Cloning, production, purification and biochemical characterization of BGL-1
T. amestolkiae has been recently postulated as a very interesting option for producing enzymatic cocktails rich in BGLs [13]. All the BGLs from this fungus characterized up to date belong to family GH3 [14, 15], which is usually considered the one encompassing the BGLs with better catalytic efficiency, although these enzymes have some limitations such as their low glucotolerance [16]. In this sense, the GH1 family contains most of the glucose-tolerant BGLs characterized so far [17]. For this reason, investigating the presence of potential glucotolerant GH1 BGLs in T. amestolkiae genome and proteome could increase the value of the cellulolytic system of this fungus.
In a previous work [13], the secretome released by T. amestolkiae growing in different carbon sources was analyzed and, in every condition tested, one potential BGL from GH1 family (protein g8384, renamed as BGL-1) was detected in very low amounts. Therefore, the bgl1 gene was cloned and expressed in P. pastoris with the goal of increasing BGL-1 production in order to analyze its glucose tolerance, kinetic constants, and physicochemical properties. After identifying the DNA sequence, RNA was extracted from 7-day old cultures of T. amestolkiae, obtaining total cDNA by retrotranscription. Amplification of the sequence of the mature bgl-1 gene concluded that the 1906 bp gene contains one intron, and encodes a 619 amino acids protein (Figure S1).
The methylotrophic yeast P. pastoris has been widely used as one of the most efficient expression systems for heterologous expression of BGLs. Some of its most notable advantages include its ability to produce correctly folded protein at high levels, or to perform complex post-translational modifications [18]. A recombinant plasmid with the sequence of bgl-1 without the intron was constructed using pPICzα as vector. Once transformed P. pastoris X-33, the transformants were screened to detect the best β-glucosidase producers. The maximal β-glucosidase activity was 75 U/mL, which is among the highest productions of BGLs reported in the literature. As can be seen in Table 1, this value is only surpassed by those found for the recombinant PtBglu3 from Paecilomyces thermophila, and bgl3A, from Talaromyces leycettanus. It is important to emphasize that the activity determined for this recombinant BGL-1 was 35-fold higher than the total β-glucosidase activity detected in cultures of T. amestolkiae [14], which contains a mixture of BGL-1, BGL-2 and BGL-3. This value confirms the very high overexpression of BGL-1 in this system.
Table 1
Comparison of the heterologous production in P. pastoris of BGLs from different microorganisms.
Enzyme Name | Microorganism | Production (U/mL) | References |
rBgl3 | Aspergillus fumigatus | 4.9 | [46] |
rBgl4 | Penicillium funiculosum | 52.8 | [5] |
bgl3A | Talaromyces leycettanus | 6,000.0 | [47] |
PtBglu3 | Paecilomyces thermophila | 274.4 | [48] |
Nfbgl1 | Neosartorya fischeri | 33.5 | [49] |
MtBgl3a | Myceliophtora termophila | 41.0 | [50] |
Bgl3B | T. leycettanus | 1.5 | [51] |
BGL-2 | T. amestolkiae | 6.0 | [14] |
BGL-3 | T. amestolkiae | 8.1 | [15] |
BGL-1 | T. amestolkiae | 75.0 | This work |
BGL-1 was purified in very high yield (around 80%) in just one step, by anion-exchange chromatography using a HiTrap QFF cartridge (Table 2), and 10 mM sodium phosphate buffer, pH 6.
Table 2
Purification of the β-glucosidase BGL-1 secreted from P. pastoris cultures.
BGL-1 Purification |
Step | Total Protein (mg) | Total Activity (U) | Specific Activity (U/mg) | Yield (%) |
Crude extracts | 39.66 | 4,176.97 | 105.48 | 100.0 |
HiTrap QFF Anion exchange | 11.82 | 3,354.15 | 282.83 | 80.1 |
Table 3
Kinetic constants of BGL-1 hydrolyzing different substrates.
