1→3)- residues (33), fitting the structure α-D-Glcp-α-(1→3)-Glcp-1,1-Hydroquinone. Similar structural reporters were observed for Res-G2 with residue A H-3/C-3 at δ 4.04/81.5, fitting a 3-substituted residue (Fig. S3, Table S1). Residue B showed a terminal-residue pattern with a downfield shifted H-5 fitting a α-D-Glcp-(1→3)- residue. These data fit the structure of α-D-Glcp-α-(1→3)-Glcp-1,1-Resorcinol. Quantification of glucosylated products showed that the yields for catechol, resorcinol and hydroquinone were 59.3%, 19.7% and 5.8%, respectively, when 150 mM sucrose and 150 mM the corresponding acceptor substrates were used.

Compared to catechol, the yields of resorcinol and hydroquinone glucosylation by GtfZ-CD2 were relatively low. Increasing the concentration of sucrose was reported to be beneficial for the stability of glucansucrase and to promote the transglycosylation reaction (10), thereby increasing the yield of glucosylated products. Therefore, we chose to use 1 M sucrose as donor substrate for subsequent experiments. Also, the concentration of the acceptor substrate is a key factor that affects the production and yield of glucosylated products. When the concentration of catechol was increased from 100 to 200 mM, the yield of glucosylated products increased from 60% to 88% (Fig. 3). However, the yield of glucosylated product decreased when the catechol concentration exceeded 200 mM, and only 0.2% was obtained with 400 mM catechol, indicating an inhibitory effect on the activity of GtfZ-CD2 at a high concentration of catechol. The total production of glucosylated catechol products was highest at 200 mM catechol, reaching 175.8 mM. For resorcinol and hydroquinone, the yield of glucosylated products increased at lower acceptor substrate concentrations, probably reflecting the increased donor/acceptor substrate ratio. At 25 mM resorcinol and hydroquinone, the yield of corresponding glucosylated products reached 57.9% and 30.9%, respectively (Fig. 3). Although the yields were higher at low acceptor substrate concentrations, the total amounts of glucosylated products were relatively low. The total amounts of glucosylated products obtained for both resorcinol and hydroquinone were highest at 200 mM, reaching 72.7 mM and 30.9 mM, respectively, with yields of 36.4% and 15.5%, respectively.

As shown above, GtfZ-CD2 has considerable higher catalytic efficiency with catechol than with resorcinol and hydroquinone. This resulted in highest yields of catechol glucosylated products. The relative position of the two hydroxyl groups on the aromatic ring are thus believed to significantly affect GtfZ-CD2-mediated glucosylation of these benzenediol compounds. Aiming to improve the glucosylation of resorcinol and hydroquinone, catechol, resorcinol, and hydroquinone were docked in the GtfZ-CD2 modeled structure constructed previously (26). The modeled structure of GtfZ-CD2 showed that the polypeptide chain adopts a U-shaped path to form domains A, B, C, IV, and V, which is similar to reported GH70 glucansucrase crystal structures (6). Notably, GtfZ-CD2 has a wide acceptor substrate binding site, which is less favorable for the binding of the relatively small benzenediol compounds. Restricting the docking space to subsite +1 resulted in productive binding modes where an acceptor hydroxyl group is favorably positioned to attack the C1 atom of the covalent-glucosyl intermediate (Fig. 4). Notably, the spatial position of this hydroxyl group is shared between all three acceptor ligands (Fig. S4) and nearly coincides (within 1.4 Å) with that of the C6-hydroxyl of the superposed structure of the Gtf180-ΔN-maltose complex (PDB: 3KLL) (8). This attacking hydroxyl group is at hydrogen-bond distance with the acid/base residue E1672 of GtfZ-CD2, an interaction that is also observed in the Gtf180-ΔN-maltose complex between the C6-OH of the glucosyl moiety in subsite +1 and the corresponding acid/base residue E1063. Additionally, this hydroxyl group may engage in hydrogen bond interactions with the transition state stabilizing residue D1744, as well as with the 2-OH group of the covalently linked glucosyl moiety. The productive binding modes of all three benzenediol compounds thus, via transglycosylation, lead to monoglucoside synthesis. The docking results for catechol show that this compound may have two productive poses, where the second (non-attacking) hydroxyl can be stabilized through hydrogen-bonded interactions with either E1672 (left panel in Fig. 4A) or with the O3 of the covalently linked glucosyl moiety (right panel of Fig. 4A). For resorcinol and hydroquinone, it was noticed that in productive poses, the second (non-attacking) hydroxyl group is surrounded by several hydrophobic amino acid side chains (Fig. 4B and C). This is obviously not favorable for productive binding of these acceptor molecules. Therefore, based on the molecular docking analysis, 6 hydrophobic amino acid residues close to the second hydroxyl group at the acceptor substrate binding site, namely, L1559, L1560, L1590 (all in domain B) as well as A1635, F1638 and F2065 (all in domain A), were selected for semirational mutagenesis and the resulting GtfZ-CD2 mutants were screened for improved glucosylation efficiency.

