A Comparison of the Transglycosylation Capacity between the Guar GH27 Aga27A and Bacteroides GH36 BoGal36A α-Galactosidases

Abstract

The transglycosylation behavior and capacity of two clan GH-D α-galactosidases, BoGal36A from the gut bacterium Bacteroides ovatus and Aga27A from the guar plant, was investigated and compared. The enzymes were screened for the ability to use para-nitrophenyl-α-galactoside (pNP-Gal), raffinose and locust bean gum (LBG) galactomannan as glycosyl donors with the glycosyl acceptors methanol, propanol, allyl alcohol, propargyl alcohol and glycerol using mass spectrometry. Aga27A was, in general, more stable in the presence of the acceptors. HPLC analysis was developed and used as a second screening method for reactions using raffinose or LBG as a donor substrate with methanol, propanol and glycerol as acceptors. Time-resolved reactions were set up with raffinose and methanol as the donor and acceptor, respectively, in order to develop an insight into the basic transglycosylation properties, including the ratio between the rate of transglycosylation (methyl galactoside synthesis) and rate of hydrolysis. BoGal36A had a somewhat higher ratio (0.99 compared to 0.71 for Aga27A) at early time points but was indicated to be more prone to secondary (product) hydrolysis in prolonged incubations. The methyl galactoside yield was higher when using raffinose (48% for BoGal36A and 38% for Aga27A) compared to LBG (27% for BoGal36A and 30% for Aga27A).

1. Introduction

α-Galactosidases (EC 3.2.1.22) are glycoside hydrolases (GHs) which catalyze the hydrolysis of terminal non-reducing α-D-galactosyl units and are found across seven GH families with the majority of the characterized enzymes being part of families GH27 and GH36 [1]. α-Galactosidases are able to act on a range of different substrates such as raffinose, stachyose and melibiose and polymeric substrates such as galactomannans (locust bean gum, LBG and guar gum) and galactoglucomannan [2] but the substrate preference varies considerably. In several cases, it was shown that GH27 α-galactosidases display activity towards galactosyl units in galactomannan and other abundant β-mannans, but this is an unusual property among GH36 α-galactosidases which often prefer to act on terminal galactoses in oligosaccharides [3,4,5]. β-Mannans from renewable sources, e.g., softwood hemicellulose, are valuable second-generation feedstocks that gain increased interest as biobased raw materials [6,7].

GH27 and GH36 enzymes belong to clan GH-D and catalyze the hydrolysis of carbohydrates through a double-displacement mechanism [8]. In the reaction, the anomeric configuration (in this case α-configuration) of the substrate is retained in the product, and such enzymes are often referred to as “retaining”. Both GH27 and GH36 utilize two aspartate residues as the catalytic residues, acting as nucleophile and acid/base catalysts, respectively [8]. A nucleophilic attack by the enzyme generates a leaving group and a covalent glycosyl-enzyme intermediate with the remaining part of the substrate (donor). The intermediate is then disrupted by a nucleophilic attack by a water molecule [8] or another molecule with an accessible hydroxyl group (acceptor) [9]. For a given reaction, the propensity for the glycosyl-enzyme intermediate to be disrupted by the acceptor (transglycosylation) or water (hydrolysis) is reflected by the ratio between the rate of synthesis and rate of hydrolysis (r_S/r_H), and such values can be used to judge the transglycosylation capacity of enzymes [10].

The ability to use the retaining GHs’ transglycosylation capacity for the synthesis of defined glycosides and oligosaccharides has gained a lot of interest [9,11,12]. In contrast to chemical glycosynthesis, glycosylation with GHs can occur under mild, benign conditions, in a strereo- and regio-selective manner and in a single reaction without the use of protective groups [9,13]. Furthermore, GHs do not require activated substrates such as nucleotide sugars used by glycosyltransferases [9] and are therefore attractive for the utilization of abundant polysaccharides [14]. A number of studies have been carried out to investigate the transglycosylation capacity of α-galactosidases, exploring their ability to utilize different donor substrates, including 4-nitrophenyl-α-D-galactopyranoside (pNP-Gal) [15,16,17,18], di- and oligosaccharides [19,20] and polymeric substrates such as guar gum galactomannan [21]. The acceptors which were explored include saccharides [19,20], various alcohols [15,17] and glycerol [21]. Recent studies showed the prebiotic potential of α-galactosylglycerol [22]. Furthermore, there is increased interest in enzymatically generated [23] α-galactooligosaccharides for prebiotic applications [23,24]. One of the challenges when using the transglycosylation approach is the risk of secondary hydrolysis, i.e., when the synthesized product is hydrolyzed. The propensity for this is dependant on the used enzyme and an important factor to consider when looking for promising enzymes [10].

We recently showed that a synergy approach greatly benefits transglycosylation reactions with polymeric donor substrates. Substantially increased yields (4.4-fold higher) of transglycosylation products (allyl mannosides and allyl galactosides) were observed when the GH27 α-galactosidase from the guar plant (Aga27A) was added to incubations of the GH5 β-mannanase from Trichoderma reesei, with galactomannan and allyl alcohol [25].