Substrate | Km (mM) | kcat (s− 1) | kcat/Km (mM− 1.s− 1) |
pNPG | 3.36 ± 0.7 | 898.31 | 267.35 |
oNPG | 2.36 ± 0.6 | 135.72 | 57.50 |
Cellobiose | 20.36 ± 3.4 | 137.77 | 6.76 |
Cellotriose | 19.39 ± 5.4 | 196.24 | 10.12 |
Cellotetraose | 17.62 ± 0.6 | 276.62 | 15.69 |
Cellopentaose | 12.41 ± 0.2 | 260.42 | 20.98 |
Cellohexaose | 9.18 ± 0.6 | 217.86 | 23.73 |
The isoelectric point of the pure BGL-1 was determined to be 6.7 by isoelectrofocusing, and its molecular mass, measured by MALDI-TOF mass spectrometry, was 88.11 kDa. Considering that the theoretical mass was 23% lower (68.05 kDa), the differences found can be attributed to P. pastoris protein hyperglycosylation [18].
Optimal activity of BGL-1 was found at pH 4 and 60 °C (Fig. 1). Although these values are in the ranges reported for other native β-glucosidases [3, 17], BGL-1 has the peculiarity of working unusually well even at more acidic pHs. This behavior was also observed in the two other known BGLs of this fungus [14, 15], which may indicate that these enzymes from T. amestolkiae are more tolerant to acidic pHs than most BGLs characterized to date.
Glucose tolerance, kinetic study and substrate specificity of BGL-1
In general, most of the β-glucosidases used for cellulose degradation belong to the GH3 family, showing high catalytic efficiency values. However, they are often inhibited by glucose, with inhibition constant values lower than 0.1 M. In contrast, some GH1 β-glucosidases are much more glucotolerant than GH3 BGLs, but they usually have lower kcat/Km values over cellooligosaccharides, which are their natural substrates [3, 16]. The inhibition constants of BGL-1 towards glucose were calculated, displaying two main features that should be highlighted. First, its Ki value was very high (3.8 M), which to the best of our knowledge, corresponds to the second highest reported [19]. Second, the activity of BGL-1 was stimulated by low concentrations of glucose (Fig. 2). This finding has been observed in other β-glucosidases, mostly belonging to the GH1 family [16], but also in few BGLs from the GH3 family [5]. The reason for this glucose-induced stimulation remains unknown, although it could be related either to an allosteric effect triggered by the binding of glucose to some part of the protein, or to an increased hydrolysis rate upon transglycosylation [20]. BGL-1 activity was improved by 1.18-fold in the presence of 0.25 M of glucose. At this point, the activity begins to decrease, although BGL-1 still retained 40% of its initial activity at 3 M glucose (Fig. 2). Both characteristics postulates BGL-1 as a candidate for industrial processes performed at high glucose concentrations.
β-glucosidases can be classified into three groups, according to their substrate preferences: cellobiases, which have high substrate specificity towards cellooligosaccharides, aryl-β-glucosidases, with very high specificity towards synthetic substrates such as p-nitrophenyl-β-D-glucopyranoside (pNPG), and β-glucosidases with broad substrate specificity, that combine both activities [17]. In this sense, the kinetic constants showed that BGL-1 has very high efficiency and good affinity on pNPG and oNPG and, in addition, this enzyme displays a remarkable activity against p-nitrophenyl-β-D-xylopyranoside (specific activity 5.3 U/mg), thus demonstrating to have some versatility. However, its catalytic efficiency against cellooligosaccharides, from cellobiose to cellohexaose is low, as it occurs with other fungal BGLs from the GH1 family (Table 4). Thus, in contrast with the high activity on oligosaccharides shown by BGL-2 and BGL-3, which could be classified as cellobiases from the family GH3, BGL-1 seems to be a member of the group of aryl-β-glucosidases.
Table 4
Comparison of the catalytic efficiency on cellobiose and glucotolerance reported for BGLs from the GH3 and GH1 families.