For each of the 6 selected amino acid residues, 11 to 16 mutant enzymes were constructed. In total, 80 mutant enzymes were obtained from semirational mutagenesis (Table S2). Our initial screening based on TLC analysis revealed that although a few mutant enzymes (F1638Y, F2065W and F2065V) were still active and retained the ability to glucosylate benzenediol compounds, most enzymes with mutations targeted at residues L1590, A1635, F1638 and F2065 were inactivated. In contrast, L1559 and L1560 were found to be more tolerant to mutations and only L1559Q were completely inactive. Based on the initial screening, 11 mutant enzymes were selected for purification and their hydrolytic activities and glucosylation abilities were further quantitatively evaluated. All the selected mutant enzymes showed comparable or even higher enzyme activities with 200 mM sucrose as the substrates (Table S3). Regarding glucosylation, the yield of glucosylated catechol products synthesized by these mutant enzymes was not significantly affected (Fig. 5A). For resorcinol glucosylation, the yield of glucosylated products produced by L1559E and L1559D increased from 31% (wild type) to about 40% (Fig. 5B). For hydroquinone glucosylation, L1559K, L1560C and L1560Y mutations remarkably improved the yield of glucosylated products, with L1560Y giving the highest yield (25.0%) (Fig. 5C). Notably, the L1560Y mutant showed an increased monoglucoside Hq-G1 yield (21.5%) but a reduced synthesis of diglucoside Hq-G2 (3.5%), compared to 9.7% and 6.2% for wild-type GtfZ-CD2, respectively. In order to confirm the decreased ability of the L1560Y mutant to elongate the glucosyl unit of Hq-G1, we incubated GtfZ-CD2 and L1560Y with 200 mM Hq-G1 (acceptor substrate) and 1 M sucrose (donor substrate). Compared to wild-type GtfZ-CD2, the L1560Y mutant synthesized diglucoside products with a significant lower yield (15.4% versus 38.1%) when 200 mM Hq-G1 was used as acceptor substrate.

GtfZ-CD2 and L1560Y followed Michaelis-Menten kinetics in the transglucosylation reactions with catechol, resorcinol, and hydroquinone as acceptor substrates. The K_m values of GtfZ-CD2 for catechol, resorcinol, and hydroquinone were determined to be 37.76 ± 4.76 mM, 43.48 ± 4.56 mM and 61.44 ± 10.55 mM, respectively. Compared to GtfZ-CD2, L1560Y showed a comparable K_m value for resorcinol, a slightly lower K_m for hydroquinone but a significantly lower (70%) K_m for catechol (Table 1). The k_cat values of GtfZ-CD2 for catechol, resorcinol, and hydroquinone were determined to be 13. 40 ± 0.58 s⁻¹, 0.55 ± 0.02 s⁻¹, and 0.50 ± 0.03 s⁻¹, respectively. L1560Y displayed more than 2 times higher k_cat values for catechol and hydroquinone and a 65% higher k_cat value for resorcinol than GtfZ-CD2 (Table 1). Taken together, the catalytic efficiency (k_cat/K_m) of L1560Y for catechol, resorcinol, and hydroquinone was more than 10 times, 60% and 3 times higher than that of GtfZ-CD2, respectively (Table 1).

