A novel strategy for e � cient disaccharides synthesis from glucose by β-glucosidase

Kangle Niu State Key Laboratory of Microbial Technology, Shandong University, Qingdao 266237, China Zhengyao Liu State Key Laboratory of Microbial Technology, Shandong University, Qingdao 266237, China Yuhui Feng State Key Laboratory of Microbial Technology, Shandong University, Qingdao 266237, China Tianlong Gao State Key Laboratory of Microbial Technology, Shandong University, Qingdao 266237, China Zhenzhen Wang State Key Laboratory of Microbial Technology, Shandong University, Qingdao 266237, China Piaopiao Zhang Yantai Huakangrongzan Biotechnology Co., Ltd, Yantai 264006, China Zhiqiang Du State Key Laboratory of Microbial Technology, Shandong University, Qingdao 266237, China Daming Gao Department of Rehabilitation Science, Graduate School of Health Science, Kobe University, Kobe 6540142, Japan Xu Fang (  fangxu@sdu.edu.cn ) Shandong University https://orcid.org/0000-0002-9196-5697

Extraction from plants is limited by the source plant and its terrestrial distribution. The biosynthesis of oligosaccharides via enzymatic synthesis technique in vitro has recently received increasing attention due to attributes including mild reaction temperature and excellent regio-and stereo-selectivity without the need for masking of functional groups (Moracci et al. 2001). Enzymatic synthesis of oligosaccharides is mainly catalyzed by glycosidases or glycosyltransferases (Perugino et al. 2004).
Synthesis of oligosaccharides by glycosidases has many advantages that include of simplicity, reliability, glycosidases, such as β-glucosidase, α-glucosidase, α-L -rhamnosidase, β-mannosidase and βgalactosidase, have been used to synthesize glycosides via the reverse hydrolysis reaction (Lu et (Lundemo et al. 2013). However, the effect of hydrophobicity of amino acid residues in the catalytic site on the catalytic activity needs to be further elucidated.
In this study, we report the ability of β-glucosidase TrCel1b from Trichoderma reesei to simultaneously catalyze the synthesis of three disaccharides (laminaribiose, sophorose, and cellobiose) from glucose.
The three-dimensional structure of TrCel1b was obtained and docked with cellobiose as the model of disaccharides. Based on the analysis of the docking results, we hypothesized that the hydropathy index of key amino acid residues in the catalytic site is closely related with disaccharide synthesis and more hydrophilic residues located in the catalytic site would enhance reverse hydrolysis activity.
To verify our deduction, the Hydropathy Index For Enzyme Activity (HIFEA) strategy was devised. Three hydrophobic amino acid residues in the catalytic site were mutated into hydrophilic residues, which generated the maximal change in the hydropathy index. Thus, three variants were obtained: TrCel1b I177S , TrCel1b I177S/I174S , and TrCel1b I177S/I174S/W173H . Additionally, the variant TrCel1b N240I was obtained by improving the hydrophobicity in the catalytic site. The production of synthesized disaccharides by the three variants were investigated. Total production (195.8 mg/mL/mg enzyme) of the synthesized disaccharides was increased by 3.5 times, compared to that of the wild type. Especially, the production of laminaribiose and sophorose reached 92.3 and 71.1 mg/mL/mg enzyme. The ndings indicate the value of the HIFEA strategy in providing a new perspective for the rational design of β-glucosidases used for the synthesis of oligosaccharides.

Expression and puri cation
The constructed vectors were introduced into E. coli BL21(DE3) for protein expression and transformants were selected on LB plates containing 10 μg/mL ampicillin as previously described (Xin et al. 2016).
These proteins were puri ed from the supernatant with His SpinTrap columns (GE Healthcare, Uppsala, Sweden) as previously described ). The puri ed protein of TrCel1b and its variants was analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis as previously described (Schagger 2006).
