Field of the Invention

The present invention relates to the field of poly- and oligosaccharides and their dietary effects. In particular it relates to a method of producing a branched a-glucan. Further aspects of the invention are a branched a-glucan comprising linear segments of (al->4) linked D-glucose units interspersed with (al->6) glucosidic linkages and having (al->4,6) branching points; a food composition; and the use of an a-glucanotransferase enzyme for reducing the digestible carbohydrates of a starch containing food material.

Background of the Invention The prevalence of obesity and being overweight is rapidly increasing worldwide. The development of foods with high satiating capacities and low energy densities may help to prevent weight gain and to stimulate weight loss. Consumption of food and drinks containing non-digestible or slowly digestible carbohydrates instead of sugars induces a lower blood glucose rise after meals compared to sugar- containing food and drinks. The most common carbohydrate in human diets is starch. This polysaccharide is produced by most green plants as an energy store. It is contained in large amounts in such staple foods as potatoes, wheat, maize, rice, and cassava. Various methods have been proposed for the chemical modification of starch and malto-oligosaccharides into non-digestible carbohydrates.

Lactic acid bacteria (LAB) are known to produce diverse extracellular polysaccharides (EPS) with applications in the food and health related industries. Examples are the a-glucans that are synthesized by the action of a single glucansucrase (GS) enzyme from sucrose. WO2001/90372 describes the formation of a branched α-glucan known as "reuteran", regarded as a health promoting food ingredient, synthesized by Lactobacillus reuteri 121 GtfA glucansucrase from sucrose. This enzyme is a member of the glycoside hydrolase family 70 (GH70). It has been observed that highly branched a-glucans can combine a reduced digestibility with a thickening effect triggered by the low pH conditions of the stomach. This thickening leads to feelings of satiety, EP1545562. EP2427565 describes the use of a GH70 glucanotransferase enzyme of L. reuteri 121 GtfB to convert starch into linear gluco-oligosaccharides containing relatively long isomalto-oligosaccharide side chains. The L. reuteri 121 GtfB displays 4,6-a-glucanotransferase (4,6-a-GTase) activity as it cleaves (al->4) linkages and forms new consecutive (al->6) glucosidic linkages. Such materials are partially resistant to digestion and hence give less glucose production on consumption, contributing to the prevention of obesity and type II diabetes.

Co-pending application PCT/EP2016/071474 describes how the GH70 family GtfD enzymes Azotobacter chroococcum NCIMB 8003 and Paenibacillus beijingensis DSM 24997 convert amylose, and starch into a-glucans with alternating (al->4)/(al->6) glucosidic linkages and (al->4,6) branching points, resembling the reuteran polymer synthesized by the L. reuteri 121 GtfA GS from sucrose. Unusually for the starch-converting GH70 family enzymes, both these GtfD enzymes are unable to synthesize consecutive (al->6) glucosidic linkages.

It would be desirable to provide further means for the enzymatic modification of starch, starch derivatives and malto-oligosaccharides in order to change their functional properties and improve their nutritional value. In particular it would be beneficial to provide enzymes to perform such modifications which are suitable for use in food manufacture and exhibit good enzyme activity and thermostability.

Any reference to prior art documents in this specification is not to be considered an admission that such prior art is widely known or forms part of the common general knowledge in the field. As used in this specification, the words "comprises", "comprising", and similar words, are not to be interpreted in an exclusive or exhaustive sense. In other words, they are intended to mean "including, but not limited to".

Summary of the invention

An object of the present invention is to improve the state of the art and to provide an improved solution for the enzymatic modification of starch and other polysaccharide or oligosaccharide into materials having reduced digestibility, or at least to provide a useful alternative. The object of the present invention is achieved by the subject matter of the independent claims. The dependent claims further develop the idea of the present invention.