The monoglucoside product of hydroquinone, α-arbutin, is one of the most common whitening agents on the market, with important applications in the cosmetic industry. The GtfZ-CD2 L1560Y mutant enzyme was found to produce α-arbutin as its predominant product (Fig. 5). We further optimized the reaction conditions for α-arbutin synthesis by the L1560Y mutant in terms of enzyme concentration, hydroquinone concentration and the ratio of sucrose/hydroquinone (Fig. S5). The production of α-arbutin first increased with increasing enzyme loading, reaching a maximum at 5 U/mL and subsequently dropped at higher enzyme loadings. The optimum hydroquinone concentration was determined to be 200 mM. Production of α-arbutin also increased with the increase of the sucrose/hydroquinone ratio. The optimum condition for α-arbutin synthesis by L1560Y was thus determined to be 5 U/mL enzyme, 200 mM hydroquinone and 1.4 M sucrose. The production of α-arbutin reached 55.8 mM with a yield of 27.9% under these conditions.

Branching sucrase GtfZ-CD2 is capable of efficiently glucosylating dihydroxylaromatic compounds catechol, resorcinol, and hydroquinone to form monoglucoside and diglucoside products. The highest glucosylation efficiency was found for catechol, which has adjacent aromatic hydroxyl groups. Structural analysis of glucosylated products demonstrated that GtfZ-CD2 catalyzed the synthesis of (α1→3) linkages, the same specificity as for branching linkage synthesis onto dextran. The observed product profiles are consistent with molecular docking results, revealing favorable productive binding modes for catechol, but less so for resorcinol and hydroquinone. Semirational mutagenesis was further performed with selected residues at the acceptor substrate binding site, generating mutant enzymes with improved resorcinol and hydroquinone glucosylation. The optimized synthesis of α-arbutin by L1560Y mutant enzyme reached 55.8 mM with a yield of 27.9%. Using benzenediols as model compounds, our present study provides insight into their binding as acceptor substrates and demonstrates the capacity of the branching sucrase GtfZ-CD2 for the glucosylation of non-carbohydrate molecules.

GtfZ-CD2 was produced and purified as previously described (26). Briefly, the O/N preculture of E. coli BL21 star (DE3) harboring p15b-gtfZ-CD2 was inoculated into fresh LB media with 100 μg/mL ampicillin. The expression of gtfZ-CD2 was induced by adding 0.1 mM IPTG when the culture reached OD_600nm of 0.4~0.6. The culture was continuously incubated at 16°C with shaking at 160 rpm for 26 h. The cells were collected by centrifugation (8000 g, 10 min) and disrupted by B-Per (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer’s instructions. GtfZ-CD2 was purified by Ni-NTA affinity chromatography. The protein concentration was determined with a Nanodrop Q5000 spectrometer by measuring the absorbance at 280 nm. The molar extinction coefficient of GtfZ-CD2 was calculated by ProParam (41) as 159870 M⁻¹cm⁻¹ and used for protein concentration determination.

The hydrolytic activity of branching sucrase was determined by measuring the released reducing sugar with 200 mM sucrose as the substrate, which was quantified using the DNS method (42) and glucose as standard. Sucrose hydrolysis results in the release of glucose and fructose, therefore one unit of enzyme activity (U) was defined as the release of 2 μmol of reducing sugar per minute. The enzyme assays were performed in 50 mM sodium acetate buffer pH 5.5, 1 mM CaCl₂ at 30°C using 200 mM sucrose as the substrate. Samples of 120 μL incubation mixture at 10 min were inactivated by adding 120 μL DNS reagent and boiled for 10 min. The reducing sugar was quantified by measuring the absorbance at OD_550nm with a Tecan Infinite M200 Pro plate reader.