Measurement of β-glucosidase activity and enzymatic synthesis of disaccharides β-glucosidase activity was measured using 50 µL of 5 mM p-nitrophenyl-β-D -glucopyranoside (pNPG) as substrate at 30 o C for 30 min as previously described method ). The unit of β-glucosidase activity was de ned as the amount of enzyme required to release total reducing sugar equivalent to 1 µmol pNP per min. The reverse hydrolysis reaction was performed in 10 ml solution containing 3.5 mg TrCel1b or its variants, 1-8 g glucose, 500 μL glycerol, and 10 mg sodium azide in 50 mM phosphate buffer or 0.2 M sodium phosphate dibasic and 0.1 M citric acid buffer and the reaction solution was kept under 30 °C. The solution was sampled at speci c time and three independent replicates were carried out. The sample was diluted 1/4 -1/2 with puri ed water before the preparation sample was analyzed by HPLC that was equipped with a refractive index detector (Hitachi, Tokyo, Japan) using an Inertsil NH 2 column at 45 °C (250 mm × 7.8 mm; Shimadzu, Kyoto, Japan). The mobile phase was 80% acetonitrile and a ow rate was 1.0 mL/min. The fractions containing disaccharides were collected and the laminaribiose, sophorose, and cellobiose were further identi ed with Thin layer chromatography (TLC) using aluminum-backed sheets of silica gel 60F 254 (0.2 mm thick; E. Merck, Germany). Elution was carried out with n-butanol: ethanol: water (5:3:2). The plates were visualized by exposure to staining solution containing 3 g phenol, 5 mL concentrated sulfuric acid and 95 mL alcohol followed by charring.
Thermodynamic parameters for the reverse hydrolysis reaction were calculated from glucose to laminaribiose, sophorose, and cellobiose at pH 7.4 with an ionic strength of 0.05 M using the eQuilibrator (Noor et al. 2013).

Statistics
The Student's t-test was performed for signi cant differences between two groups of data. P<0.05 was considered statistically signi cant and standard deviations (SD) were calculated at least in triplicate.
The reverse hydrolysis reaction was performed with puri ed TrCel1b in 50 mM phosphate buffer at pH 7.4 under 30 o C. The results of TLC and HPLC indicate that the three disaccharides -laminaribiose, sophorose, and cellobiose were simultaneously produced through the reverse hydrolysis reaction with inexpensive glucose as the glucosyl donor ( Fig. 1 and Fig. S3). Moreover, the production of laminaribiose and sophorose was similar, with the production of cellobiose being markedly lower (Fig. S4 A-L, A-S, A-C).
To improve the production of disaccharides, interaction between TrCel1b and cellobiose as the model of disaccharide was predicted with Autodock 1.5.6 (Fig. 2B). The amino acid residues surrounded by the glucose moiety in aglycone subsite (+1 subsite) reportedly have a signi cant effect on synthesis capacity compared to the residues surrounded by the glucose moiety of cellobiose at the -1 subsite (Frutuoso et  in the aglycone subsite is surrounded by two hydrophobic residues (W173 and I174) and two hydrophilic residues (Y178 and N240), within a distance of 3.1 Å. The predicted distances between W173, I174, Y178, or N240 and cellobiose were 3.1, 2.8, 2.1, and 2.1 Å, respectively. The extremely hydrophobic residue I177 was also found at the +2 subsite at a distance of 6.4 Å. The hydropathy index of I, W, Y, and N was 4.5, −0.9, −1.3, and −3.5, respectively.
To verify our hypothesis, the HIFEA strategy was applied to improve the reverse hydrolysis activity by reducing the average hydropathy index of key amino acid residues in catalytic site (I ah ) of TrCel1b. I ah was de ned as the sum of the hydropathy index of amino acid residues 173, 174, 177, and 240 divided by their number, namely, I ah = (I h, 173 + I h, 174 + I h, 177 + I h, 240 ) / 4. Three hydrophobic amino acid residues in the catalytic site were mutated into hydrophilic residues and three variants TrCel1b I177S , TrCel1b I177S/I174S , and TrCel1b I177S/I174S/W173H were obtained. Additionally, the variant TrCel1b N240I was obtained by improving the hydrophobicity in the catalytic site. The I ah of TrCel1b and its variants was calculated according to the hydropathy index. The I ah of TrCel1b, TrCel1b I177S , TrCel1b I177S/I174S , TrCel1b I177S/I174S/W173H , and TrCel1b N240I was 1.15, −0.175, −1.5, −2.075, and 3.15, respectively, which revealed that I ah changed along with the mutation. The hydrophobic interaction between these key amino acid residues and disaccharides was weakened when I ah was reduced by the mutation of the hydrophobic residues located in the catalytic site of W173, I174, and/or I177 to the hydrophilic residues, which facilitated the release of disaccharide. Finally, production of disaccharides synthesized by reverse hydrolysis was improved. On the contrary, reverse hydrolysis was repressed when I ah was increased by improving the hydrophobicity in the catalytic site.

