Structural Comparison of Different Galacto-oligosaccharide Mixtures Formed by β-Galactosidases from Lactic Acid Bacteria and Bifidobacteria

The LacLM-type β-galactosidase from Lactobacillus helveticus DSM 20075 expressed in both Escherichia coli (EcoliBL21Lhβ-gal) and Lactobacillus plantarum (Lp609Lhβ-gal) was tested for their potential to form galacto-oligosaccharides (GOS) from lactose. The Lh-GOS mixture formed by β-galactosidase from L. helveticus, together with three GOS mixtures produced using β-galactosidases of both the LacLM and the LacZ type from other lactic acid bacteria, namely, L. reuteri (Lr-GOS), L. bulgaricus (Lb-GOS), and Streptococcus thermophilus (St-GOS), as well as two GOS mixtures (Br-GOS1 and Br-GOS2) produced using β-galactosidases (β-gal I and β-gal II) from Bifidobacterium breve, was analyzed and structurally compared with commercial GOS mixtures analyzed in previous work (Vivinal GOS, GOS I, GOS III, and GOS V) using high-performance anion-exchange chromatography with pulsed amperometric detection (HPAEC-PAD), high-performance size-exclusion chromatography with a refractive index (RI) detector (HPSEC-RI), and one-dimensional 1H NMR spectroscopy. β-Galactosidases from lactic acid bacteria and B. breve displayed a preference to form β-(1→6)- and β-(1→3)-linked GOS. The GOS mixtures produced by these enzymes consisted of mainly DP2 and DP3 oligosaccharides, accounting for ∼90% of all GOS components. GOS mixtures obtained with β-galactosidases from lactic acid bacteria and B. breve were quite similar to the commercial GOS III mixture in terms of product spectrum and showed a broader product spectrum than the commercial GOS V mixture. These GOS mixtures also contained a number of GOS components that were absent in the commercial Vivinal GOS (V-GOS).


■ INTRODUCTION
β-Galactosidases (lactases; EC 3.2.1. 23), which catalyze the hydrolysis of the β-1,4-D-glycosidic linkage of milk sugar lactose and its structurally related compounds to yield the monosaccharides glucose and galactose, are of importance to several biotechnological processes in the food industry. 1 The well-known applications of β-galactosidases in lactose hydrolysis include development of low-lactose and lactose-free dairy products for lactose-intolerant consumers, improvement of technological and sensory properties of foods containing lactose, and utilization of cheese whey to diminish economic and environmental problems. 2−5 In addition to their hydrolytic activity, β-galactosidases possess transgalactosylation activity to form galacto-oligosaccharides (GOS) from lactose. The mechanism of these enzymes involves two steps of which the cleavage of lactose and the formation of a covalently linked galactosyl−enzyme intermediate while releasing glucose occurs in the first step, followed by the transfer of galactosyl moieties from the donor sugar lactose to an acceptor, which can be water, lactose, or any of the sugar products in the reaction mixture. 6,7 In the trans-galactosylation mode, galactosyl moieties of lactose are transferred to other saccharides instead of water to yield GOS with different degrees of polymerization as reaction products. 8,9 GOS are known as a mixture of carbohydrates consisting of non-lactose disaccharides, various trisaccharides, and higher oligosaccharides with the degree of polymerization up to eight (DP8). 10 Sugar units in GOS typically join together with β-(1→4) and β-(1→6) linkages, but β-(1→2) and β-(1→3) linkages are also reported. 11−15 Currently, GOS are considered as dominant functional food ingredients fulfilling the criteria of "prebiotics" 16 as reported in the literature on their modulatory effects on gut microbiota including stimulation of beneficial bacteria such as bifidobacteria and lactobacilli as well as inhibition of "undesirable" bacteria, maintenance of gut health, beneficially affecting the bowel functions, and colitis prevention. 17,18 In addition to these confirmed beneficial effects, postulated physiological benefits of GOS include suppression of intestinal disturbances and colorectal cancer, lowering serum cholesterol levels, reducing risk of cardiovascular diseases, increasing absorption and retention of divalent minerals, and promoting immune responses. 10,16,19,20 Although GOS are not a significant element in human breast milk, they were shown to have a similar bifidogenic effect 21 and therefore are used as key fortificants in infant formulas. 19 A number of β-galactosidases from microbial sources have been isolated and purified for the production of GOS. 3 Among the β-galactosidase-producing microorganisms, some species of lactic acid bacteria, predominantly lactobacilli, and bifidobacteria have attracted great attention and have been studied intensively because of their probiotic potential and their generally recognized as safe (GRAS) status. It is anticipated that GOS synthesized by lactobacillal and bifidobacterial βgalactosidases would have structural characteristics, which will be preferentially utilized by this group of beneficial intestinal microorganism that facilitates the growth and metabolic activity of gut microbiota. 