Glucosylation products were obtained by incubating GtfZ-CD2 with sucrose and benzenediol compounds (i.e., catechol, resorcinol, and hydroquinone) in 50 mM sodium acetate buffer pH 5.5, 1 mM CaCl₂. Reaction mixtures (10 μL) were injected into a HPLC system equipped with a refractive index detector and analyzed on Rezex ROA-Organic Acid H + (8%) column (Phenomenex, Torrance, CA, USA) using 0.005 N sulfuric acid as mobile phase with a flow rate of 0.6 mL/min. The temperature of the column was set to 50°C.

Enzymatic reactions (10 mL) were performed by incubating 1 M sucrose and 200 mM benzenediol compounds (i.e., catechol, resorcinol and hydroquinone) with 0.5 U/mL GtfZ-CD2 in 50 mM sodium acetate buffer 1 mM CaCl₂ for 120 h. The glucosylated products were partitioned 3 times with an equal volume of 2-butanol. The combined 2-butanol phase was concentrated to dryness by rotary evaporation at 45°C and the dry residues were dissolved in 2 mL deionized water. The extracted glucosylation products were fractionated on a Bio-Gel P-2 column (50 × 2.5 cm), eluted with 10 mM NH₄HCO₃ at a flow rate of 48 mL/h. Fractions containing pure glucosylated products were collected and lyophilized for further structural analysis.

The purified glucosylated products were injected into a LC-MS system (LTQ-Orbitrap velos pro ETD) for mass spectrometry analysis using negative mode. One-dimensional ¹H NMR, 2D ¹H-¹H and 2D ¹³C-¹H correlation spectra were recorded on a Bruker 600 MHz Spectrometer (Core Facilities For Life and Environmental Sciences, Shandong University) at a probe temperature of 300 K. Prior to NMR analysis, samples of purified glucosylated products were prepared by dissolving them in D₂O (Cambridge Isotope Laboratories, Inc., MA, USA) followed by freeze-drying. After repeating this process for two times, samples were dissolved in 0.6 mL D₂O. Chemical shifts (δ) were expressed in ppm by reference to internal acetone (δ ¹H 2.225; δ ¹³C 31.07). MestReNova (Mestrelabs Research SL, Santiago de Compostela, Spain) was used to process all the recorded spectra.

Different concentrations of benzenediol compounds (100 to 400 mM catechol, 25 to 400 mM resorcinol and 25 to 400 mM hydroquinone) were incubated with 1 M sucrose and 0.5 U/mL GtfZ-CD2 in 50 mM sodium acetate buffer pH 5.5, 1 mM CaCl₂ at 30°C for 120 h. The reactions were stopped by incubating at 95°C for 10 min. The glucosylated products were quantified by HPLC using respective purified standards.

A homology model of GtfZ-CD2, which was generated with the Phyre2 server (43) in intensive mode using residues 1364 to 2621 of the GtfZ sequence in the previous study (26), was used for molecular docking. The model covered 87% of the input sequence with >90% confidence. Construction of the model for the covalent glucosyl intermediate at the GtfZ-CD2 D1634 residue was guided by the model of the covalent glucosyl intermediate of glucansucrase Gtf180-ΔN (6). Molecular docking studies of GtfZ-CD2 were performed with catechol, resorcinol and hydroquinone using AutoDock Vina (44). The docking space was confined to the acceptor substrate binding subsite +1 as guided by the location of the covalent glucosyl intermediate and the size of the docked substrates. The homology model and docking results were visualized using PyMOL (The PyMOL Molecular Graphics System, Schrödinger, LLC).

