4,6-α-Glucanotransferase activity occurs more widespread in Lactobacillus strains and constitutes a separate GH70 subfamily

Family 70 glycoside hydrolase glucansucrase enzymes exclusively occur in lactic acid bacteria and synthesize a wide range of α-d-glucan (abbreviated as α-glucan) oligo- and polysaccharides. Of the 47 characterized GH70 enzymes, 46 use sucrose as glucose donor. A single GH70 enzyme was recently found to be inactive with sucrose and to utilize maltooligosaccharides [(1→4)-α-d-glucooligosaccharides] as glucose donor substrates for α-glucan synthesis, acting as a 4,6-α-glucanotransferase (4,6-αGT) enzyme. Here, we report the characterization of two further GH70 4,6-αGT enzymes, i.e., from Lactobacillus reuteri strains DSM 20016 and ML1, which use maltooligosaccharides as glucose donor. Both enzymes cleave α1→4 glycosidic linkages and add the released glucose moieties one by one to the non-reducing end of growing linear α-glucan chains via α1→6 glycosidic linkages (α1→4 to α1→6 transfer activity). In this way, they convert pure maltooligosaccharide substrates into linear α-glucan product mixtures with about 50% α1→6 glycosidic bonds (isomalto/maltooligosaccharides). These new α-glucan products may provide an exciting type of carbohydrate for the food industry. The results show that 4,6-αGTs occur more widespread in family GH70 and can be considered as a GH70 subfamily. Sequence analysis allowed identification of amino acid residues in acceptor substrate binding subsites +1 and +2, differing between GH70 GTF and 4,6-αGT enzymes.

Today (November 2011), 135 GH70 sequences are reported in the CAZy database (Cantarel et al. 2009), of which 47 have been experimentally characterized (http:// www.cazy.org). Three-dimensional structures have been reported for the N-terminally truncated GTF-180 enzyme of L. reuteri 180 (Vujičić-Žagar et al. 2010), which makes a branched glucan consisting of α1→6 and α1→3 glycosidic bonds (Kralj et al. 2004a;van Leeuwen et al. 2009), the Nand C-terminally truncated mutan [(1→3)-α-glucan with about 10% α1→6 glycosidic bonds] synthesizer GTF-SI of Streptococcus mutans (Ito et al. 2011) and DSR-E of Leuconostoc mesenteroides NRRL B1299, which attaches glucose moieties on dextran via α1→2 glycosidic bonds (e.g., branches) (Brison et al. 2012). The catalytic domain of these enzymes has a (β/α) 8 -fold, which is circularly permutated compared to the arrangement of the catalytic domain in the evolutionary related enzymes of the GH13 and GH77 families. Note that also the GH13 amylosucrase enzymes synthesize α-glucans from sucrose (Potocki de Montalk et al. 2000), though only linear (1→4)-α-glucans, whereas all other GH13 enzymes either hydrolyze or disproportionate α-glycosidic bonds between two glucose moieties (Stam et al. 2006;Kelly et al. 2009b). In the GH70 family, as well as in the GH13 and GH77 families, the catalytic nucleophile (Asp1025; numbered according to , acid/base (Glu1063), and transition state stabilizer (D1136) are located at the C-terminal ends of the βstrands, as revealed by a sucrose bound structure of GTF-180 (Vujičić-Žagar et al. 2010). Despite the availability of crystal structures and the observation that mutations near the active site can alter the ratio of α-glycosidic bonds synthesized (Hellmuth et al. 2008), the reaction specificity of GH70 is not well understood, limiting the design of GH70 variants capable of synthesizing predefined α-glucan polymers.
Here, we describe the cloning, expression, and characterization of the second and third GH70 enzyme inactive with sucrose as donor substrate. The two GH70 genes of L. reuteri DSM20016 and L. reuteri ML1 encode enzymes that utilize maltooligosaccharides as substrate to synthesize linear α-glucans with α1→6-linked glucose segments at the non-reducing end of α1→4-linked glucose segments.

Materials and methods
Cloning of the gtfW gene A truncated version of the 4,6-αGT-W encoding gtfW gene (GenBank accession number ABQ83597) was amplified from the genomic DNA of L. reuteri DSM 20016 using the primers For (NcoI) 5′-GAT GCA TCC ATG GGC ATA GAT GGT AAG AAC TAC CAC TTC GC-3′ and Rev (BamHI) 5′-ATA TCG ATG GAT CCT ATT AGT GAT GGT GAT GGT GAT GAA TAT TTT CTT GGT TTG CAT AGT AAT CTG C-3′ and high fidelity DNA polymerase (Fermentas), and cloned in pET15b (Novagen), yielding pET15b_4,6-αGT-W. This N-terminally truncated version of 4,6-αGT-W (amino acids 458-1363) carries a fused (His) 6 -tag at its C-terminal. The removal of the N-terminal variable region is based on the cloning and expression of other GTFs, as this improves protein expression without affecting the catalytic properties (Kralj et al. 2004b).