In this study, we further examine the transglycosylation capacity of Aga27A and compare it to a GH36 α-galactosidase from Bacteroides ovatus (BoGal36A) [5]. The two enzymes were chosen because they belong to different families and were both shown to efficiently cleave galactosyl side groups from galactomannan [5,26], but their transglycosylation capacity has not previously been thoroughly investigated. Furthermore, the activity towards internally attached galactosyls in galactomannan observed for BoGal36A stands out as an unusual property for a GH36 enzyme [5]. In order to expand the repertoire of enzymes that may be used with complex natural glycan donors in transglycosylation, BoGal36A and Aga27A were screened for their ability to utilize three different glycosyl donors, pNP-Gal, raffinose and galactomannan (locust bean gum, LBG), combined with a range of acceptors. The first screening was carried out using mass-spectrometry which was followed by screening using HPLC. We evaluated the stability of the studied enzymes under the screening conditions and quantified their transglycosylation products with raffinose and LBG as donors to evaluate their transglycosylation capacity.

The presented screening procedure represents a quick and effective way to discover enzymes with the potential to utilize renewable polysaccharides for the production of well-defined glycosides. The study thus explores the utilization of polysaccharides as donor substrates in transglycosylation reactions for α-galactosidases, a subject which so far is relatively unexplored.

2. Materials and Methods

2.1. Chemicals

The α-galactosidase E-AGLGU from Cyamopsis tetragonoloba (Aga27A) and 4-Nitrophenyl-α-D-galactopyranoside (pNP-Gal) were purchased from Megazyme (Bray, Ireland). 4-Nitrophenol (pNP), raffinose, sucrose, galactose (Gal), methyl-α-galactopyranoside (Me-Gal), locust bean gum (LBG) (galactosyl to mannosyl ratio 1:4), methanol (MeOH), 1-propanol (PrOH), allyl alcohol (allyl-OH), propargyl alcohol (propargyl-OH), glycerol and kanamycin (Kan) were purchased from Sigma Aldrich (St Louis, MO, USA). HisPurTM Ni-NTA resin was purchased from Thermo Scientific (Waltham, MA, USA).

2.2. Expression and Purification of BoGal 36A and Confirmation of Identity for Aga27A

BoGal36A was expressed in E. coli and purified by HisTag chromatography as described previously [5], with some modifications. The cells were lysed with a French press instead of glass beads, and the buffer component in the binding, wash and elution buffer was 20 mM sodium phosphate, pH 7.4 and the imidazole concentration of the elution buffer was increased to 250 mM. Mini-PROTEAN^® TGX™ Precast Gels 12% (BioRad, Hercules, CA, USA) was used for sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE).

The expected identity of Aga27A was confirmed with trypsin cleavage using Trypsin Gold MS Grade (Promega, Madison, WI, USA) according to the manufacturer´s guidelines and analysis by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) collected using a 4700 Proteomics Analyzer (Applied Biosystems, Waltham, MA, USA) in positive reflector mode. In-solution digestion (20 µL) was performed with 10 µg Aga27A and 100 ng trypsin in 50 mM NH₄HCO₃ pH 7.8 at 37 °C for four hours, after which 100 ng trypsin were added, and the digestion continued overnight. To acidify the sample, trifluoroacetic acid was added to 0.5%, and 0.5 µL digested protein were spotted with 0.5 µL alpha-cyano-4-hydroxycinnamic acid (3 mg/mL). Bruker Protein Calibration Standard I (Bruker, Billerica, MA, USA) was used as a calibration standard. The peak list generated from the mass spectra was used in MASCOT to search an in-silica trypsin digested Swiss-Prot database for matching peptide fragments.

2.3. Enzyme Activity Determination

α-Galactosidase activity was determined by incubating the appropriately diluted enzyme with 1 mM pNP-Gal [5]. The reactions (150 µL) were carried out at 37 °C for 10 min in 50 mM sodium citrate (NaCi) buffer pH 6 for BoGal36A and 50 mM sodium acetate (NaAc) pH 4.5 for Aga27A. The reactions were stopped by the addition of 50 µL 1M Na₂CO₃. The released pNP was detected with a spectrophotometer at 405 nm and correlated to a standard curve of pNP (40 µM to 250 µM).

2.4. Screening for Transglycosylation with pNP-Gal and MALDI-TOF MS Analysis

The ability of the two enzymes to perform transglycosylation was evaluated by incubating triplicate reactions containing 8 nkat/mL BoGal36A or Aga27A with 40 mM pNP-Gal alone or with 10 v/v% methanol, 5 v/v% propanol, 10 v/v% glycerol, 5 v/v% allyl alcohol or 2.5 v/v% propargyl alcohol. The reactions were incubated in the assay buffers at 37 °C for one hour.

