A Highly Glucose Tolerant ß-Glucosidase from Malbranchea pulchella (MpBg3) Enables Cellulose Saccharification.

β-glucosidases catalyze the hydrolysis β-1,4, β-1,3 and β-1,6 glucosidic linkages from non-reducing end of short chain oligosaccharides, alkyl and aryl β-D-glucosides and disaccharides. They catalyze the rate-limiting reaction in the conversion of cellobiose to glucose in the saccharification of cellulose for second-generation ethanol production, and due to this important role the search for glucose tolerant enzymes is of biochemical and biotechnological importance. In this study we characterize a family 3 glycosyl hydrolase (GH3) β-glucosidase (Bgl) produced by Malbranchea pulchella (MpBgl3) grown on cellobiose as the sole carbon source. Kinetic characterization revealed that the MpBgl3 was highly tolerant to glucose, which is in contrast to many Bgls that are completely inhibited by glucose. A 3D model of MpBgl3 was generated by molecular modeling and used for the evaluation of structural differences with a Bgl3 that is inhibited by glucose. Taken together, our results provide new clues to understand the glucose tolerance in GH3 β-glucosidases.

Influence of metal ions and β-mercaptoethanol. The effects of metal ions and reducing agents on enzymatic activity of the MpBgl3 are presented in Fig. 4. Significant inactivation was observed only in presence of HgCl 2 , and no significant effects were observed for any of the other substances tested. The lack of effect with EDTA suggests that the MpBgl3 does not require a metal ion cofactor for activity. circular dichroism and molecular modelling of MpBgl3. The secondary structure of the purified MpBgl3 was analyzed by Far-UV CD spectrum (Fig. 5). The spectrum presents a positive peak at 194 nm and two negative peaks at 207 and 222 nm. These are characteristic of proteins that contain both α-helix and β-sheets, and deconvolution of the CD spectrum yields an estimate of 25% α-helix structure and 20% β-sheet structure.
The protein modelling C-score value of 0.81 for MpBgl3 indicates the reliability of the predicted tridimensional structure. With the aim of improving the quality of the model, energy minimization was performed using the Chiron server, which reduced the clash ratios in the modelled structure from 0.027 to 0.018 (Fig. S1).
The overall stereochemical quality of the predicted tridimensional structure of MpBgl3 was evaluated using PROCHECK and Verify3D programs available at SAVE (The Structure Analysis and Verification Server) platform are presented in Table S2. The SAVE analyses showed that the majority of residues are in the favored region The structural basis for glucose tolerance of the MpBgl3. To investigate the structural basis for glucose tolerance in Bgl3 enzymes, the amino acid sequence of the MpBgl3 was aligned with the glucose intolerant GH3 β-glucosidase from A. niger (AnBgl3; 46.1% sequence identity), A. oryzae (AoBgl3; 46.6% sequence identity), and A. aculeatus (AaBgl3; 47.2% sequence identity) (Fig. 6) 20,21 . For further analysis, the crystal structure of   the glucose intolerant AaBgl3 (PDB entry 4IIG, K I for glucose = 3.7 mM 20,22 was superimposed on the modelled structure of the glucose tolerant MpBgl3with a rmsd of 1.02 Å (over 751 aligned Ca atoms).
The modelled 3D-structure and sequence alignments suggests that the MpBgl3 conserves the catalytic retaining mechanism that is typical of GH3 enzymes, where the glycone-binding site is fully conserved, together with the Asp222 and Glu464 residues which act as the nucleophile and the acid/base, respectively (Figs. 6 and 7A). Although the glycone-binding site is completely conserved, comparisons of the aglycone-binding site identified significant differences, which may be responsible for the change in the topology and electrostatic properties of the entrance to the catalytic site (Fig. 7B,C). It has been previously suggested that changes in the shape and the electrostatic properties of the aglycone-binding site were responsible for modulating the glucose tolerance of the GH1 β-glucosidases 23 . It was further suggested that GH1 Bgls are more glucose tolerant than GH3 Bgl3 because of the deeper catalytic cavity, a less accessible catalytic site entrance, and a reduced negatively charged patch in the aglycone-binding site that decreases the access of glucose in the glycone-binding site, resulting in an enzyme that is more tolerant to glucose 23 .