Cloning of the gtfML4 gene Previously, the 3′-end fragment of the 4,6-αGT-ML4 encoding gtfML4 gene (GeneBank accession number AAU08003.1) has been identified in L. reuteri ML1, located upstream of a glucansucrase gene (gtfML1). The 5′-end of the gtfML4 gene was obtained by inverse PCR. Briefly, genomic DNA of L. reuteri ML1 was digested by BcuI or NsiI, and the products were ligated at a concentration of~10 ng/μl using T4 DNA ligase to obtain circular fragments. The ligation mixtures served as PCR template in an inverse PCR with the oligonucleotides HL115, 5′-TGA TCG TCC AGA TGT AGC-3′ and HL116, 5′-CCA GTT ACT TTC ATA GAG G-3′. This yielded PCR fragments of 2 kbp (BcuI) and 8 kbp (NsiI), which were cloned in the pCR-XL-TOPO vector (Invitrogen). The obtained plasmids were used for DNA sequencing, yielding the 5′-end of the gtfML4 gene.
For 4,6-αGT-W activity, the optimal pH and temperature were determined over the pH range of 3.5-6.5 (in 25 mM sodium acetate buffer) and the temperature range of 20-60°C, using 100 mM maltose as substrate by following the release of glucose in time using the GOPOP kit (Megazyme) (Kaper et al. 2007). This assay could not be used with 4,6-αGT-ML4 as it is hardly active with maltose. The optimal pH and temperature of 4,6-αGT-ML4 were therefore determined using 50 mM maltotetraose as substrate and by following appearance of products on TLC plates.
The thermal inactivation rate of 4,6-αGT-W was determined by incubating the enzyme at a concentration of 0.26 mg/ml in sodium acetate buffer (50 mM, pH 5.5) with CaCl 2 (1 mM) at temperatures from 20°C to 55°C in a water bath for 10 min. Samples were then cooled in an ice bath, and residual activity was measured by following the release of glucose using maltose as substrate. The T 50 is defined as the temperature at which 50% of the initial enzyme activity is retained after 10-min incubation.
Kinetic properties of 4,6-αGT-W were determined with maltose (0-500 mM) as substrate and by following the release of glucose. Reactions were initiated by the addition of 4,6-αGT-W enzyme at a concentration of 64 nM. The rates were fitted to the Michaelis-Menten equation.
High-pH anion-exchange chromatography Carbohydrate samples were analyzed on a 4×250 mm Car-boPac PA-1 column using a Dionex DX500 workstation (Dionex), run with a gradient of 30-600 mM sodium acetate in 100 mM NaOH (1 ml/min), and detected with an ED40 pulsed amperometric detector. Calibration was done by running samples with known concentrations of glucose and maltotetraose.
Matrix-assisted laser-desorption ionization time-of-flight mass spectrometry MALDI-TOF-MS experiments were performed on an Axima™ mass spectrometer (Shimadzu), equipped with a nitrogen laser (337 nm, 3 ns pulse width). Positive-ion mode spectra were recorded using the reflector mode at a resolution of 5000 FWHM and delayed extraction (450 ns). The accelerating voltage was 19 kV with a grid voltage of 75.2%; the mirror voltage ratio was 1.12, and the acquisition mass range was 200-3,000 Da. Samples (1 μl) were mixed in 1:1 ratio with 10 mg/ml 2,5-dihydroxybenzoic acid in acetonitrile/water01:1 (v/v). The compounds fly as sodium ion adduct.

Linkage analysis
Samples (~2 mg) were permethylated using CH 3 I and solid NaOH in Me 2 SO as described previously (Ciucanu and Kerek 1984). After hydrolysis with 2 M trifluoroacetic acid (2 h, 120°C), the partially methylated monosaccharides were reduced with NaBD 4 (2 h at room temperature, aqueous solution). Conventional work-up, comprising neutralization (by adding 4 M acetic acid) and removal of boric acid by co-evaporation with methanol, followed by acetylation with pyridine/acetic anhydride (1:1, v/v) (30 min, 120°C), yielded mixtures of partially methylated alditol acetates, which were analyzed by gas-liquid chromatography electron impact mass spectrometry.