Identi cation of products synthesized by TrCel1b wildtype and its variants
Puri ed TrCel1b I177S , TrCel1b I177S/I174S , TrCel1b I177S/I174S/W173H , and TrCel1b N240I displayed a similar molecular weight of 74 kDa, compared to the size of TrCel1b (Fig. S5), consistent with the molecular weight predicted by the ExPASy website (https://web.expasy.org/compute_pi/). The products synthesized by these variants, except TrCel1b N240I , were laminaribiose, sophorose, and cellobiose, respectively (Fig. 1,  lane 7-9). The hydrolysis activities of the variants decreased as the I ah of variants decreased (Fig. 3A) and β-glucosidase activity of TrCel1b I177S/I174S/W173H was almost lost, compared to 0.35 U/mg soluble protein of TrCel1b. However, there was no signi cant change of β-glucosidase activity between TrCel1b and TrCel1b N240I . The production of disaccharides was enhanced as the I ah of the variants decreased (Fig. 3B). The disaccharide production of TrCel1b I177S/I174S/W173H increased by 3.5 times, reaching 68.5 mg/mL, compared with that of TrCel1b. On the contrary, the variant TrCel1b N240I with an I ah of 3.15 displayed no disaccharide synthetic activity. As shown in Fig. S4, the disaccharides laminaribiose, sophorose, and cellobiose were synthesized by TrCel1b I177S , TrCel1b I177S/I174S , and TrCel1b I177S/I174S/W173H the same as that of TrCel1b. Moreover, the production of laminaribiose, sophorose, and cellobiose synthesized by TrCel1b I177S , TrCel1b I177S/I174S , and TrCel1b I177S/I174S/W173H were greater than those of TrCel1b (Fig. S4). These results indicated that the decreased I ah of TrCel1b was favorite for the reverse hydrolysis reaction.
Glucose was used as the substrate (10, 20, 40, 60, and 80%) for reverse hydrolysis and the laminaribiose, sophorose and cellobiose production was measured by high-performance liquid chromatography (HPLC) (Fig. S4). The production of laminaribiose, sophorose, and cellobiose was greatly improved as glucose concentration increased from 10% to 80%. Moreover, we investigated the effects of pH value on synthesis activity in 0.2 M sodium phosphate dibasic and 0.1 M citric acid buffer (Fig. S6), and then it was proved that the optimal pH for TrCel1b I177S , TrCel1b I177S/I174S , and TrCel1b I177S/I174S/W173H were 7, 6, and 6, respectively. The activity of TrCel1b I177S/I174S and TrCel1b I177S/I174S/W173H increases by 55% and 46% in comparison with TrCel1b at pH 6 ( Fig. S6). When the hydrophobic residue isoleucine at the +1 subsite of TrCel1b is mutated to hydrophilic residue serine, the reverse hydrolysis activity of TrCel1b is improved; when W173 at +1 subsite of TrCel1b is mutated to histidine, protonation of histidine may occur under the weakly acidic environment, which is unfavorable to reverse hydrolysis activity (Lundemo et al. 2017;Seidle et al. 2005). Thus, the total disaccharides production synthesized by TrCel1b I177S/I174S/W173H is slightly lower than that of TrCel1b I177S/I174S at pH 6 (Fig. S6).
The production of laminaribiose and sophorose was increased when the I ah value of the variants was decreased. Laminaribiose production was the highest among the three disaccharides. The maximal laminaribiose production by TrCel1b I177S , TrCel1b I177S/I174S , and TrCel1b I177S/I174S/W173H reached 20.0, 27.6, 32.3 mg/mL, and increased 1.8, 2.8, and 3.5-fold, compared to that of TrCel1b, respectively (Fig. S4). Sophorose production was markedly higher than that of cellobiose, and its maximal production of TrCel1b I177S , TrCel1b I177S/I174S , and TrCel1b I177S/I174S/W173H reached 17.8, 19.9, and 24.9 mg/mL, respectively (Fig. S4). These results revealed a direct relationship between the reverse hydrolysis activity and the I ah value of TrCel1b.
As shown in Fig. 3B, the increased disaccharides production from TrCel1b I177S to TrCel1b I177S/I174 is more than two times higher than that from TrCel1b I177S/I174S to TrCel1b I177S/I174/W173H . As shown in Fig. 2B, I177 is located in +2 site and then W173 and I174 are located in +1 site. Our results were consistent with previous report that the position of amino acid residue as well as I ah value has an important in uence on the disaccharides synthesis (Rosengren et al., 2014). We speculated that I177 has a more signi cant in uence on the entrance of glucose and the release of disaccharides, compared to W173 and I174 owe to +2 site is closer to the entrance of enzyme, compared to +1 site.