Comparative analysis of the MpBgl3 and AaBgl3 catalytic sites indicated that the shape and electrostatic properties of the deep active-site entrance is associated with GH3 Bgl glucose tolerance similarly as observed for GH1 family enzymes. The depth of the catalytic cleft in the AaBgl3 is greater than for the MpBgl3 (Fig. 7B). Furthermore, the Arg247 residue in the MpBgl3 is replaced by Phe305 in the AaBgl3 ( Fig. 6 and 7B,C), and in the other GH3 Bgls that are inhibited by glucose this position is also occupied by aromatic residues (Phe in AoBgl3 and Tyr in AnBgl3), demonstrating that this difference may be important in modulating glucose tolerance (Fig. 6). In addition, the Trp224 residue in MpBgl3 is replaced by residues with less bulky side chains in the GH3 Bgl3 that are sensitive to glucose, such as Thr, Ala and Gly (Gly282 in AaBgl3) (Fig. 6). As shown in Fig. 7B,C, the presence of the Trp224 side chain in the MpBgl3 contributes not only to the restriction of the catalytic site entrance, but also pushes the Arg247 toward Trp15, and by changing the orientation of Trp15 further contributes to the narrowing of the entrance to the active site. Although the Trp15 residue in MpBgl3 is highly conserved in the different GH3 Bgls, the substitution of an aromatic residue at position 305 in glucose intolerant GH3 Bgl3 (for example Phe305 in AaBgl3) by Arg247 in MpBgl3 contributes to the loss of the aromatic stacking interaction with Trp15, perhaps facilitating the narrowing of the catalytic site entrance. Finally, the double substitution of residues Arg247 and Arg394 in the MpBgl3 by residues Phe305 and Ser436 in the AaBgl3 (Tyr and Ser in AnBgl3; Phe and Asp in AoBgl3), respectively, induce important changes on the active site entrance, introducing a more basic charge in the aglycone-binding site (Fig. 7D). Taken together, these results suggest that glucose tolerance by GH3 Bgl3 can be modulated by the depth, space and electrostatic characteristics of the catalytic site entrance in an analogous manner to the GH1 enzymes.

Discussion
β-glucosidases are important enzymes that catalyze the rate-limiting reaction in the conversion of cellobiose to glucose in the saccharification of cellulose for second-generation ethanol production. Due to this important role, many groups around the world are focusing on the field of Bgls for several biochemical and biotechnological applications, aiming the optimization in bioreactor production, the reuse through enzymatic immobilization, improvement in activity by site-direct mutagenesis and others. The search for glucose tolerant enzymes has a great importance for the sugar and alcohol industry. In this study we standardized the production of Bgls with a buffered medium (Lummy medium 19 ), that improved the Bgl3 physical-chemistry characteristics, a glucose hyper tolerant Bgl3 from a thermophilic fungus Malbranchea pulchella (MpBgl3) was purified and characterized.
The MpBgl3 was produced using 1% (w/v) cellobiose as carbon source in Lummy medium. This culture medium was chosen for being simple, buffered at pH 6.0 and to preferentially induce Bgl production. The purification performed by tangential ultrafiltration and ion exchange chromatography in DEAE-fractogel proved to be a fast, simple and an efficient protocol. Many Bgls have been purified using similar strategies [24][25][26] and the purity of MpBgl3 was confirmed by SDS-PAGE, zymogram and by mass spectrometry analyses.