NMR spectroscopy
Resolution-enhanced 1D 500-MHz 1 H NMR spectra were recorded in D 2 O on a Varian Inova Spectrometer (NMR Center, University of Groningen) or a Bruker DRX-500 spectrometer (Bijvoet Center, Department of NMR Spectroscopy, Utrecht University) at probe temperatures of 300 K. Prior to analysis, samples were exchanged twice in D 2 O (99.9 atm% D, Cambridge Isotope Laboratories, Inc.) with intermediate lyophilization and then dissolved in 0.6 ml D 2 O. Suppression of the HOD signal (on the Bruker DRX-500 spectrometer only) was achieved by applying a WEFT pulse sequence. Chemical shifts (δ) are expressed in parts per million by reference to internal acetone (δ 2.225 for 1 H).
Previously, it was shown that L. reuteri ML1 possesses a gtfML1 gene, which encodes a glucansucrase and that the direct upstream DNA sequence shows similarity to GH70 genes (Kralj et al. 2004a), though a full gene sequence was not obtained. This gene fragment was designated as (part of) gtfML4, and the encoded protein fragment showed most sequence similarity to 4,6-αGT-B of L. reuteri 121. Therefore, we isolated the full gtfML4 gene by inverse PCR. The gene is 4863 nucleotides and encodes a protein of 180 kDa, which has the typical arrangement of a GH70 family enzyme, with an N-terminal signal sequence of 39 amino acids as predicted by the SignalP 3.0 (Bendtsen et al. 2004), a large variable N-terminal region (amino acids 40-738), a catalytic domain (amino acids 739-1486), and a C-terminal region (amino acids 1487-1620). 4,6-αGT-ML4 shares 96% sequence identity with GTF-106B of L. reuteri TMW1.106, which has been reported to slowly hydrolyze sucrose but does not synthesize a polymer from sucrose (Kaditzky et al. 2008) and 86% and 46% sequence identity with 4,6-αGT-B and 4,6-αGT-W, respectively.

4,6-αGT-W and 4,6-αGT-ML4 disproportionate maltooligosaccharides
The substrate preference of the 4,6-αGT-W and 4,6-αGT-ML4 enzymes was explored by incubating them with various oligosaccharides. As shown by TLC, both enzymes use linear maltooligosaccharides [(1→4)-α-D-glucooligosaccharides] as substrate, forming a range of shorter and longer products (Fig. 2), but not sucrose (the typical substrate of GH70 enzymes), trehalose, raffinose, 1-kestose, nystose, isomaltose, and isomaltopentaose. The difference between the two enzymes is that 4,6-αGT-W already efficiently disproportionates the disaccharide maltose, whereas 4,6-αGT-ML4 requires maltotetraose or longer maltooligosaccharides (Fig. 2). Besides oligomeric material, TLC analysis showed that both enzymes also produce polymeric material [degree of polymerisation (DP) >10] (Fig. 2). Note that the polymeric material is not of very high molecular weight. For the homolog enzyme 4,6-αGT-B, we showed that the larger The initial rate of 4,6-αGT-W with maltose was determined by measuring the release of glucose (as maltose + maltose→a DP3 compound + glucose) using glucose oxidase, displaying a k cat of 37±4 s −1 and a K M of 150± 40 mM. The enzyme also processes maltopentaose, yielding maltotetraose + glucose (hydrolysis) or maltotetraose + a DP6 compound (transglycosylation). With 43 mM of maltopentaose as substrate, the initial rate of glucose formation was 56% of that of maltotetraose formation, showing that the enzyme is rather hydrolytic at the start of the reaction. However, when reaction products start to accumulate, transglycosylation becomes more efficient, as the reaction products are better acceptor substrates than maltopentaose. 4,6-αGT-W and 4,6-αGT-ML4 form α1→6 glycosidic bonds HPAEC analysis of the reaction products of the incubations of 4,6-αGT-W and 4,6-αGT-ML4 with maltooligosaccharides [(1→4)-α-D-glucooligosaccharides] of DP2 to DP7 (for DP7, see Suppl. Info. Fig. S2) revealed the formation of products with retention times different from those of maltooligosaccharides. The elution profiles of both enzyme incubations per DP also showed that the product ensembles are not identical, despite the fact that both reaction mixtures contain the same type of glycosidic linkages, at nearly identical ratios, as discussed below.
Purification and characterization of 4,6-αGT-W products To gain more insight in the carbohydrate structures made by 4,6-αGT-W, the reaction mixture obtained from maltose was separated by HPAEC (Fig. 4). The isolated compounds were analyzed by MALDI-TOF-MS for their molecular mass and for their purity in terms of DP. Fractions containing multiple oligosaccharides were separated again by HPAEC using a different elution gradient. The structures of the isolated compounds were established by 1D 1 H NMR spectroscopy making use of a 1 H NMR library data base of α-glucans  (Fig. 4). For 1 H NMR spectra, see (Suppl. Info. Fig. S4).