Enzyme | Km (mM) | kcat (s− 1) | kcat/Km (s− 1/mM) | Ki (mM) or Glucose tolerance |
| | | | |
GH3 | | | | |
T. leycettanus Bgl3A [47] | 10.4 | 786 | 75.8 | 14.0 |
T. amestolkiae BGL-2 [14] | 1.1 | 630 | 567 | 1.0 |
T. amestolkiae BGL-3 [15] | 0.5 | 1,594 | 3,308 | 1.6 |
P. funiculosum rBgl4 [5] | 1.2 | 4,513 | 3,610 | 60% of residual activity at 500 mM |
| | | | |
GH1 | | | | |
Metagenomic library BGL mutant M3 [16] | 49.2 | 48.4 | 1.1 | 50% of residual activity at 250 mM |
Metagenomic library BGL mutant V174C [16] | 45.1 | 83.1 | 1.7 | 50% of residual activity at 3000 mM |
Soil metagenomic library BGL [19] | ND* | ND | ND | 4,280 |
Thermoanaerobacterium thermosaccharolyticum BGL [52] | 7.9 | 120 | 13.3 | 600 |
Thermoanaerobacterium aotearoense BGL [53] | 25.4 | 740.5 | 29.1 | 800 |
Actinomadura amylolytica AaBGL1 [54] | 95.3 | 10.7 | 0.1 | 40% of residual activity at 2000 mM |
Actinomadura amylolytica AaBGL2 [54] | 187.7 | 16.6 | 0.1 | 40% of residual activity at 500 mM |
Thermotoga thermarum BGL [55] | 35.5 | 19.0 | 0.5 | 1,500 |
BGL-1 (this work) | 20.4 | 137.8 | 6.8 | 3,780 |
*ND = no determined |
However, it is interesting to remark that, in spite of the relatively good kcat values of BGL-1 in the hydrolysis of cellooligosaccharides, the Km values are poor when compared with those observed for other BGLs from the GH3 family (Table 4). This result confirms that, like other BGLs from the GH1 family, BGL-1 has low affinity for these substrates, which could limit its applicability in hydrolytic processes.
Finally, an interesting discovery was made when examining the regioselectivity in hydrolysis reactions catalyzed by BGL-1. Enzymatic activity was tested against cellobiose, sophorose, laminaribiose and gentiobiose. While the activity over laminaribiose and gentiobiose was low, the activity over cellobiose and sophorose was considerably high, and BGL-1 was 5 fold more active on sophorose (535.82 U/mg) than on cellobiose (110.27 U/mg). These results indicated that BGL-1 could be considered as a versatile β-1,2 BGL, due to its preference for β-1,2 bonds but also being able to hydrolyze β-1,4 bonds. This behavior, although initially unexpected, seems to be more common as has recently been reported by Heins et al. [21], in a recent study in which more than 170 GH1 family enzymes were analyzed, using a high throughput screening approach. In said study, it can be seen how the activities obtained against sophorose were superior to those detected in the hydrolysis of cellobiose in a large number of the enzymes tested, which can establish a pattern within the proteins of this family. But, it is interesting to note that, so far, a very low number of β-1,2-glycosidases have been completely characterized. For example, some glucanases and glucosidases induced by β-1,2-glucan have been discovered in Acremonium sp., a filamentous anamorphic fungus [22]. However, their amino acid sequences have not been determined, which precludes their comparison with the BGL-1 from T. amestolkiae. Recently, a BGL with activity on β-1,2 bonds was reported in Listeria innocua, and related with the β-1,2-glucan metabolism in this bacteria [23], although the physiological role of these BGLs capable of synthesizing the β-1,2 disaccharide, remains poorly understood. However, it is well known that sophorose is the most powerful inducer of cellulases in T. reesei [24]. A recent report describes the production of this disaccharide by transglycosylation catalyzed by intracellular BGLs of this fungus, and the regulatory role of another BGL that hydrolyzes this compound, triggering the synthesis of cellulases [25]. Therefore, taking into account that BGL-1 is produced by T. amestolkiae in all conditions assayed, and that this enzyme can hydrolyze and synthesize β-1,2 bonds, its physiological role could be related to the regulation of the induction of the cellulolytic system in this organism.