8,22 In the present study, we describe the formation of GOS from lactose using the recombinant LacLM-type β-galactosidase from the industrially important lactic acid bacterium Lactobacillus helveticus DSM 20075. Individual GOS components of the resulting GOS mixtures were identified and structurally compared with different GOS mixtures formed by β-galactosidases from several lactic acid bacteria and bifidobacteria, including Lactobacillus reuteri, Lactobacillus delbrueckii subsp. bulgaricus, Streptococcus salivarius subsp. thermophilus, and Bifidobacterium breve, in comparison with some commercial GOS products studied previously. 23 Significantly, the information on the GOS profiles obtained from this study will provide deeper insights into lactic acid bacterial and bifidobacterial β-galactosidases with respect to the spectrum of their GOS products.

■ MATERIALS AND METHODS
Chemicals. All chemicals and analytical standards of D-glucose, Dgalactose, and lactose used in this study were purchased from Sigma-Aldrich (St. Louis, MO) unless stated otherwise and were of the highest purity available. Vivinal GOS (V-GOS; 59% GOS, 21% lactose, 19% glucose, and 1% galactose) and the GOS mixtures, GOS I, GOS III, and GOS V are the commercial GOS standards from previous work. 23 Isopropyl-β-D-thiogalactopyranoside (IPTG) was obtained from Roth (Karlsruhe, Germany). Peptide pheromone IP-673 was synthesized by Caslo (Lyngby, Denmark). The Glucose (HK) Assay Kit for the determination of D-glucose was obtained from Megazyme (Wicklow, Ireland).
β-Galactosidase Assays. The determination of β-galactosidase activity was performed in 50 mM sodium phosphate buffer (NaPB; pH 6.5) at 30°C for 10 min using 22 mM o-nitrophenyl-β-Dgalactopyranoside (oNPG) or 600 mM lactose as previously described. 24 When the artificial chromogenic substrate oNPG was used, the change in absorbance due to the release of o-nitrophenol (oNP) was measured at 420 nm using a spectrophotometer (Beckman DU 800, Palo Alto, CA). One unit of oNPG activity was defined as the amount of β-galactosidase liberating 1 μmol of oNP per minute under the specified conditions. When the natural substrate lactose was used, the release of glucose in the reaction mixture was determined using the Glucose (HK) Assay Kit according to the manufacturer's instructions. One unit of lactose activity referred to the amount of βgalactosidase liberating 1 μmol of glucose per minute under the specified condition.
Preparation of Recombinant β-Galactosidases from L. helveticus. Escherichia coli BL21 Star (DE3) carrying pET21-lacLMLh and Lactobacillus plantarum TLG02 carrying p609lacLMLh, which contain the β-galactosidase encoding genes lacLM of L. helveticus DSM 20075, were cultivated and used for production of recombinant L. helveticus β-galactosidases as described previously. 25 Heterologous expression of β-galactosidase in E. coli BL21 Star (DE3) was performed in 300 mL of Luria-Bertani (LB) broth containing 100 μg/mL ampicillin. The recombinant E. coli strain was grown at 37°C with shaking to an OD 600 of 0.6. IPTG was then added into the culture to a final concentration of 1 mM, and the culture was further incubated at 25°C for 18 h. The expression of β-galactosidase in L. plantarum TLG02 carrying p609lacLMLh was performed in 1 L of MRS broth. The recombinant L. plantarum strain was grown at 30°C to an OD 600 of 0.3. Inducing peptide pheromone IP-673 26 was then added to a final concentration of 25 ng/mL, and the culture was further incubated at 30°C for 4 h. The cultures were harvested by centrifugation (4000g, 4°C for 15 min), washed twice with 50 mM NaPB (pH 6.5), and resuspended in the same buffer containing 1 mM PMSF. Cell disruption was performed using a French press (Aminco, Silver Spring, MD). After centrifugation (10,000g for 15 min at 4°C), the crude recombinant enzymes from E. coli BL 21 (DE3) (EcoliBL21Lhβ-gal) and L. plantarum TLG02 (Lp609Lhβ-gal) were collected and stored at −20°C. EcoliBL21Lhβ-gal was then purified to apparent homogeneity. The purification was carried out by immobilized metal affinity chromatography (IMAC) using a prepacked 1 mL HisTrap HP column with Ni-Sepharose resin (GE Healthcare, Uppsala, Sweden) as described previously. 25 Active fractions were collected, desalted, and concentrated by ultrafiltration using an Amicon ultra centrifugal filter unit with a 30 kDa cutoff membrane (Millipore, MA, USA). The purified EcoliBL21Lhβ-gal was kept in 50 mM NaPB (pH 6.5) at 4°C.