Discussion
It was generally accepted that all GH70 enzymes synthesize α-glucans from sucrose, until we recently demonstrated that L. reuteri 121 encodes a GH70 enzyme (called 4,6-αGT-B) that uses (1→4)-α-glucans, but not sucrose, as substrate Dobruchowska et al. 2012). Even though it is known that some GH70 enzymes possess a low disproportionating activity with α-glucans in addition to their main activity (Binder et al. 1983), it was surprising to find a GH70 enzyme to be so effective in utilizing (1→4)-α-glucans as glucose donor because the glycosidic linkage of sucrose is labile compared with the α1→4 glycosidic bond. Phylogenetic tree analysis of the GH70 protein sequences revealed that 4,6-αGT-B clusters with a few hypothetical proteins (Kralj et al. 2011) (Suppl. Info . Fig. S1). Here, we show that the hypothetical GH70 proteins of L. reuteri strains DSM 20016 and ML1 encode enzymes that convert maltooligosaccharides [(1→4)-α-D-glucooligosaccharides] into linear α-glucans (in fact isomalto/maltooligosaccharides) with about 50% α1→6 glycosidic bonds. It thus appears that 4,6-αGT activity is a common activity within the GH70 family.
The 4,6-αGT products are synthesized by cleaving off the non-reducing glucose moiety of a (1→4)-α-glucan and transferring the glucose moiety to the HO-6 function of the non-reducing end of an acceptor α-glucan chain [a (1→4)α-glucan or a (1→4)-α-glucan already elongated with a number of (α1→6)-linked glucose residues] (Fig. 6). The enzymes possess, in addition, minor hydrolytic and α1→4 transglycosylase activities. All three disproportionating GH70 4,6-αGT enzymes characterized so far synthesize α1→6 glycosidic bonds and occasionally an α1→4 glycosidic bond. In contrast, the common GH70 GTF enzymes synthesize all four types of α-glycosidic linkages possible. It appears unlikely that this is an intrinsic property of the non-sucrose utilizing type of GH70 enzymes. Therefore in the future α1→2 and α1→3 synthesizing non-sucrose utilizing type of GH70 enzymes may be found encoded by the ever increasing number of bacterial genome sequences, unless of course such synthetic capacities are not beneficial for the bacteria.
The distinct reaction specificities of 4,6-αGTs and GTFs Currently it is puzzling why 4,6-αGTs use maltooligosaccharides, but not sucrose, as glucose donor. Because GTFs cleave the glycosidic linkage between the glucosyl and fructosyl moieties of sucrose, and 4,6-αGTs the α1→4 glycosidic linkage between two glucosyl moieties, structural differences are expected at substrate binding subsite +1, which holds either a glucosyl or fructosyl unit (see Fig. 6 for subsite nomenclature). Whether the differences at subsite +1 (Table 1) are the only requirements to interchange the reaction specificities of GH70 enzymes or that additional mutations more remote from the catalytic center are required currently is unknown. In addition, acceptor subsite +2 has a few different amino acids (Table 1), which might contribute to the 4,6-αGT reaction specificity. For GTFs, it is known that amino acid substitutions at subsite +1 (in position 1140) and subsite +2 (in positions 1137 and 1141) drastically alter the ratio of αglycosidic linkages in the α-glucan polymers made by the enzymes (Shimamura et al. 1994;van Leeuwen et al. 2009). Further mutational analysis of the 4,6-αGT acceptor subsites thus may allow elucidation and modification of their glycosidic linkage specificity. Fig. 5 4,6-αGT-W and 4,6-αGT-ML4 generate α-amylase resistant αglucans. TLC analysis of the product mixtures from maltooligosaccharide (G2-G7) incubations with 4,6-αGT-W (a) and 4,6-αGT-ML4 (b) (see Fig. 2), after treatment with a high-dose of pig pancreatic αamylase. S standard: glucose (G1) to maltoheptaose (G7); Pol polymer. Lanes 2-7 are the product mixtures from G2 to G7, generated by the sequential 4,6-αGT/α-amylase incubations. The upper panel shows as controls isomaltopentaose (Iso5; not degraded by α-amylase) and maltoheptaose (dp7; degraded to maltose and glucose by α-amylase). Note that Iso5 is not entirely pure Another remarkable difference is that 4,6-αGTs synthesize linear α-glucans only, whereas GTFs often produce branched polymers. Since linear α-glucans take up less space than branched ones, 4,6-αGTs may possess a narrower acceptor binding cleft. In this regard, it is noteworthy that the loop connecting domains IV and B (residues 932-943 in GTF-180) is eight residues longer in 4,6-αGTs compared to GTFs (Fig. 1). The crystal structures of GTF-180 and GTF-SI (Vujičić-Žagar et al. 2010;Ito et al. 2011) show that part of this loop delineates the active site; a longer loop at this position thus may result in a more restricted binding cleft in 4,6-αGTs. To further study this hypothesis, 3D structural information is needed, and currently, we are trying to obtain crystal structures of 4,6-αGT proteins.