The reactions were diluted 10 times with Milli-Q water and thereafter analyzed with MALDI-ToF MS by adding 0.5 or 1 µL of the diluted sample onto 0.5 or 1 µL matrix (10 mg/mL 2,5-dihydroxybenzoic acid, DHB) on a stainless steel (Applied Biosystems, Waltham, MA, USA) or a 384 MTP BigAnchor (Bruker, Billerica, MA, USA) target plate, respectively, and then dried under warm air. The mass to charge ratio (m/z) of compounds present in the samples was then determined with either a 4700 Proteomics Analyzer (Applied Biosystems, Waltham, MA, USA) or an AutoFlex Speed (Bruker, Billerica, MA, USA) MALDI-ToF mass spectrometer, using positive reflector mode. Data were analyzed with Data explorer version 4.8 (Applied Biosystems, Waltham, MA, USA) or flexAnalysis 3.4 (Bruker, Billerica, MA, USA). Peaks with m/z values that matched expected galactosylated acceptors were interpreted as such products.

2.5. Screening for Transglycosylation with Raffinose or LBG and MALDI-TOF MS Analysis

Transglycosylation capabilities with natural substrates were evaluated by triplicate incubations of 12 nkat/mL BoGal36A or 8 nkat/mL Aga27A with 0.4M of the oligosaccharide raffinose or 0.4 w/v% of the polymeric substrate LBG at 37 °C for 24 h. The reaction buffers, acceptor concentrations and MALDI-TOF analysis were as described above in Section 2.4.

2.6. Enzyme Stability in the Presence of Acceptors

To determine enzyme stability in the presence of the acceptors, BoGal36A or Aga27A was incubated with individual acceptors at the concentrations listed above in Section 2.4, and the retained activity was determined using the assay above. The buffers were 50 mM sodium citrate pH 6 (BoGal36A) and 50 mM sodium acetate pH 4.5 (Aga27A) using triple incubations (each with triple activity assay) at 37 °C over 24 h (sampling at 0, 1, 3, 6 and 24 h). The retained activity was normalized against the 0-h sample, for which no immediate deactivation was observed.

2.7. Screening of Transglycosylation Products with Raffinose and LBG using HPLC

Following the MS analyses described above, further screening of the transglycosylation capacity of the two enzymes was carried out using high-performance liquid chromatography (HPLC) to confirm the production of quantifiable transglycosylation products. Reactions containing 12 nkat/mL BoGal36A or Aga27A with 0.4 M raffinose as the donor and acceptor of methanol, propanol or glycerol buffer, at earlier mentioned concentrations, and 50 mM sodium citrate pH 6 (BoGal36A) or 50 sodium acetate pH 4.5 (Aga27A) were incubated for 24 h at 37 °C. The enzymes were deactivated by heating at 95 °C for 10 min and then prepared for HPLC. Prior to injection, the samples were appropriately diluted with Milli-Q water and then mixed with 1:3 sample:acetonitrile (ACN) and filtered through a 0.2 µm polytetrafluoroethylene (PTFE) filter. Samples were then separated by hydrophilic interaction chromatography (HILIC) using a LUNA NH2 column (3 µm, 250 mm 4.6 mm) (Phenomenex, Torrance, CA, USA) (30 °C) on an UltiMate 3000 HPLC system (Thermo Scientific, Waltham, MA, USA) connected to a Corona^TM Veo^TM Charged Aerosol Detector (CAD) (Thermo Scientific, Waltham, MA, USA). In total, 10 µL of the sample was injected with a 1 mL/min isocratic flow of 30:70 H₂O:ACN. Galactose (0.1 to 10 mM), sucrose (0.1 to 5 mM), raffinose, glycerol and methyl galactoside were used as external standards and were used to suggest product identity based on elution time and for quantification of these molecules. Reaction mixtures were also analyzed with MALDI-TOF MS as described above.

2.8. Evaluation of Transglycosylation Capacity with Raffinose or LBG and Methanol using HPLC

After the HPLC screening described in the previous paragraph, the transglycosylation capacity of BoGal36A and Aga27A was evaluated using three types of reactions. In the first, each enzyme was incubated with raffinose and methanol under the same reaction conditions as described in Section 2.7. The second reaction consisted of 0.4 w/v% LBG with 10 v/v% methanol added in the appropriate buffer and incubated with 33 nkat/mL BoGal36A or 12 nkat/mL Aga27A for 24 h at 37 °C. The third reaction was with 40 mM raffinose and 10 v/v% methanol in the appropriate buffer, using 33 nkat/mL BoGal36A or 12 nkat/mL Aga27A with incubation up to 3 h at 37 °C (sampling at zero, half an hour, one two and three hours). All reactions were terminated and analyzed as above with the exception that samples were diluted at 1:2 (sample:ACN) instead of 1:3. Standards of methyl-α-galactopyranoside (0.5 to 5 mM) and galactose (1 to 5 mM) were used to quantify the products.