GOS Formation by Recombinant β-Galactosidases from L. helveticus. Batch conversions of lactose for the formation of GOS were carried out using recombinant β-galactosidases from L. helveticus, crude enzyme Lp609Lhβ-gal, and purified enzyme EcoliBL21Lhβ-gal. The reactions were performed based on the enzyme properties obtained from our previous study. 25 Reaction conditions were 205 g/ L initial lactose concentration in 50 mM NaPB (pH 6.5) containing 1 mM MgCl 2 , β-galactosidase (1.5 U Lac /mL of reaction mixture), and constant agitation (300 rpm). Different reaction temperatures varying from 37−50°C were applied for crude enzyme Lp609Lhβ-gal, while those varying from 30−50°C were applied for purified enzyme EcoliBL21Lhβ-gal. Samples were withdrawn periodically to monitor their residual enzyme activities with oNPG as the substrate. Sugar compositions and size distribution of the GOS mixtures were analyzed by HPAEC-PAD, HPSEC-RI, and one-dimensional 1 H NMR spectroscopy.
Other GOS Preparations. The purified GOS mixtures Lr-GOS, free from monosaccharides and unconverted lactose, produced using recombinant LacLM-type β-galactosidase from L. reuteri L103 were prepared as previously described by Maischberger et al. 14 Preparation of the GOS mixtures, Lb-GOS and St-GOS, using LacZ-type βgalactosidases from L. delbrueckii subsp. bulgaricus DSM 20081 15 and S. salivarius subsp. thermophilus DSM 20259, 2 respectively, and Br-GOS1 and Br-GOS2 using purified LacZ-type β-galactosidases β-gal I and β-gal II from B. breve DSM 20213, 12 respectively, was carried out under the conditions that yielded the highest GOS for each enzyme as reported previously. The GOS mixtures were analyzed by HPAEC-PAD, HPSEC-RI, and one-dimensional 1 H NMR spectroscopy.
High-Performance Anion-Exchange Chromatography with Pulsed Amperometric Detection (HPAEC-PAD). Carbohydrate contents in the reaction mixtures were analyzed by HPAEC-PAD at the Food Biotechnology Lab (BOKU; Vienna). HPAEC-PAD analyses were performed on a Dionex DX-500 system (Dionex Corp., Sunnyvale, CA) as previously described in detail. 27 The GOS yield (%) was defined as the percentage of GOS produced in the samples compared to initial lactose.