The reaction specificity of dextran dextrinases comes closest to that of 4,6-αGTs, both transferring single glucose moieties from the non-reducing end of maltooligosaccharides to synthesize isomalto/maltooligosaccharides (Yamamoto et al. Fig. 6 Schematic diagram of the reactions catalyzed by 4,6-αGT enzymes. The open squares indicate the glucose moiety transferred by the enzyme. The sugar binding subsite nomenclature is according to Davies et al. (1997), in which the glycosidic bond is cleaved between donor subsite −1 and acceptor subsite +1 1993a; Kralj et al. 2011). The crucial difference between the two types of enzymes is that dextrin dextrinases form branched carbohydrates (Yamamoto et al. 1993b;Wang et al. 2011;Sims et al. 2001;Tsusaki et al. 2009). A second difference is that dextrin dextrinases disproportionate isomaltooligosaccharides (Yamamoto et al. 1993a), whereas isomaltooligosaccharides are not a substrate of 4,6-αGTs. This is relevant as this means that dextran dextrinases can disproportionate their own reaction products, whereas 4,6-αGTs cannot. A key question that remains to be answered is whether dextran dextrinases are evolutionary related to the GH70 enzymes. Based on their activity they combine the disproportionation activity of 4,6-αGTs and the branching activity of GH70 glucansucrases. The answer to this question requires the identification of the dextran dextrinase (gene) sequence.
A further difference is that GH70 enzymes, including 4,6-αGTs, are more hydrolytic than the GH13 and GH77 transglycosylases. For example, the k cat(transglycosylation) /k cat (hydrolysis) ratio of the GH13 Neisseria polysaccharea amylosucrase is 20 (Albenne et al. 2004), that of GH13 Bacillus circulans CGTase is 90 (van der Veen et al. 2001), GH13 branching enzymes have ratios of >>1,000, and the GH77 4α-glucanotransferases of Thermus thermophilus and T. brockianus have ratios of 5,000 (Kaper et al. 2007;Jung et al. 2011). GH70 GTFs, in contrast, are predominantly hydrolytic in the early phase of sucrose conversion, and they become transglycosylases when better acceptor substrates become available as result of the transglycosylation activity or when good acceptor substrates are added at the start of a reaction (Kralj et al. 2004b). Here, we find that also 4,6-αGT-W has about 50% hydrolytic activity at the start of a reaction with low concentrations of linear (1→4)-α-D-glucooligosaccharides and becomes a better transglycosylase once products appear that again function as acceptor substrate. The acceptor subsites of GH70 enzymes seem thus to be optimized for recognizing and utilizing the products made by the enzyme as acceptor substrates in a polymerization reaction.
Applications and future perspectives of 4,6-αGTs 4,6-αGTs are an exciting type of novel enzymes that convert (1→4)-α-glucan substrates into linear α-glucan products with a high percentage of α1→6 glycosidic linkages (i.e., isomaltooligosaccharides built on maltooligosaccharides). These enzymes are potentially very useful for the synthesis of carbohydrate displaying compounds. Indeed, GH13 and GH70 enzymes are being used for the glycosylation of a variety of compounds (Desmet and Soetaert 2011;Leemhuis et al. 2010;Homann and Seibel 2009). The fact that 4,6-αGTs synthesizes mainly α1→6 glycosidic bonds ensures that the products are water-soluble, which is beneficial for increasing the solubility of poorly water-soluble compounds such as drug molecules.
Because 4,6-αGTs generate α-glucans with a high degree of α1→6 bonds, their products are water-soluble, which is advantageous for applications such as energy and soft drinks. The α1→6 linkages, in addition, make the products resistant to α-amylase digestion and the products as such are expected to pass the human stomach and small intestine and enter the colon were they can serve as carbon source for health promoting bacteria. Thus, 4,6-αGTs are expected to convert readily degradable maltooligosaccharides into a novel type of "resistant" α-glucans and are thus potentially of great interest to the food industry .