It is interesting to note that when evaluated for metal ions influence MpBgl3 did not present any significant activation and for this reason, MpBgl3 is not a metalloprotein. In contrast, mercury chloride could reduce 90% of MpBgl3 activity, and the majority of Bgls show a reduced total or partial activity in the presence of Hg +2 . The relative effectiveness of the heavy metal ions as inhibitors has been reported to decrease the activity in the following order: +2 27 , and from all these heavy metals only the Hg +2 was able to inactivate the MpBgl3. It is know that the Hg 2+ can inhibit the enzymatic activity acting on thiol sites present in the enzyme active site 28 , or by acting on R groups at the enzyme surface by changing the 3D structure and consequently its activity 28 . In the case of MpBgl3 it can be explained by model structure that shows Cys amino acids forming a disulfide bridge (Fig. 6), and it could be contributing to this result. The effect of ions on other Bgls is quite varied. For example, Aureobasidium pullulans Bgl retained its activity in the presence of all ions tested 15 , on the other hand, Sporidiobolus pararoseus Bgl was inhibited only by Ag +2 and Hg +2 , and partially inhibited by Cu +2 and Zn +2 29 . Finally, Penicillium pinophilum Bgl was inhibited by Cu +2 and Pb +2 30 , and although the majority of metal ions do not inhibit Bgls activity, inhibition by Ag + , Hg +2 , Cu +2 and Fe +3 has been frequently reported 31,32 .
The kinetic parameters of the MpBgl3 were determined, with a K m of 0.33 mM, V max of 13.67 U/mg and K cat of 26.6 s −1 , and they were compared to those of other Bgls. These enzymes from different organisms present significant differences in size and kinetic parameters, i.e. two Bgls were reported from A. oryzae, one with molecular weight (MW) of 130 kDa, K m of 0.75 mM, V max of 456 U/mg and K cat of 651 s −1 . The other had a MW of 100 kDa, K m of 0.48 mM, V max of 264 U/mg and K cat correspondent to 373 s −1 , using pNPG as substrate 24 . The A. niger Bgl showed MW of 95 kDa, K m of 8 mM and V max of 166 U/mg, for the same substrate 33    www.nature.com/scientificreports www.nature.com/scientificreports/ The optimum MpBgl3 activity was estimated at 50 °C and pH 6.0, and it was similar to the Bgls from different organisms. Bgls usually exhibit optimal temperatures in the range of 40 °C to 60 °C and optimum pH in the range of 4.0 to 6.0 ( Table 2) 15,24,29,35 . The MpBgl3 is a versatile enzyme that can be used and applied for several proposes since it is thermostable at 40 °C, but it also retained considerable activity for 4 hours at 50 °C, and in a pH range from 5.0 to 8.0 for 24 hours. The pH and temperature stability of Bgls may vary from one organism to another, but several authors have reported that Bgls for thermophilic fungi are stable at pH values ranging from 4.0 to 6.0 and at temperatures from 40 to 60 °C 15,24,29,35 .
The MpBgl3 showed hypertolerance to glucose concentrations of up to 1 M, which is an impressive result comparing to others Bgl3s from other fungi. Decker et al. 20 21 showed that just 4 g/L of glucose was enough to strongly inhibit the Bgls activities from A. oryzae and A. niger. In the present work it was not possible to calculate MpBgl3 Ki because the glucose concentration values at which the enzyme was tested did not inhibit it.
Another objective of this work was to study the structural basis of this effect by modelling of the 3D structure. A 3D structural model for the MpBgl3 was calculated based on the amino acid sequence similarity with the glucose intolerant AnBgl3 from A. niger (PDB entry 4IIG). Modelling in the presence of glucose inferred that the active site region of the MpBgl3 as well as the amino acids are important in the interaction with glucose. Previous studies suggested that changes in the shape and the electrostatic properties of the aglycone-binding site were responsible for modulating the glucose tolerance for Bgl1 23 . It was already published that Bgl1 used to present greater glucose tolerance then Bgl3 due to the deeper catalytic cavity and less accessible catalytic site entrance, reducing the negatively charged patch in the aglycone-binding site that decreases the access of glucose 23 . In this work, comparative analysis showed that although MpBgl3 was a Bgl3 it presented the shape and electrostatic properties of the deep active-site entrance similar as observed for Bgl1 enzymes. In other words, the finds of this work suggested that glucose tolerance by MpBgl3 could be modulated by the depth, space and electrostatic characteristics of the catalytic site entrance in an analogous manner to the GH1 enzymes. These results represent a new perspective for those working on the improvement of enzyme cloning and expression, or those working with site-directed mutagenesis as a perspective to improve β-glucosidase performance.