To compare the transglycosylation capacity between the two enzymes, the rate of transglycosylation over the rate of hydrolysis, r_S/r_H [10], was calculated for BoGal36A and Aga27A, respectively, according to Equation (1), with methyl galactoside being the transglycosylation product and galactose the hydrolysis product (molar concentrations are used).

$$\frac{{\mathrm{r}}_{\mathrm{S}}}{{\mathrm{r}}_{\mathrm{H}}}=\frac{\left[\mathrm{Methyl} \mathrm{galactoside}\right]}{\left[\mathrm{Galactose}\right]}$$

(1)

The yield in the reactions was calculated based on how much of the consumed glycosyl substrate was turned into methyl galactoside, hereafter referred to as product yield, Equation (2). The amount of consumed substrate was calculated as the sum of the released methyl galactoside and galactose (molar concentrations are used).

$$\mathrm{Product} \mathrm{yield}=\frac{\left[\mathrm{Methyl} \mathrm{galactoside}\right]}{\left[\mathrm{Methyl} \mathrm{galactoside}\right]+\left[\mathrm{Galactose}\right]}$$

(2)

3. Results and Discussion

3.1. Enzyme Preparation

BoGal36A was expressed and purified as described previously [5], with the exceptions listed in Section 2.2. The purified BoGal36A had a specific activity of 1170 nkat/mg when incubated with 1 mM pNP-Gal in 50 mM sodium citrate pH 6.0 at 37 °C for 10 min.

The purified BoGal36A was run on an SDS-PAGE gel together with the commercially available Aga27A (Figure S1). Both enzymes were >95% pure and had apparent molecular weights (m.w.) corresponding well with theoretical ones, with BoGal36A possessing an apparent m.w. of 83 kDa compared to the theoretical of 81 kDa while Aga27A had an apparent m.w. of 42 kDa compared to the theoretical of 40 kDa. The identity of the commercial Aga27A preparation was analyzed with peptide mass fingerprinting and, as expected, produced a significant hit to the guar α-galactosidase with UniProt entry P14749 (Figure S2) [27].

3.2. Initial Screening for Transglycosylation Ability Using pNP-Gal as Donor

Initial screening of the transglycosylation capacity of Aga27A and BoGal36A was carried out with pNP-Gal as the donor substrate. For this, reactions were set up with 40 mM pNP-Gal as the donor with methanol, propanol, glycerol, allyl alcohol or propargyl alcohol as potential acceptors. In addition, reactions with pNP-Gal as the sole reactant were set up in order to assess if self-condensation [28] occurred, i.e., a transglycosylation event where pNP-Gal functions as both the donor and acceptor. pNP-Gal was chosen as a screening substrate due to the liberation of pNP during catalysis (for both hydrolysis and transglycosylation), which produced a direct colorimetric indication as to whether the enzymes were active or not under the tested conditions. All reactions liberated pNP.

MALDI-TOF MS analysis of reactions with pNP-Gal as the sole substrate showed that Aga27A catalyzed self-condensation, generating both pNP-galactobioside and pNP-galactotrioside (Figure 1), while BoGal36A appeared to lack this ability. Both enzymes catalyzed transglycosylation using pNP-Gal as the donor and propanol as the acceptor (Figure 1). For Aga27A, self-condensation products were observed in these reactions as well. Similar data were obtained for the two enzymes when incubated with pNP-Gal and the other tested acceptors (Figures S3 and S4). Thus, both enzymes used all acceptors with pNP-Gal as glycosyl donor, summarized in

Table 1. Screening of transglycosylation products of BoGal36A and Aga27A with pNP-Gal as donor substrate and different acceptors. N.D., not detected.

Acceptor	BoGal36A	Aga27A	Product
pNP-Gal	N.D.	+	pNP-Gal-Gal
10% Methanol	+	+	Methyl-Gal
5% Propanol	+	+	Propyl-Gal
10% Glycerol	+	+	Glyceryl-Gal
5% Allyl-OH	+	+	Allyl-Gal
2.5% Propargyl-OH	+	+	Propargyl-Gal

Enzyme, pNP-Gal and acceptor incubations and product analysis with MALDI-TOF MS as presented in Figure 1, Figures S3 and S4. Enzyme concentration 8 nkat/mL, pNP-Gal concentration 40 mM, acceptor concentration (v/v %): 10% methanol, 5% propanol, 10% glycerol, 5% allyl alcohol, 2.5% propargyl alcohol. Time 1 h, temperature 37 °C.

.

As both enzymes displayed a capacity to utilize a wide range of acceptors, further evaluations of their properties were carried out to determine their potential as catalysts in transglycosylation reactions.