Detailed structural analyses of the GOS mixtures were also performed using HPAEC-PAD at the laboratory of Groningen Biomolecular Sciences and Biotechnology Institute (University of Groningen). GOS syrup samples were diluted by pipetting 20.2 ± 0.40 μL of GOS syrup into 1.990 ± 0.007 mL of Milli-Q water containing 1.01 ± 0.02 mM fucose as an internal standard. Reference GOS samples, used to identify peaks, were diluted ∼400 times first by diluting a volume of thick GOS syrup 1:1 with Milli-Q and then diluting it 200 times by adding ∼10 μL in 1.990 ±0.007 mL of Milli-Q water containing a 1.01 ± 0.02 mM fucose internal standard. GOS samples (10 μL, ∼1.0 mg/mL) were profiled on a Dionex ICS-3000 workstation (Dionex, Amsterdam, The Netherlands) equipped with a CarboPac PA-1 column (250 × 2 mm, Dionex) and an ICS-3000 ED pulsed amperometric detector (PAD). Oligosaccharides were eluted using a complex gradient of A: 100 mM NaOH, B: 600 mM NaOAc in 100 mM NaOH, C: Milli-Q water, and D: 50 mM NaOAc. The fractionations were performed at 0.25 mL/min with 10% A, 85% C, and 5% D in 25 min to 40% A, 10% C, and 50% D followed by a 35 min gradient to 75% A and 25% B directly followed by 5 min washing with 100% B and reconditioning for 7 min with 10% A, 85% B, and 5% D. 23,28,29 High-Performance Size-Exclusion Chromatography with a Refractive Index Detector (HPSEC-RI). Size distribution analysis of GOS samples (10 μL injections of ∼10 mg/mL GOS) was performed on a Rezex RSO-01 oligosaccharide Ag + (4%) column (200 × 10 mm; Phenomenex, Utrecht, The Netherlands) using a Waters 2690XE Alliance HPLC system (Waters, Etten-Leur, The Netherlands) equipped with a Waters 2410 RI detector. Elutions were carried out with Milli-Q water at a flow rate of 0.3 mL/min. The peaks were identified and quantified in relation to a calibration curve (0.05−20 mg/mL) of Glc, Gal, Lac, and isolated GOS fractions up to DP5.
NMR Spectroscopy. Samples of ∼10 mg of GOS were exchanged two times with 300 μL of 99.9 atom % D 2 O with intermediate lyophilization and finally dissolved in 650 μL of D 2 O containing 25 ppm acetone as an internal standard (δ 1 H 2.225). One-dimensional 1 H NMR spectra were recorded at a probe temperature of 25°C on a Varian Inova 500 spectrometer (NMR Department, University of Groningen, The Netherlands). Spectra were recorded with a 4500 Hz spectral width at 16,000 complex data points using a WET1D pulse to suppress the HOD signal. All spectra were processed using MestReNova 12 (Mestrelabs Research SL, Santiago de Compostela, Spain), applying manual phase corrections and Whittaker Smoother baseline corrections.
Statistical Analysis. All experiments and measurements were conducted at least in duplicate, and the standard deviation (SD) never exceeded 5%. The data are expressed as the mean ± SD with significant digits when appropriate.

■ RESULTS AND DISCUSSION
Lactose Conversion and GOS Formation by Recombinant β-Galactosidases from L. helveticus. The thermophilic lactic acid bacterium L. helveticus is extensively used as starter cultures for various milk fermentation processes. The strain is capable of utilizing lactose by exhibiting intracellular βgalactosidase activity. 30,31 It was shown previously that two recombinant β-galactosidases from L. helveticus DSM 20075, EcoliBL21Lhβ-gal and Lp609Lhβ-gal, were able to convert lactose using a batch mode of conversion and were found to be promising candidates for the production of prebiotic GOS. 25 It was also shown in our previous study that both enzymes are stable in the presence of high concentrations of lactose, retaining more than 75 and 60% of the initial activity, respectively, after 24 h of incubation at 50°C. 25 Therefore, there is great potential for application of these recombinant βgalactosidases in lactose conversion processes at temperatures up to 50°C.