Transglycosylation profile and regioselectivity of BGL-1
In order to test the transglycosylation capacities of BGL-1 a screening with a variety of potential acceptors, including sugars, sterols, phenolic compounds, or amino acids (Table S1) was performed according to a methodology previously developed [26] and also applied for studying the transglycosylation profile of other T. amestolkiae BGLs [27]. Unfortunately, in contrast with the good results obtained with the GH3 BGLs of this fungus against a large and diverse panel of glycosylation acceptors, BGL-1 only showed potential for transglycosylating p-nitrophenol sugar derivatives, like pNPG, pNPGal, or pNPX, which ruled out most of the potential acceptors for transglycosylation tested with the wild type enzyme. On the other hand, taking into account the saccharide nature of those acceptors, the regioselectivity of the transglycosylation was assessed analyzing by NMR the products of a model reaction set up with pNPG as donor and 13C-labelled glucose as acceptor. The NMR spectra of the compounds detected in the mentioned reaction were compared with those from sophorose, cellobiose and laminaribiose, confirming their coincidence with the pattern from sophorose (Fig. 3). This result showed that BGL-1 transglycosylated with high selectivity towards the β-1,2 manner.
Conversion of BGL-1 into glycosynthases by rational design
Historically, enzyme engineering has been successfully implemented to enhance the transglycosylation activity of glycosidases and, simultaneously, attenuate hydrolysis. With the aim of expanding the transglycosylation capacities of the recombinant BGL-1, it was converted into a glycosynthase. This kind of enzymes were first reported by Withers and coworkers [9], who noticed that a mutated glycosyl hydrolase lacking its catalytic nucleophile can use activated glycosyl fluoride donors with the opposite anomeric configuration for synthesizing glycosides, without hydrolyzing the products. This approach has been successfully applied for instance to convert glycosidases from GH1 family into glycosynthases, from GH36 in galactosynthases, or from GH29 in fucosynthases [10].
In this work, the replacement of the catalytic nucleophile of BGL-1, a glutamic acid at position 521, by a glycine (BGL-1-E521G) or a serine (BGL-1-E521S) produced two novel versions of BGL-1. These mutations have been shown to be much more efficient than the alanine replacement in the synthesis of oligosaccharides and p-nitrophenol derivatives [8]. Both versions of the protein were produced in P. pastoris and purified with the same strategy used for BGL-1. The purified glycosynthases showed their ability to use α-GlcF for synthesizing glycosides, and the most efficient mutant was selected from a comparative assay developed using 10 mM α-GlcF as donor and 10 mM pNPG as acceptor. Product formation was analyzed by HPLC, and the outcomes from this experiment showed the glycine mutant as the more efficient with a transglycosylation yield 2-fold higher that for the serine mutant. This is in concordance with data reported in the literature, which are explained considering that the rigid serine side-chain could hamper the departure of the fluoride, which is instead stimulated in the glycine mutants [28–30]. In a similar way, other authors justify this different behavior between the mutants considering the lack of a side-chain in glycine, and thus of reduced steric hindrance compared with the alanine or serine side-chains [31].
Once selected the glycine mutant, the kinetic parameters of this new glycosynthase employing α-GlcF and pNPG, for the formation of pNPG plus glucose, were determined. The results revealed that the affinity of BGL-1-E590G for pNPG (Km 90.14 mM) was higher than for α-GlcF (Km 260.86 mM), although the catalytic constants were similar for both substrates (kcat 0.11 s− 1 and 0.08 s− 1, respectively). The results obtained displayed slightly worst performance in Km than reported for a xylosynthase [32], but were similar in terms of kcat, confirming that the obtained glycosyntases could be a valuable starting point for optimizing biocatalytic transglycosylation reactions.