3.3. Screening for the Ability to Utilzse Raffinose as the Donor Substrate

Having established t’e enzyme’s ability to perform transglycosylation with pNP-Gal and various acceptors, their ability to utilize natural donor substrates was investigated. The trisaccharide raffinose (0.4 M) was investigated as a donor saccharide, with the same conditions as Section 3.2. The reactions were extended to 24 h since they were expected to be slower than for pNP-Gal [5] and analyzed with MALDI-TOF MS (Figures S5 and S6). The most striking difference compared to using pNP-Gal was that BoGal36A was able to catalyze self-condensation, likely producing a raffinose molecule with an extra galactose unit added (Figure 2).

While Aga27A had the ability to utilize all the screened acceptors with raffinose as a donor, BoGal36A used methanol, propanol and glycerol but showed no signs of performing transglycosylation with allyl alcohol or propargyl alcohol under the tested conditions (

Table 2. Screening of transglycosylation products from incubations with raffinose as donor and different acceptors with BoGal36A and Aga27A. N.D., not detected.

Acceptor	BoGal36A	Aga27A	Product
Raffinose	+	+	Raffinose-Gal
10% Methanol	+	+	Methyl-Gal
5% Propanol	+	+	Propyl-Gal
10% Glycerol	+	+	Glycerol-Gal
5% Allyl-OH	N.D.	+	Allyl-Gal
2.5% Propargyl-OH	N.D.	+	Propargyl-Gal

Enzyme, raffinose and acceptor incubations and product analysis with MALDI-TOF MS as presented in Figure 2, Figures S5 and S6. Enzyme concentration 12 nkat/mL, raffinose concentration 0.4 M, acceptor concentration (v/v %): 10% methanol, 5% propanol, 10% glycerol, 5% allyl alcohol, 2.5% propargyl alcohol. Time 24 h, temperature 37 °C.

, Figure 2 and Figure S5). This is in contrast to when pNP-Gal was used as the donor substrate, where it was able to use both of these acceptors. The reason for this observation is not clear, but contributing factors may be comparably slower kinetics for raffinose [5], giving too little product formation to allow detection and/or occurrence of secondary hydrolysis [10]. In addition, enzyme active site architecture and substrate interactions may contribute, as discussed in Section 3.8.

3.4. Screening for the Ability to Utilize Locust Bean Gum as the Donor Substrate

MALDI-TOF MS screening for transglycosylation with the polymeric galactomannan locust bean gum (0.4 w/v% LBG) as the donor substrate was carried out with acceptor concentrations as described in Section 2.5 in the presence of either 12 nkat/mL BoGal36A or 8 nkat/mL Aga27A at 37 °C over 24h. Aga27A maintained its ability to perform transglycosylation (Figure 3 and Figure S7) with all of the screened acceptors except propargyl alcohol (

Table 3. Screening of transglycosylation products with BoGal36A and Aga27A using LBG as a donor and different acceptors. N.D., not detected.

Acceptor	BoGal36A	Aga27A	Product
10% Methanol	+	+	Methyl-Gal
5% Propanol	N.D.	+	Propyl-Gal
10% Glycerol	N.D.	+	Glycerol-Gal
5% Allyl-OH	N.D.	+	Allyl-Gal
2.5% Propargyl-OH	N.D.	N.D.	Propargyl-Gal

Enzyme, LBG and acceptor incubations and product analysis with MALDI-TOF MS are presented in Figure 3 and Figure S7. Enzyme concentration 8 nkat/mL, LBG concentration 0.4 w/v%, acceptor concentration (v/v %): 10% methanol, 5% propanol, 10% glycerol, 5% allyl alcohol, 2.5% propargyl alcohol. Time 24 h, temperature 37 °C.

). For BoGal36A, on the other hand, a peak corresponding to transglycosylation products was only detected with methanol as the acceptor.

The ability of Aga27A to utilize a polymeric substrate as a donor substrate with a wide range of acceptor molecules makes it a particularly interesting enzyme. We recently showed that Aga27A was able to produce allyl galactoside when incubated with LBG [25], and to our knowledge, only one other α-galactosidase was described where a polymeric substrate was used as glycosyl donor, with glycerol being the only evaluated acceptor [21].

3.5. Analysis of Transglycosylation Products of BoGal36A and Aga27A Using HPLC

To screen for the extent of the transglycosylation ability of the two enzymes, product formation in reactions with BoGal36A or Aga27A with raffinose as the donor substrate and 10 v/v% methanol, 5 v/v% propanol or 10 v/v% glycerol as the acceptor was evaluated by HPLC. An initial HPLC screening was carried out in order to possibly detect transglycosylation products (Figure 4). For this, the reactions were incubated at 37 °C for 24 h and had a substrate consumption between 25 and 60%, determined by the amount of released sucrose, a product from the cleavage of the galactosyl unit from raffinose. For both BoGal36A and Aga27A, the incubations with methanol, propanol and glycerol had peaks at 3.8, 3.4 for and 4.4 min, respectively, none of which were present in the hydrolysis controls for either enzyme. The reactions were confirmed to contain the m/z of the expected transglycosylation product for the respective acceptor with MALDI-ToF MS (Figure 4), and the elution time of the additional peak with methanol as acceptor matched the elution time of the methyl galactoside standard.