Accordingly, the present invention provides in a first aspect a method of producing an a-glucan with a ratio of branching of at least 8% comprising contacting a polysaccharide or oligosaccharide substrate comprising at its non-reducing end at least two (al->4) linked D-glucose units with an a-glucanotransferase enzyme capable of cleaving (al->4) glucosidic linkages and making new (al->6) glucosidic linkages without forming consecutive (al->6) glucosidic linkages, to form a glucose polymer having linear segments of (al->4) linked D-glucose units interspersed with (al->6) glucosidic linkages and having (al->4,6) branching points, wherein said α-glucanotransferase (for example a GtfB type of enzyme) comprises an amino acid sequence having at least 70 % identity to SEQ ID NO:l.

In a second aspect, the invention relates to an a-glucan comprising linear segments of (al->4) linked D-glucose units interspersed with (al->6) glucosidic linkages and having (al->4,6) branching points wherein the a-glucan has a ratio of branching of at least 8%; comprises less than 1 wt.% consecutive (al->6) linkages; has an average molecular mass between 1 x 10 ³ Da and 5 x 10 ⁴ Da; and at least 85 wt.% of the α-glucan comprises (al->4) linked D-glucose units having a degree of polymerisation from 2 to 7. A third aspect of the invention relates to a food composition comprising an a-glucan comprising linear segments of (al->4) linked D-glucose units interspersed with (al->6) glucosidic linkages and having (al->4,6) branching points wherein the α-glucan has a ratio of branching of at least 8%; comprises less than l wt.% consecutive (al->6) linkages; has an average molecular mass between 1 x 10 ³ Da and 5 x 10 ⁴ Da; and at least 85 wt.% of the α-glucan comprises (al->4) linked D- glucose units having a degree of polymerisation from 2 to 7.

A further aspect of the invention is the use of an α-glucanotransferase enzyme (for example a GtfB enzyme) that comprises an amino acid sequence having at least 70 % identity to SEQ ID NO:l, or has an amino acid sequence of SEQ ID NO:l, for reducing the digestible carbohydrates of a starch containing food material. Still further aspects of the invention are a bacteria comprising a nucleic acid sequence having at least 95 % identity to SEQ ID NO:l, a bacteria selected from the group consisting of Lactobacillus reuteri strains CNCM 1-2451 (NCC 2603) and CNCM 1-2452 (NCC 2613), an α-glucanotransferase enzyme comprising an amino acid sequence having at least 90% identity to SEQ ID NO:l, and an expression vector comprising a nucleic acid sequence encoding a polypeptide possessing at least 90% sequence identity to SEQ ID NO:l.

The inventors have identified novel GH70 family proteins in the genome of L. reuteri CNCM 1-2451 (NCC 2603) and L. reuteri CNCM 1-2452 (NCC 2613). These enzymes are very similar to each other and are designated GtfB. The GtfB GH70 subfamily mostly comprises 4,6-a-glucanotransferases synthesizing consecutive (al->6) linkages, but surprisingly the activity of these novel enzymes resembles that of the GtfD 4,6-a-glucanotransferases identified in non-lactic acid bacterial strains. Studies of the L. reuteri CNCM 1-2452 GtfB enzyme acting on amylose show that it is unable to synthesize consecutive (al->6) glucosidic bonds, and instead synthesizes a low-molecular-mass reuteran-like polymer consisting of linear (al->4) sequences connected by alternating linear (al->4)/(al->6) linkages and (al->4,6) branching points.