Transglycosylation of selected acceptors and analysis of the products
BGL-1-E521G was tested as the catalyst for transglycosylation towards aryl-glycoside acceptors, pNPG, pNPX, pNPGal, and some phenolic compounds, vanillin, hydroxytyrosol, gallic acid, and epigallocatechin gallate (EGCG), using α-GlcF as the donor (Fig. 4). With other glycosynthases pNP-sugars have been frequently used as preferential acceptors of transglycosylation, generating a variety of products, from the expected pNP-disaccharides, to pNP-oligosaccharides of different length and regioselectivity [8]. Besides these acceptors, phenolic compounds are very interesting targets for transglycosylation, because of the possibility of obtaining value-added glycosides from this type of compounds. These molecules have shown a variety of beneficial properties related to human health, and have been reported to confer cardiovascular protection, and to exert a positive effect in neurodegenerative diseases and cancer [33]. One of the main disadvantages of these substances when used in treatments is their low bioavailability, and their glycosylation, which can increase its solubility, has been proposed as a potential solution. In this context, various studies have demonstrated the interesting properties of hydroxytyrosol, vanillin and gallic acid [34–36], and EGCG has recently attracted attention as a potential therapeutic agent [37, 38], even in its glycosylated forms [39]. In a first approximation TLC analysis of the reaction mixtures was used to detect the synthesis of glycosides from the selected acceptors, identifying positive spots for each potential glycoside (figure S4). The presence of the expected compounds was confirmed by obtaining by mass spectrometry (MS) the molecular weight of the newly synthesized molecules. All the molecules were detected in its sodium adduct form in their mass spectra (Table 5). It is interesting to highlight that MS analysis revealed the presence of additional products with higher molecular weights in all reactions. The peaks corresponding to saccharides with two (G2), three (G3), four (G4) and five (G5) glucose units were thus detected, showing that this glycosynthase can also generate oligosaccharides. The presence of non-fluorinated derivatives of these molecules could be due to self-hydrolysis of α-GlcF or the fluorinated derivatives during overnight reactions. Besides, considering the ability of the native BGL-1 hydrolizing pNPX, and the capacity of BGL-1-E521G to interact with transglycosylation acceptors with xylose, it opens up the possibility of synthesizing oligosaccharides with xylose, or using D-xylosyl fluoride as potential donors of the reactions, which may expand the applications of the enzyme.
Table 5
ESI-MS data glycosides obtained by transglycosylation catalyzed by BGL-1-E521G. All the glycosides were detected as Na+ adducts.
Glycoside | | Intensity | m/z |
EGCG-glucose | | 533,662 | 643.2 |
Vanillin-glucose | | 43,977 | 337.1 |
Hydroxytyrosol- glucose | | 22,386 | 339.1 |
Gallic acid-glucose | | 48,294 | 355.1 |
pNPX-glucose | | 307,637 | 456.1 |
pNPGal-glucose | | 569,165 | 486.1 |
pNPG-glucose | | 430,001 | 486.2 |
G2 | | 449,035 | 365.1 |
G3 | | 428,105 | 527.2 |
G4 | | 287,540 | 689.3 |
G5 | | 59,761 | 851.3 |
The reactions rendering products with higher intensity in the TLC and MS analysis (those corresponding to pNPG, pNPX, pNPGal, and EGCG as acceptors) were submitted to HPLC analysis to determine the transglycosylation yield and to purify the main glycoside products for further NMR analysis. The conversions were 73.5% of initial pNPG, 89.8% for pNPX, and 36.6% for pNPGal, and, more interestingly, the glucoside of EGCG was obtained with a very significant yield of 48.8%. Besides, a second EGCG derived product was also detected and purifiedBesides, a second EGCG product was also detected and purified. The convsersion rate of 48.8% in transglycosylation of EGCG is among the best reported in the literature, although yields of 58% [40] and 91% [41], have also been reported for the enzymatic synthesis of the same compound using a cyclodextrin glucanotransferase and a dextransucrase, respectively. However, since the production of this glycoside has not been optimized, adjusting the reaction conditions for BGL-1-E521G transglycosylating EGCG could generate higher yields, and will be explored in future works.