3.6. Stability in the Presence of Acceptor Molecules

In the previous sections, we presented a rapid screening methodology using MALDI-ToF MS, which was followed by a screening of detectable transglycosylation products using HPLC. While some of the screening reactions displayed the absence of transglycosylation products, this does not exclude that products would be formed under other conditions using the same enzyme and reactants. There is a possibility that factors such as reduced activity towards the more complex substrates raffinose and LBG [5], secondary hydrolysis of transglycosylation products and enzyme instability in the presence of acceptors affected product formation. Previous studies of α-galactosidase transglycosylation reactions with alcohols showed an enzyme-destabilizing effect by alcohol acceptors [17,29]. In order to address this destabilization effect and its potential influence on the results observed in the previous sections, the stability of both BoGal36A and Aga27A in the acceptors was investigated.

Both enzymes displayed high stability when incubated at 37 °C for 24 h in the buffer. BoGal36A retained 80% and Aga27A 56% of their respective initial activity (Figure 5). Very similar values were obtained in the presence of 10 v/v% glycerol (data not shown). When incubating in the presence of the other acceptors, both enzymes were more unstable, but no immediate decrease in activity was observed upon the addition of any of the acceptors. Propargyl alcohol (2.5 v/v%) had the most destabilizing effect, BoGal36A had no detectable activity after one hour and Aga27A had <5% remaining activity after six hours. While BoGal36A had almost no activity left in the presence of 5 v/v% allyl alcohol or propanol after three hours, Aga27A had 33% and 19% remaining activity, respectively. Overall, Aga27A was the more stable enzyme of the two under the tested conditions, with the exception of 10 v/v% methanol. In this condition, BoGal36A had higher stability compared to Aga27A, retaining 42% of the initial activity after 24 h, while Aga27A had no detectable activity at this timepoint but had 28% residual activity at 3 h.

The stability data show that both enzymes displayed activity in the presence of all acceptors, which is consistent with the release of pNP when using pNP-Gal as a donor (Section 3.2). With the exception of glycerol, activity declined with time, the effect being most pronounced with BoGal36A incubated with propargyl alcohol, followed by allyl alcohol and propanol (Figure 5). Since propargyl alcohol and allyl alcohol did not generate products with BoGal36A using raffinose and LBG (Table 2 and Table 3) but did with pNP-Gal (Table 1), enzyme destabilization in combination with slower kinetics using the more complex natural substrates raffinose and LBG [5] is a plausible cause. The combined data also support the strategy of using pNP-Gal for screening general acceptor usage since it provides easy overall activity monitoring and a faster reaction is an advantage in cases where enzyme destabilization occurs.

3.7. Evaluation of Transglycosylation Capacity with Natural Substrates

3.7.1. Comparison of Transglycosylation Properties with Raffinose and Locust Bean Gum

Methanol was chosen as the acceptor for continued comparison of the transglycosylation properties of the two enzymes using HPLC, as both enzymes retained at least 30% of their initial activity during the first three hours under the used conditions and the fact that a product standard (methyl galactoside) was commercially available. Reactions were set up with raffinose or LBG as the donor and 10 v/v% methanol as the acceptor. First, a 24-h incubation time was chosen as quantitative product formation was confirmed at this time point with raffinose (Section 3.5). In addition, to quantify the transglycosylation product methyl galactoside, the hydrolysis product galactose was also quantified. Quantification of both products can be used to gain insight into the transglycosylation capacity of retaining GHs by calculating r_S/r_H, Equation (1) [10,25]. Theoretically, this ratio reflects the propensity for the glycosyl unit in the enzyme–glycosyl intermediate to be released by a water molecule (hydrolysis) or an acceptor molecule (transglycosylation).

The first comparison of transglycosylation capacity was made in reactions with 0.4 M raffinose as the donor and 10 v/v% methanol incubated with 12 nkat/mL of BoGal36A or Aga27A for 24 h at 37 °C. Both enzymes produced a substantial amount of both transglycosylation and hydrolysis products (

Table 4. Comparative table of transglycosylation capacity of BoGal36A and Aga27A with methanol as acceptor and raffinose or LBG as a donor.