The more open architecture of the L. reuteri CNCM 1-2452 GtfB active site may explain its ability to synthesize branched products, whereas the L. reuteri 121 GtfB 4,6-a-GTase, due to a tunnel extending beyond its active site, only forms linear products. Based on in vitro digestibility studies, branched types of polymers, especially highly branched with relative small size of branches, are less and/or more slowly digested by human gastrointestinal tract enzymes, opening new perspectives for the application of these enzymes for the reduction of glycemic index of starchy products [PCT/EP2016/071474]. L. reuteri bacteria have a long history of safe use in food, providing an advantage for their use by the food industry. The L. reuteri CNCM 1-2452 GtfB, and its homolog encoded by L. reuteri strain NCC 2603 represent new evolutionary intermediates between GH13 and GH70 families. The L. reuteri CNCM I- 2452 GtfB enzyme provides a valuable biocatalyst for the conversion of starch present in food into carbohydrates with attenuated blood glucose release. ID ^Χ Η NM analysis of the branched a-glucan formed by the L reuteri CNCM 1-2452 GtfB enzyme revealed the formation of (al->4) and (al->6) linkages. Methylation analysis of the a-glucan revealed the presence of terminal, 4-substituted, 6-substituted, and 4,6-disubstituted glucopyranose residues. The presence of 6-substituted, and 4,6-disubstituted glucopyranose residues means that the GtfB enzyme forms (al->6) linkages in linear and branched orientations, respectively. No evidence was observed for two consecutive (al->6)-linked glucopyranose residues by 2D NMR spectroscopy analysis. Also, the branched α-glucan synthesized by the L. reuteri CNCM 1-2452 GtfB enzyme was resistant to the endo-(al->6)-hydrolase activity of dextranase, further confirming the absence of consecutive (al->6) linkages in this polysaccharide. Thus, all the branched residues are (al->4,6)-a-D- Glcp-(al->4)- residues. Also, all 6-substituted glucopyranose residues detected by methylation analysis must be (al->4)-linked and are connecting (al->4) glucan chains forming alternating (al->6)/(al->4) linkages in the linear part of the α-glucan structure. This is in contrast to the action of branching enzymes with E.C. 2.4.1.18 activity disclosed in EP1943908. Such branching enzymes only create (al->4,6) branching points but do not create (al->6) linkages in the linear part of the a-glucan structure, and so do not form linear segments of (al->4) linked D-glucose units interspersed with (al->6) glucosidic linkages (sometimes referred to as "alternating" (al->4) and (al->6) glucosidic linkages). Description of the Drawings

Figure 1 is a table showing the sequence alignment of conserved motifs l-IV in the catalytic domain of novel GtfB-like proteins identified in the NCC genome database and other GH70 starch and sucrose acting enzymes: (A) GtfB-like enzymes showing differences in some of the residues in motifs I I and IV forming the substrate-binding site, (B) Characterized GtfB-like enzymes, (C) GtfC-like and GtfD-like 4,6- a-GTase enzymes, (D) sucrose-active GSs enzymes. The seven conserved amino acid residues in GH70 enzymes (indicated by the numbers 1 to 7 above the sequences) are also conserved in the L. reuteri CNCM 1-2452, L reuteri CNCM 1-2451 and L delbrueckii CNCM 1-5166 GtfB proteins identified in the NCC genome database, while six of these seven amino acid residues are conserved for S. thermophilus CNCM 1-5168 and S. thermophilus CNCM 1-5167. Amino acids that constitute the catalytic triad are in bold and slightly shaded. The "hot-spots" 1029, 1065, 1137 and 1140 (L reuteri Gtf 180 GS numbering) are outlined with boxes. Symbols: N U, nucleophile; A/B, general acid/base; TS, transition state stabilizer.

Figure 2 shows a homology model for L. reuteri CNCM 1-2452 (also named L. reuteri NCC 2613) GH70 enzyme. Tertiary structure prediction was accomplished by using the Phyre2 server and the L. reuteri 121 GtfB-ΔΝΔν as template. (A) Overall 3D model structure of L. reuteri CNCM 1-2452 GH70 enzyme. Domains A, B, C and IV are indicated; the proposed catalytic residues in the active site are shown in stick representation (B) Close-up of the active sites regions of the L. reuteri CNCM 1-2452 GtfB enzyme and the L. reuteri 121 GtfB. with loops Al, A2 and B highlighted; the sequence alignment of these loops in the two enzymes is also shown. In L. reuteri CNCM 1-2452 GtfB, the much shorter loops Al and B predict a much more open substrate binding groove than observed in the L. reuteri 121 enzyme. (C) Superposition of the maltopentaose bound in subsites -1 to -5 of the L. reuteri 121 GtfB (PDB: 5J BF) with the L. reuteri CNCM 1-2452 GtfB model. Residues near the binding groove are indicated.