In this study, we looked in more detail at the lactose conversions catalyzed by these two recombinant β-galactosidases from L. helveticus DSM 20075 at various process temperatures and performed a detailed analysis of the GOS mixtures formed. As shown in Figure 1A,B, lactose conversions catalyzed by the crude enzyme Lp609Lhβ-gal and the purified enzyme EcoliBL21Lhβ-gal approached completion (>98%) within 12 h at the examined process temperatures varying from 30 to 50°C. The rate of lactose conversion considerably increased with increasing reaction temperature. The conversions at 50°C occurred most rapidly and achieved ∼99% lactose conversion after 6 and 4 h with the crude enzyme Lp609Lhβ-gal and the purified enzyme EcoliBL21Lhβ-gal,

Br-GOS1
Br-GOS2 respectively. The conversion of lactose was thus faster when using the purified enzyme EcoliBL21Lhβ-gal than when using the crude enzyme Lp609Lhβ-gal under similar process conditions. The suppressed catalytic efficiency of the crude enzyme may be caused by the interference of the impurities, such as other proteins and substances in the cell-free extract. 32 The highest total GOS yield was ∼26% (∼53 g/L) at approximately 90% lactose conversion at 37, 42, and 50°C ( Figure 2A) when the crude enzyme Lp609Lhβ-gal was used in batch conversions with an initial lactose concentration of 205 g/L. The time needed to obtain 90% lactose conversion decreased from a reaction time of 7 h at 37°C to 5 h at 42°C and 4 h at 50°C ( Figure 1A). Trans-galactosylation may thus become more pronounced than hydrolysis at higher reaction temperatures 33 although the increased temperature hardly affected the maximum GOS yield. 15,33 Interestingly, the maximum GOS yield increased to ∼33% (∼68 g/L) when using the purified enzyme EcoliBL21Lhβ-gal under reaction conditions similar to those used for the crude enzyme Lp609Lhβ-gal. This highest GOS yield was reached when approximately 90% of the initial lactose was converted after 6 and 4 h of reaction at 30 and 37°C, respectively ( Figures 1B  and 2B). The maximum GOS yield obtained with the purified enzyme EcoliBL21Lhβ-gal was achieved in a much shorter time at elevated reaction temperatures after 2 h at 42°C (∼70% lactose conversion) and only after 1 h at 50°C (∼60% lactose conversion) (Figures 1B and 2B). The highest GOS yield of approximately 33% obtained with the purified enzyme EcoliBL21Lhβ-gal ( Figure 2B) is comparable to the reported GOS yields for other lactobacillal β-galactosidases, for instance 31% for L. pentosus β-galactosidase, 34 38% for L. reuteri βgalactosidase, 27 and 41% for L. sakei β-galactosidase. 13 Synthesis of GOS by β-galactosidases is kinetically controlled as a consequence of the competition between hydrolysis and trans-galactosylation, 9 and the GOS formed via trans-galactosylation are also potential substrates for hydrolysis. The total amount of GOS decreased dramatically after reaching their maximum yields as a result of the degradation of GOS by crude Lp609Lhβ-gal and purified EcoliBL21Lhβ-gal (Figure 2A,B, respectively), and consequently, hydrolysis prevailed over trans-galactosylation when the substrate lactose was depleted. This phenomenon was well observed and reported in the literature. 15,35 Identification of Individual GOS Components  Figure 3). The enzyme thus showed a strong preference for the formation of β-D-(1→3) and β-D-(1→6) linkages in accordance with other lactobacillal β-galactosidases reported previously. [13][14][15]34 The composition of the GOS mixture formed in batch conversion at 30°C with an initial lactose concentration of 205 g/L using the purified enzyme EcoliBL21Lhβ-gal was determined to be 19.3% disaccharides (DP2), 11.5% trisaccharides (DP3), and 1.2% tetrasaccharides (DP4) ( Table 2).
Structural Comparison of Different GOS Mixtures. GOS mixtures produced using β-galactosidases from different lactic acid bacteria (Lh-GOS, Lr-GOS, Lb-GOS, and St-GOS) and B. breve (Br-GOS1 and Br-GOS2) were analyzed and structurally compared with previously studied commercial GOS of known compositions (V-GOS, GOS I, GOS III, and GOS V). 23 All the GOS samples, except Lr-GOS, contain the monosaccharides (galactose, peak 1 and glucose, peak 2) and unconverted lactose (peak 5) (Figure 3). The Lr-GOS mixture was purified after the conversion to remove the monosaccharides and unconverted lactose. 14 A major component found in all GOS samples is the disaccharide allolactose (peak 4, Figure 3). Trans-galactosylation is known to involve intermolecular and intramolecular reactions of which the latter reaction is the result of direct galactosyl transfer to D-glucose to Graphical presentations of individual GOS structures are shown in; 24 [ ] represents branched elongation. b • structure present; ○ structure absent; and UNK is abbreviated for unknown structure. c Peak numbers correspond to the HPAEC-PAD chromatograms presented in Figure 3.