Donor	Enzyme	[Me-Gal] (mM)	[Gal] (mM)	Product Yield (%) ([Me-Gal]/([Me-Gal]+[Gal]))	rS/rH ([Me-Gal]/[Gal])
0.4 w/v% LBG	BoGal36A (24 h)	0.7 (±0.1)	2.1 (±0.4)	27	0.33 (±0.06)
	Aga27A (24 h)	1.0 (±0.1)	2.4 (±0.1)	30	0.42 (±0.05)
400 mM raffinose	BoGal36A (24 h)	41 (±7)	70 (±11)	37	0.58 (±0.01)
	Aga27A (24 h)	35 (±11)	55 (±13)	39	0.63 (±0.05)
40 mM raffinose	BoGal36A (30 min)	1.3 (±0.1)	1.4 (±0.2)	48	0.99 (±0.06)
	BoGal36A (2h)	3.1 (±0.1)	4.2 (±0.3)	42	0.74 (±0.07)
	Aga27A (30 min)	2.3 (±0.3)	3.3 (±0.4)	41	0.71 (±0.03)
	Aga27A (3 h)	4.0 (±0.6)	6.4 (±1.0)	38	0.63 (±0.04)

). With these values, the fraction of transglycosylation products (methyl galactoside) of the total release of galactosyl units can be calculated and is presented as the “Product yield”, Equation (2), in Table 4 (37% for BoGal36A and 39% for Aga27A). The product ratio [Me-Gal]/[Gal] is equal to the apparent ratio of the rate of synthesis over the rate of hydrolysis (r_S/r_H) over the time period used (see Section 2.8). The product ratios of the 24h 400mM raffinose incubations presented in Table 4 are informative for a first comparison of the transglycosylation propensity. BoGal36A had a r_S/r_H of 0.58 (±0.01) while Aga27A reached a r_S/r_H of 0.63 (±0.05) (Table 4). The values for transglycosylation yields are similar to those that were reported for other α-galactosidases with other types of donors and acceptors [17,23,29], including Aga27A with LBG as the donor and allyl alcohol as the acceptor [25]. As can be concluded from the presented values, both enzymes appear to have more or less equivalent transglycosylation capacity under the tested conditions.

In order to evaluate the transglycosylation capacity against polymeric substrates, reactions with 0.4% (w/v) LBG and 10% (v/v) methanol were set up as described in Section 3.4, with the exception of the concentration of BoGal36A which was increased to 33 nkat/mL to reach sufficient product for quantification. As was the case with raffinose, when using LBG, both enzymes produced similar amounts of methyl galactoside and had comparable yields to one another (Table 4). The data gives methyl galactoside product yields of 27% and 30% for BoGal36A and Aga27A, respectively. These product yields using LBG are lower than with raffinose but very similar to the product yield previously reported for Aga27A with 3% LBG using 10% allyl alcohol as the acceptor [25]. Slower kinetics with LBG galactomannan compared to oligomeric substrate, as observed for BoGal36A [5], could potentially contribute to a higher degree of secondary hydrolysis and lower yield. The r_S/r_H values were in the same range for both enzymes (0.42 for Aga27A 0.33 and for BoGal36A) but 33–44% lower than the values for raffinose (Table 4).

3.7.2. Determination of Initial r_S/r_H Ratios and Assessment of Secondary Hydrolysis in Raffinose and Methanol Reactions

In order to further compare and evaluate the transglycosylation properties of the two enzymes, the product formation over the initial reaction time was analyzed. Reactions with 40 mM raffinose and 10 v/v% methanol using 33 nkat/mL BoGal36A or 12 nkat/mL Aga27A at 37 °C were followed by HILIC HPLC-analysis for up to 3 h.

Both enzymes had continued production throughout the 3-h time period (Figure 6a,b), with the highest productivity during the first half-hour. Between the two- and three-hour timepoints, BoGal36A had a slight decrease in the methyl galactoside concentration (Figure 6a) while it increased slightly for Aga27A (Figure 6b). The decrease in methyl galactoside concentration is an indication of secondary hydrolysis in the BoGal36A reaction, which was further supported when the apparent r_S/r_H was plotted for both enzymes (Figure 6c,d). Both BoGal36A and Aga27A had the highest value of r_S/r_H at the 30-min timepoint with a ratio of 0.99 and 0.71, respectively. For BoGal36A, this value then decreased throughout the reaction time, going down to 0.64 after three hours (Figure 6c), while for Aga27A, the apparent r_S/r_H remained stable throughout the incubation time, with only a slight decrease to 0.63 after three hours (Figure 6d).

Based on the concentration of products at 30 min and the timepoint with the highest concentration of methyl galactoside for the respective enzyme (Table 4), product yields were calculated. For Aga27A, the product yield after 30 min was 41% and 38% after three hours. BoGal36A had product yields of 48% and 42% at 30 min and 2h, respectively. This is a sign of secondary hydrolysis for BoGal36A, as reflected by the decrease in r_S/r_H over time (Table 4, Figure 6c).

Thus, the two enzymes seem to be quite equal in transglycosylation capacity, even if BoGal36A displayed a 39% higher initial r_S/r_H (Table 4). This advantage appears to be counteracted by secondary hydrolysis by BoGal36A. Similar findings were observed for β-galactosidases and other GHs [10,28,30]. When comparing the yield and the r_S/r_H-values obtained at 40 mM with those obtained at 400 mM raffinose (Section 3.7.1), the differences are surprisingly small (Table 4), indicating that a large donor concentration range could be applicable.