Products characterization by NMR
The purified EGCG, pNPG, pNPX and pNPGal derived glucosides synthetized by transglycosylation catalyzed by BGL-1-E521G were analyzed by 1H and 13C-NMR in order to confirm their structure and assign their regiochemistry. The HMBC spectra of the two EGCG-glucose derivatives (figures S5 and 6) showed a correlation between the anomeric position of glucose and the meta-carbon (3’’/5’’) of the gallate aromatic ring, indicating the position of the linkage between the phenolic and the sugar moieties in these glycoconjugates (Fig. 5). In addition, the second sugar unit in the EGCG-disaccharide is attached to the O-2 of the first one through a β-linkage, as deduced from the value of the coupling constant of the anomeric proton (7 Hz) and in accordance with the spectral assignment (Table 6). The results from the NMR analysis of the three pNP derivatives are shown in Figs. 6, 7 and 8 and their NMR spectra in figures S7, S8 and S9. All of them indicate the regioselectivity of the glycosynthase that specifically forms pNP-disaccharides, incorporating the second sugar unit through a β-1,2 linkage. Finally, the different molecular species produced in a crude transglycosylation reaction mixture of α-GlcF (donor) and pNPG (acceptor) were also analysed by NMR. Interestingly, this sample showed significant heterogeneity as observed in the anomeric region of the 1H-13C HSQC spectrum of the mixture (Fig. 9). The presence of some unreacted acceptor pNPG-glucose but not starting donor α-GlcF, was confirmed. Unexpectedly signals tentatively assigned to α-F-sophorose were observed indicating that the α-GlcF itself with the α configuration could fit in the acceptor site. Besides, free glucose was also identified, confirming the auto-hydrolysis of the fluorinated substrate during the reaction. To note, this free glucose also worked as acceptor of α-GlcF, as deduced from the presence of sophorose among the reaction products. Again, the newly-synthetized disaccharides were linked by β-1,2 bonds, confirming the total regioselectivity of the synthase, which follows the same behavior than BGL-1, just being able to transglycosylate in this position.
Table 6
Chemical shifts for EGCG-glucose and EGCG-sophorose.
EGCG-Glucose | EGCG-Sophorose |
| 1H | 13C | | 1H | 13C |
2 | 5.08 | 77.06 | 2 | 5.03 | 77.05 |
3 | 5.60 | 68.89 | 3 | 5.57 | 68.71 |
4 | 2.86 | 24.91 | 4 | 2.84 | 24.86 |
3.01 | 2.97 |
4a | --- | 99.04 | 4a | --- | 99.00 |
5 | --- | 155.18 | 5 | --- | 155.34 |
6 | 6.08 | 95.72 | 6 | 6.06 | 96.01 |
7 | --- | 155.17 | 7 | --- | 155.34 |
8 | 6.08 | 95.72 | 8 | 6.06 | 96.01 |
8a | --- | 155.18 | 8a | --- | 155.21 |
1’ | --- | 106.33 | 1’ | --- | 106.39 |
2’ | 6.48 | 106.45 | 2’ | 6.47 | 106.40 |
3’ | --- | 145.42 | 3’ | --- | 145.51 |
4’ | --- | 132.10 | 4’ | --- | 132.62 |
5’ | --- | 145.44 | 5’ | --- | 145.65 |
6’ | 6.48 | 106.45 | 6’ | 6.47 | 106.40 |
1’’ | --- | 120.59 | 1’’ | --- | 120.58 |
2’’/6’’ | 7.09 | 108.61 | 2’’/6’’ | 7.06 | 108.61 |
3’’ | --- | 144.30 | 3’’ | --- | 144.59 |
4’’ | --- | 139.73 | 4’’ | --- | 139.90 |
5’’ | --- | 144.27 | 5’’ | --- | 144.59 |
6’’/2’’ | 7.06 | 112.27 | 6’’/2’’ | 7.05 | 112.18 |
7’’ | --- | 166.50 | 7’’ | --- | 166.43 |
1 Glc | 4.97 | 100.33 | 1 Glc | 5.09 | 99.22 |
2 Glc | 3.53 | 72.60 | 2 Glc | 3.78 | 80.98 |
3 Glc | 3.54 | 75.32 | 3 Glc | 3.69 | 75.39 |
4 Glc | 3.49 | 68.65 | 4 Glc | 3.53 | 68.29 |
5 Glc | 3.22 | 75.97 | 5 Glc | 3.21 | 75.63 |
6 Glc | 3.41 | 59.73 | 6 Glc | 3.41 | 59.83 |
3.66 | 3.66 |
| | | 1’ Glc | 4.77 | 102.77 |
| | | 2’ Glc | 3.25 | 73.85 |
| | | 3’ Glc | 3.43 | 75.70 |
| | | 4’ Glc | 3.36 | 69.36 |
| | | 5’ Glc | 3.29 | 76.09 |
| | | 6’ Glc | 3.25 | 60.33 |
| | | 3.53 |