In conclusion, the present study reports the purification, biochemical, kinetic characterization and 3D-modelling of a ß-glucosidase GH3 (MpBgl3) from the thermophilic fungus M. pulchella. The hyperglucose tolerance of the MpBgl3 is of interest in industrial applications since glucose tolerant Bgls are not inhibited by feedback. When included in an enzyme cocktail for biomass saccharification, these tolerant enzymes may improve the hydrolysis efficiency by shifting the equilibrium towards product formation. Further work is currently in progress in order to evaluate the role of glucose-tolerant Bgls on biomass hydrolysis.   www.nature.com/scientificreports www.nature.com/scientificreports/ (180 rpm) for 72 h at 40 °C. The mycelia were subsequently, separated from the liquid medium by vacuum filtration on Whatman filter paper number 1, and the crude filtrate was used as the source of extracellular MpBgl3.
Purification of MpBgl3 secreted by M. pulchella. The two-step purification of MpBgl3 was performed at 4 °C, in which 100 mL of the crude enzyme extract was concentrated and fractionated by tangential filtration using a Vivaspin ™ 20 membrane (50 and 100 kDa cutoff, GE Healthcare Life Sciences, Uppsala, Uppland, SE). In this step the proteins greater than 50 kDa and smaller than 100 kDa were recovered in a total volume of 10 mL. the pH of this recovered fraction was adjusted to 7.0 with 25 mM Tris-HCl buffer pH 7.0, and loaded onto a The thermal stability of the MpBgl3 was evaluated at temperatures of 40 °C, 50 °C, 55 °C and 60 °C. In these experiments, the enzyme was incubated without the substrate and aliquots were withdrawn at predetermined times for enzyme assay. The pH stability of the purified enzyme free of substrate was evaluated at 25°C, at pre-defined incubation times, in a pH range varying from 2 to 10, using different buffers. After the incubation period, the enzyme activity was assayed as described above. The buffers used were: McIlvaine (pH 2-8), 50 mM glycine (pH 9-10). The results were expressed in Residual Activity (%), where the 100% value was the enzymatic activity before incubation.
Kinetic characterization of MpBgl3. Determination of the kinetic parameters (V max and K m ) of pNPG hydrolysis by the purified MpBgl3 were determined in McIlvaine buffer pH 6.0 and 50 °C, and values of V max and K m were estimated using the SigrafW software 39 2 ), EDTA and β-mercaptoethanol was evaluated. The final concentration of each tested compound in the enzymatic reaction was 10 mM. Control sample was taken as the assay in the absence of any of the compounds tested. In these experiments the enzyme was previously dialyzed against distilled water.
Glucose effect. To evaluate the effect of glucose on the MpBgl3 activity, the assay described at Measurement of BGL activity was performed in the presence of different glucose concentrations. The final concentrations of glucose tested for the pure MpBgl3 were 0.05 M, 0.1 M, 0.25 M, 0.5 M and 1 M at optimum pH and temperature. All experimental activities were expressed relative to the 100% activity measured without the addition of glucose.
MpBgl3 modeling. The modelling of the M. pulchella GH3 three-dimensional structure was performed using the I-TASSER server [40][41][42] . The best model was selected based on I-TASSER C-score values. Energy minimization of the selected tridimensional model was performed using Chiron server 43 . The evaluation of the three-dimensional model was performed using the PROCHECK 44 and Verify3D 45,46 programs via the SAVES (The Structure Analysis and Verification Server) platform. The 2Struc (The Secondary Structure Server) platform was used to calculate the secondary structure composition of the M. pulchella GH3 model using the DSSP algorithm 47 . ethical approval. The authors declare that no experiments on humans or animals were performed for this article.