To study the effect of mutation on the interaction between protein and disaccharide, cellobiose was docked with TrCel1b and its variants using Autodock software and the result of docking was analyzed by LigPlot. The hydrophobic interaction between cellobiose and the amino acid residues of TrCel1b I177S/I174S/W173H (Fig. 4B) became weak, compared to that of TrCel1b (Fig. 4A). The number of residues that hydrophobically interacted with cellobiose in TrCel1b I177S/I174S/W173H and TrCel1b was 16 and 20, respectively. The hydrophobic interactions between the W173 and I174 residues and cellobiose disappeared since the two residues were mutated into the hydrophilic residues ( Fig. 4A and 4B). These results were veri ed using Discovery studio (Fig. S7).
The relationship between the I ah value of TrCel1b and reverse hydrolysis activity Compared with TrCel1b, the I ah of TrCel1b I177S/I174S/W173H decreased, which was bene cial for the release of the cellobiose product. On the contrary, the I ah of TrCel1b N240I was enhanced and the reverse hydrolysis activity of TrCel1b N240I was completely lost (Fig. 3B) owing to the mutation of the hydrophilic residue N240 to the hydrophobic residue isoleucine (Fig. 4C). The ndings provided an obvious indication of a direct relationship between the I ah value of TrCel1b and reverse hydrolysis activity. When the I ah value of TrCel1b became negative, the reverse hydrolysis activity was enhanced. On the contrary, when the I ah value of TrCel1b increased, the reverse hydrolysis activity was abolished. The ndings are consistent with our hypothesis that the hydropathy index of key amino acid residues in the catalytic site is closely related with disaccharide synthesis.  Comparison of disaccharides production synthesized by β-glucosidases In this study, an β-glucosidase TrCel1b from T. ressei was shown to simultaneously synthesize laminaribiose, sophorose, and cellobiose using a high concentration glucose as substrate. As shown in Table 1, the ∆ r G' of laminaribiose, sophorose, or cellobiose synthesis is <0 indicating it is realizable that laminaribiose, sophorose, and cellobiose were produced from glucose by TrCel1b. Ravet et al. (1993) reported that the disaccharides were produced by β-glucosidase derived from almonds. However, most of these disaccharides were gentiobiose, rather than laminaribiose and sophorose. This was the reason why the equilibrium constant (K eq ) of the reaction to synthesize laminaribiose, sophorose, and cellobiose (Table 1) was markedly lower than that of gentiobiose (53.8×10 -3 ). There are few reports on laminaribiose and sophorose synthesis, re ecting their low production. To improve the production of disaccharides synthesized by TrCel1b, protein engineering was performed using the HIFEA strategy. The production of laminaribiose, sophorose, and cellobiose synthesized by TrCel1b I177S/I174S/W173H was increased 3.5, 2.6, and 3.9-fold, respectively, compared to that of TrCel1b (Fig. S4). Compared with reported β-glucosidases from different species (Table S1), the maximal production of laminaribiose and sophorose by TrCel1b I177S/I174S/W173H reached 92.3 and 71.1 mg/mL/mg enzyme, respectively, and were higher than the results produced by β-glucosidases from Aspergillus niger, Corynascus sp., Penicillium verruculosum, T. reesei (Semenova et al. 2015), and almond (Ravet et al. 1993). To our knowledge, this is the highest production of laminaribiose and sophorose simultaneously synthesized by β-glucosidase.

Conclusions
Our results reveal that β-glucosidase TrCel1b from T. reesei simultaneously synthesized laminaribiose, sophorose, and cellobiose. Three variants (TrCel1b I177S , TrCel1b I177S/I174S , and TrCel1b I177S/I174S/W173H ) with improved disaccharide production were obtained using the HIFEA strategy. The I ah of β-glucosidase TrCel1b is an important factor for the production of laminaribiose, sophorose, and cellobiose. The decreased I ah value of TrCel1b improved the synthetic activity and reduced the hydrolytic activity. The HIFEA strategy is implicated as a new avenue for the production of high value-added disaccharides.  cellobiose. Cellobiose is depicted in red, the catalytic amino acid residues -E171 and E383 are depicted in magenta, W173, I174, I177 and N240 are depicted in yellow, blue, green, and orange, the hydrophilic amino acid residues around cellobiose are depicted in cyan The speci c β-glucosidase activity (A) of TrCel1b and its variants and the total disaccharides production (B) synthesized by TrCel1b and its variants using 80% (w/v) glucose as the substrate for 72 hours at 30 oC. Total disaccharides concentration represents the sum of the production of laminaribiose, sophorose and cellobiose synthesized by TrCel1b and its variants. *p<0.05; **p<0.01; ***p<0.001. The statistically signi cant difference was performed between TrCel1b and its variants