Figure 3 shows a TLC analysis of the products produced by 40 μg ml ^"1 of the L. reuteri CNCM 1-2452 GtfB-ΔΝ (A) and L. reuteri 121 GtfB (B) 4,6-a-glucanotransferase enzymes from incubations with 25 mM malto-oligosaccharides (DP2-DP7), 0.6 % (w v ^"1 ) amylose V, 0.6 % (w v ^"1 ) amylopectin, and 0.6 % (w v ^"1 ) potato soluble starch. The reaction mixtures were incubated at 37°C and pH 5.5 (L reuteri CNCM 1-2452 GtfB) or pH 5.0 (L reuteri 121 GtfB) during 24 h. S, standard; Gl, glucose; G2, maltose; G3, maltotriose; G4, maltotetraose; G5, maltopentaose; G6, maltohexaose; G7, maltoheptaose; AMV, amylose V; AM P, amylopectin; STR, potato soluble starch; Pol, polymer. Figure 4 shows the characterization of product mixtures formed by the incubation of 0.6% (w v ^"1 ) amylose V with 40 μg ml ¹ of L. reuteri CNCM 1-2452 GtfB-ΔΝ, L. reuteri 121 Gtf B , and P. beijingensis Gtf D for 24 h at 37 °C and pH 5.5 (L reuteri CNCM 1-2452 Gtf B), pH 5.0 (L reuteri 121 Gtf B) and pH 7.0

1

(P. beijingensis GtfD). (A) H NM spectrum (D2O, 298K) of the generated products. The anomeric signals indicated as Ga/β and Ra/β correspond to free glucose and reducing -(l->4)-D-Glcp units, respectively. Chemical shifts are shown in parts per million (ppm) relative to the signal of internal acetone (δ 2.225). (B) HPSEC molecular mass distribution of the reaction products formed.

Figure 5 shows 500-MHz ID ^X H NMR spectrum, 2D TOCSY spectra (mixing time 150 ms), and 2D "C^H HSQC spectrum (D ₂ 0, 298K) of the a-glucan generated by the L reuteri CNCM 1-2452 GtfB-ΔΝ enzyme, isolated by size-exclusion chromatography on a Biogel P2 column. The reaction products were obtained from 0.6% (w v ^"1 ) amylose V, incubated with 40 μg ml ^"1 of the L. reuteri CNCM 1-2452 GtfB- ΔΝ enzyme for 24 h at 37°C and pH 5.5. Peaks for (al->4) and (al->6) anomeric signals have been indicated. Structural reporter peaks a: H-4 for 6-substituted Glcp, b: H-4 for terminal Glcp, c: for H-4 for 4-substituted Glcp, d: H-6a for 6-substituted Glcp and e: H-6b for 6-substituted Glcp. Figure 6 shows an HPAEC-PAD profile of the oligosaccharide mixture formed upon the incubation of L. reuteri CNCM 1-2452 GtfB-ΔΝ (20 μg ml - ¹ ) with maltoheptaose for t = 1 h, 3 h and 24 h (pH 5.5, 37 °C). The identity of peaks was assigned using commercial oligosaccharide standards. Gl, glucose; G2-G7, maltose to maltoheptaose; iso-G2, isomaltose; iso-G3, isomaltotriose; Pa, panose.