3.8. Evaluation of Transglycosylation Capacity Aga27A and BoGal36A

In this paper, we presented a transglycosylation screening strategy for retaining α-galactosidases, ultimately using natural donor saccharides (i.e., raffinose and galactomannan) and various acceptors. Mass spectrometry was used as the first rapid product screening method. The results with Aga27A show the strength of utilizing pNP-Gal as an initial screening substrate for evaluation of the transglycosylation capacity of enzymes that are known to act on polymeric substrates, as the acceptor utilization with pNP-Gal translated well to LBG. pNP-Gal has several advantages over LBG as an initial screening substrate, e.g., that α-galactosidases often have a higher catalytic efficiency toward pNP-Gal [5,31,32]. This would allow for high throughput screening of a range of different conditions, e.g., varying enzyme, donor or acceptor concentrations. The procedure may be expanded to other acceptors, although several of the herein used acceptors generated potentially applicable glycosides. Allyl glycosides may effectively be used in polymerizations or thiol-ene glycosylation reactions [33]. Enzymatically synthesized alkyl glycosides gain interest as surfactants [34]. Furthermore, galactosylglycerol has shown prebiotic potential [22].

The lower stability for BoGal36A in the presence of some of the acceptors (Figure 5), in combination with slower kinetics with raffinose, and even more so with galactomannan, compared to pNP-Gal [5], likely negatively affected the product formation for these donors (Table 2, Table 3). These limitations can be overcome by molecular enzyme development [28] and choosing and optimizing reaction conditions (time, temperature) and usage of stabilizing components for a given application [1,2]. These possibilities also further motivate the use of pNP-Gal as a first screening substrate.

BoGal36A was able to perform self-condensation with raffinose but not with pNP-Gal (Table 1 and Table 2). The discrepancy in self-condensation behavior could have several explanations. The increase in substrate concentration is one possible explanation. It was shown that increased acceptor concentrations, both for saccharide and non-saccharide acceptors, can increase the yield in transglycosylation reactions with α-galactosidases [15,20,35,36]. Using longer saccharides as acceptors was shown to have a beneficial effect on the amount of transglycosylation products synthesized by a GH27 α-galactosidase [16]. In the paper, they suggested that the enzyme has several saccharide binding subsites and that these interactions would be beneficial for transglycosylation [16], which was also proposed for β-mannanases in regard to +2 subsites [37,38]. Subsite architecture may also play an important role for non-saccharide acceptors, highlighted in Malbert et al. where a single amino acid substitution in the acceptor region of a glucansucrase led to an 8-fold increase in transglycosylation towards a flavonoid [39].

Following the performed screening and evaluation, optimizing reaction conditions may increase the yield. In addition, an interesting approach to fundamentally increase the transglycosylation capacity of α-galactosidases could be the structure-independent mutagenesis approach which successfully was applied to enzymes of several GH-families, demonstrating a 2 to 50-fold increase of r_S/r_H [28]. Since the parental wild-type enzyme for this approach should have a reasonable transglycosylation capacity for the targeted donor/acceptor pair [28], the screening procedure for α-galactosidases presented here would be a valuable tool for the selection of candidates for enzyme engineering.

The ability to effectively utilize natural saccharides from renewable sources, such as galactomannan and galactoglucomannan, is essential for biorefinery approaches with the increasing interest in biocatalysis [40,41,42]. While transglycosylation reactions with galactomannan as the donor substrate were explored in some studies, these were often limited to single GH-family studies [14,21]. We foresee that our successful enzyme synergy approach for transglycosylation with polymeric substrates using an α-galactosidase and a β-mannanase [25] can be further explored. A key element in developing this approach is the successful identification of enzymes with good transglycosylation capacity, which would benefit from an established screening procedure.

Combining synergistic enzymes with approaches such as the enzyme engineering method discussed above [28] could lead to improvements in overall transglycosylation yields and increase the possibility of utilizing renewable biomass for the production of chemicals such as alkyl glycosides or reactive glycosyl monomers [14,21,34].

4. Conclusions

Both BoGal36A and Aga27A utilized all of the tested acceptor’s methanol, propanol, allyl alcohol, propargyl alcohol and glycerol with pNP-Gal as the donor. The altered acceptor utilization of BoGal36A with the more complex substrates is probably not dependent on the structure per se, as discussed above, and could probably be overcome by changing the reaction conditions, which motivates the use of pNP-Gal as the initial donor substrate for the screening of acceptor utilization, even when complex natural saccharides are the ultimate donors. Overall, Aga27A appears to be the more promising of the two enzymes. It displayed higher stability in all of the acceptors with the exception of methanol. Under the tested conditions, Aga27A utilized a wider combination of acceptors and donors compared to BoGal36A and had a lower propensity for secondary hydrolysis. Further work includes exploring the mechanistic principles underlying the differences between BoGal36A and Aga27A, as this could give a better understanding of the factors that govern an enzyme’s capability to perform transglycosylation and propensity for the secondary hydrolysis of transglycosylation products.