Generally, β-(1→6) and β-(1→3) linkages were formed predominantly in all GOS mixtures prepared using βgalactosidases from lactic acid bacteria and B. breve, not only with linear elongations but also with branched elongations (Table 1). There also was no evidence of β-(1→4) elongation for oligomers formed when using these enzymes. The results obtained from HPAEC-PAD analyses were supported by onedimensional 1 H NMR spectroscopy data. A reference library of 1 H NMR spectroscopy data for GOS was established previously. 23,28,29 In the one-dimensional 1 H NMR spectra ( Figure 4, Table 3), the presence of the major components allolactose (HPAEC-PAD peak 4, Figure 3) and 6′galactosyllactose (HPAEC-PAD peak 6, Figure 3) was confirmed by the presence of peak d (∼δ 5.22 ppm). The occurrence of Galp-(1→6)-Galp elements was reflected by the presence of the structural-reporter-group signal at ∼4.06 ppm (peak i). The presence of 6-substituted Glc was shown by the presence of peak e (δ 4.22) and peak f (δ 4.16). 23,28,29 The presence of 3-substituted reducing Gal (HPAEC-PAD peak 38, Figure 3) was confirmed by the NMR peak q (δ 4.10), corresponding to the H-5 signal of a 3-substituted reducing Gal residue in α-configuration. 28 The presence of 2-substituted reducing Glc was shown by the one-dimensional 1 H NMR peak at δ 5.45 (peak a) and occurred only in minor levels in St-GOS, Br-GOS1, and Br-GOS2.
The structural building blocks of different GOS mixtures strongly depend on the β-galactosidases used during their synthesis. β-Galactosidases from lactic acid bacteria and B. breve show the preference to form β-(1→6)and β-(1→3)linked GOS. It was reported that the administration of a GOS mixture containing β-(1→3) as well as β-(1→4) and β-(1→6) linkages proved to have a better bifidogenic effect than a mixture containing GOS with β-(1→4) and β-(1→6) linkages. 39 Recently, we have reported the specific growth stimulation of certain desired intestinal bacteria by one of our GOS mixtures, which contained mainly oligosaccharides of β-(1→6) and β-(1→3) glycosidic linkages, and our GOS mixture stimulated the growth of several probiotic strains more than the two commercial preparations, which contain mainly β-(1→ 4)-linked GOS. 40 GOS mixtures produced with these enzymes consist mainly of DP2 and DP3 oligosaccharides, accounting for ∼90% of all GOS components. The major disaccharides  Journal of Agricultural and Food Chemistry pubs.acs.org/JAFC Article present in these GOS mixtures are allolactose, 6′-galactobiose, 3′-galactobiose, and 3′-galactosyl glucose. Sanz et al. 41 investigated the prebiotic potential of a number of disaccharides and found that β-D-Galp-(1→6)-D-Gal (6′galactobiose), one of the major disaccharides in our GOS mixtures, is a highly prebiotic molecule. In our recent study, we have reported that some bifidobacteria preferentially utilized DP2 GOS and some lactobacilli often utilized only the DP2 and DP3 molecules. 42 Trisaccharides are the main constituents in Lr-GOS and Lb-GOS of which 6′-galactosyllactose and 3′galactosyllactose are the major components. The tetrasaccharide 3′-3′-digalactosyllactose is present in all GOS mixtures produced using lactic acid bacterial and bifidobacterial βgalactosidases, whereas 6′-6′-digalactosyllactose is only found in the GOS mixtures formed with lactic acid bacterial βgalactosidases (Lh-GOS, Lr-GOS, Lb-GOS, and St-GOS). The two tetrasaccharides, which are the β-D-Galp-(1→3) elongation of 6′-galactosyllactose and the β-D-Galp-(1→6) elongation of 3′-galactosyllactose, are found in the GOS mixtures formed with lactic acid bacterial β-galactosidases and also in Br-GOS2. These GOS mixtures contain a number of GOS components that are not present in the commercial Vivinal GOS mixture. The information on individual GOS obtained in this study as analyzed by HPAEC-PAD and confirmed with 1 H NMR spectroscopy has given more insights into the basic structural elements of the GOS mixtures produced using lactic acid bacterial and bifidobacterial β-galactosidases, and it will be of interest for studies on the correlation between individual GOS structures and their prebiotic potential.