Figure 7 shows HPAEC-PAD profiles of the oligosaccharide mixtures generated by incubating the L. reuteri CNCM 1-2452 GtfB-ΔΝ enzyme (40 μg ml ^"1 ) (A) and L. reuteri 121 GtfB enzyme (40 μg ml ^"1 ) (B) with 0.35% amylose V (AMV) (donor substrate) or amylose V with 25 mM maltose or 25 mM isomaltose (acceptor substrates) for 24 h at 37°C. The identity of peaks was assigned using commercial oligosaccharide standards. Gl, glucose; G2-G4, maltose to maltotetraose; iso-G2-iso-G5, isomaltose to isomaltopentaose; Pa, panose. Figure 8 shows thin-layer chromatography analysis of the L. reuteri CNCM 1-2452 GtfB-ΔΝ polysaccharide (5 mg ml ^"1 ) after digestion with excess amounts of (A) Aspergillus oryzae a-amylase, (B) Chaetomium erraticum dextranase and (C) Klebsiella planticola pullulanase Ml for 48 h at 37°C. For comparison, the reuteran-like polymers produced by A. chroococcum NCIMB 8003 GtfD and P. bejingensis GtfD, and the IMMP product (~ 90% (al->6) linkages) synthesized by L reuteri 121 GtfB were subjected to the same enzymatic treatments. Lanes 1-5: reaction products generated by the enzymatic treatment of the L. reuteri CNCM 1-2452 GtfB-ΔΝ polymer, A. chroococcum GtfD polymer, P. beijingensis GtfD HMM polymer, P. beijingensis GtfD LMM polymer, and L. reuteri 121 GtfB polymer, respectively. Lane 6, positive controls for the a-amylase, dextranase and pullulanase digestions: amylose (A), dextran (B) and pullulan (C). Lane S, standard: glucose (Gl) to maltoheptaose (G7); Pol, polymer. Figure 9 shows HPAEC-PAD profiles of the oligosaccharides formed by digesting the L. reuteri CNCM I- 2452 GtfB polymer (A), P. beijingensis GtfD LMM polymer (B), P. beijingensis GtfD HMM polymer (C), and A. chroococcum GtfD polymer (D) using pullulanase Ml. Established oligosaccharide structures are included. The identity of peaks 1-16 was assigned using commercial oligosaccharide standards and by comparison with the profile of the pullulanase hydrolysate of reuteran [S.S. van Leeuwen et al., Carbohydr. Res. 343 (2008) 1251-1265.]

Figure 10 is a visual representation of composite structures for the L. reuteri CNCM 1-2452 (also named L. reuteri NCC 2613) GtfB-ΔΝ LMM polymer, the A chroococcum NCIMB 8003 GtfD HMM polymer, and the HMM and LMM P. beijingensis GtfD polymers [PCT/EP2016/071474] formed from amylose V. The composite structures contain all structural features established for the respective products. Quantities of each structural element fit with the combined data of ID ^X H NMR integration and methylation analysis, as well as enzymatic degradation studies with pullulanase.

Figure 11 shows a plot of in-vitro digestion results for wheat flour modified with different quantities of L. reuteri CNCM 1-2452 GtfB enzyme. Percentage glucose release is plotted against time (minutes).

Detailed Description of the invention Consequently the present invention relates in part to a method of producing an a-glucan with a ratio of branching of at least 8% comprising contacting a polysaccharide or oligosaccharide substrate comprising at its non-reducing end at least two (al->4) linked D-glucose units with an a-glucanotransferase enzyme capable of cleaving (al->4) glucosidic linkages and making new (al->6) glucosidic linkages without forming consecutive (al->6) glucosidic linkages, to form a glucose polymer having linear segments of (al->4) linked D-glucose units interspersed with (al->6) glucosidic linkages and having (al->4,6) branching points, wherein said α-glucanotransferase comprises (for example consists of) an amino acid sequence having at least 70 % identity to SEQ ID NO:l (for example at least 75, 80, 85, 90, 95, 96, 97, 98, or 99% identity to SEQ ID NO:l). The α-glucanotransferase enzyme in the method of the invention may be capable of cleaving (al->4) glucosidic linkages and transferring malto- oligosaccharides up to DP7 (for example up to DP5). The α-glucanotransferase enzyme in the method of the invention may be a GtfB type of enzyme. The α-glucanotransferase enzyme in the method of the invention may be a GtfB enzyme from a bacterium selected from the group consisting of L. reuteri CNCM 1-2451, L. reuteri CNCM 1-2452, Streptococcus thermophilus CNCM 1-5167, S. thermophilus CNCM 1-5168, Lactobacillus delbrueckii sbsp. delbrueckii CNCM 1-5166.