Quikchange site-directed mutagenesis was performed as previously described (38), using primer pairs for each targeted residue (Table 2). The plasmid p15b-gtfZ-CD2, constructed in a previous study (26), was used as the template for site-directed mutagenesis. The PCR products were cleaned up with the PCR cleaning up kit (Omega Biotech, Doraville, USA) and then transformed into E. coli DH5α. For each mutation, 25 colonies were selected for plasmid isolation, and mutations were identified by nucleotide sequencing (Tsingke, Qingdao, China). All the mutants constructed are summarized in Table S2. The isolated mutant plasmids were transformed into E. coli BL21 star (DE3) and the mutant enzymes were produced as described for the wild-type enzyme but in a small scale (50 mL). The induced cells were collected by centrifugation (8000 g, 10 min) and the wet weights were determined. Cells were lysed with B-per reagent (4 mL/g wet weight). The supernatants obtained were used as crude enzymes, which were incubated with 150 mM catechol, resorcinol, or hydroquinone and 150 mM sucrose in 50 mM sodium acetate buffer, pH 5.5, 1 mM CaCl₂ at 30°C for 120 h. Synthesis of glucosylated products was evaluated by TLC analysis and compared with that of wild-type enzyme by evaluating the intensity of transglucosylation product spots. Reaction mixtures (1 μL) were spotted on TLC sheets (Merck Silica Gel 60 F254, 20 × 20 cm), which were developed with 2-butanol/acetic acid/water = 2:1:1. A mixture of glucose and malto-oligosaccharides (DP2-7) was used as a standard. The TLC plates were stained with 10% (vol/vol) sulfuric acid solution containing 2 g/L orcinol for carbohydrate detection.

Based on the initial screening results, 11 mutant enzymes were selected for further characterization. The enzymes were purified and were incubated (0.5 U/mL) with 1 M sucrose and 200 mM catechol, resorcinol or hydroquinone in 50 mM sodium acetate buffer pH 5.5, 1 mM CaCl₂, at 30°C for 120 h. The synthesis of glucosylated products was quantified by HPLC using respective standards.

Kinetic parameters of GtfZ-CD2 and L1560Y for the acceptor substrates catechol, resorcinol and hydroquinone were determined using different concentrations of catechol, resorcinol, and hydroquinone (ranging from 1 to 200 mM). The concentration of the donor substrate sucrose was kept constant at 1 M. All the reactions were performed in 50 mM sodium acetate buffer pH 5.5, 1 mM CaCl₂ at 30°C. With catechol as acceptor substrate, reactions were stopped after 20 min by incubating at 95°C for 10 min. In case of resorcinol and hydroquinone, reactions were run for 4 h before being terminated. Glucosylated products were quantified by HPLC using respective standards. One unit of benzenediol glucosylation activity was defined as the formation of 1 μmol of the corresponding glucosylated product per minute.

The L1560Y mutant enzyme was selected for α-arbutin synthesis based on its improved monoglucoside synthesis with hydroquinone as acceptor substrate. The production of α-arbutin was optimized by varying the enzyme concentration, hydroquinone concentration and the ratio of sucrose/hydroquinone. All reactions were performed in 50 mM sodium acetate buffer pH 5.5, 1 mM CaCl₂, at 30°C for 72 h. Different concentrations of mutant enzyme L1560Y (0.5 to 15 U/mL) were incubated with 1 M sucrose and 200 mM hydroquinone. For hydroquinone concentration optimization,10 to 300 mM hydroquinone was incubated with 5 U/mL GtfZ-CD2 and 1 M sucrose. Regarding the optimization of the sucrose/hydroquinone ratio, reactions were performed with 200 mM hydroquinone and various concentration of sucrose to achieve different ratios of sucrose/hydroquinone (from 1:1 to 7:1). The synthesis of α-arbutin was quantified by HPLC with the standard.

All relevant data are included in the main body or the Supplementary file of this article.