SEQ ID NO:l is the sequence of the L. reuteri CNCM 1-2452 GtfB enzyme. SEQ ID NO:4 is the sequence of the Streptococcus thermophilus CNCM 1-5168 GtfB enzyme (which is identical to the sequence of the Streptococcus thermophilus CNCM 1-5167 GtfB enzyme). SEQ ID NO:5 is the sequence of the Lactobacillus delbrueckii sbsp. delbrueckii CNCM 1-5166 enzyme. SEQ ID NO:19 is the sequence of the L. reuteri CNCM 1-2451 GtfB enzyme.

L. reuteri CNCM 1-2452, also named NCC 2613, was deposited with the Collection Nationale de Cultures de Microorganismes (CNCM), Institut Pasteur, 25 rue du Docteur oux, F-75724 PARIS Cedex 15, France, on 19 ^th April 2000 and given the deposit number 1-2452.

L. reuteri CNCM 1-2451, also named NCC 2603, was deposited with the Collection Nationale de Cultures de Microorganismes (CNCM), Institut Pasteur, 25 rue du Docteur Roux, F-75724 PARIS Cedex 15, France, on 19 ^th April 2000 and given the deposit number 1-2451. L. delbrueckii sbsp. delbrueckii CNCM 1-5166, also named NCC 828, was deposited with the Collection Nationale de Cultures de Microorganismes (CNCM), Institut Pasteur, 25 rue du Docteur Roux, F-75724 PARIS Cedex 15, France, on 14 ^th February 2017 and given the deposit number 1-5166.

S. thermophilus CNCM 1-5167, also named NCC 903, was deposited with the Collection Nationale de Cultures de Microorganismes (CNCM), Institut Pasteur, 25 rue du Docteur Roux, F-75724 PARIS Cedex 15, France, on 14 ^th February 2017 and given the deposit number 1-5167.

S. thermophilus CNCM 1-5168, also named NCC 2408, was deposited with the Collection Nationale de Cultures de Microorganismes (CNCM), Institut Pasteur, 25 rue du Docteur Roux, F-75724 PARIS Cedex 15, France, on 14 ^th February 2017 and given the deposit number 1-5168.

Lactobacillus fermentum CNCM 1-5068, also named NCC 2970 was deposited with the Collection Nationale de Cultures de Microorganismes (CNCM), Institut Pasteur, 25 rue du Docteur Roux, F-75724 PARIS Cedex 15, France, on 8th March 2016 and given the deposit number 1-5068. Polysaccharides are polymeric carbohydrate molecules composed of long chains of monosaccharide units bound together by glycosidic linkages. Oligosaccharides are saccharide polymers containing a small number (typically three to nine) of monosaccharides. An example of a substrate comprising at its non-reducing end at least two (al->4) linked D-glucose units is amylose. In the present specification, the abbreviation Gtf refers to glucanotransferase. Single (al->6) glucosidic linkages between one or more (al->4) glucosidic linkages as may be formed in the method of the invention are sometimes referred to as "bridging" (al->6) linkages. The notation (al->4) may be used instead of a(l->4) to refer to a l->4 a linkage, but these are equivalent, as are (al->6) and a(l->6).

In the context of the present invention, the ratio of branching is defined as the total number of branching anhydroglucose units (AGU), i.e. AGU being bound to three other units, with respect to the total number of AGU of a molecule. The ratio of branching can be determined by methods known in the art, such as methylation with gas chromatography. The α-glucan produced by the method of the invention may have a ratio of branching of at least 8%, for example at least 10%, for example at least 15%. An embodiment of the present invention provides a method of producing an α-glucan with a ratio of branching of at least 8% comprising contacting a polysaccharide or oligosaccharide substrate comprising at its non-reducing end at least two (al->4) linked D-glucose units with a L. reuteri GtfB enzyme comprising (for example consisting of) an amino acid sequence having at least 90% identity to SEQ ID NO:l (for example at least 95, 96, 97, 98, or 99% identity to SEQ ID NO:l). GH70 enzymes active on starch possess a Tyr residue, replacing the 1065 (L reuteri 180 Gtfl80 numbering) residue of motif III which is well-conserved in GSs.

The GtfB protein sequences of L. reuteri CNCM 1-2451, L. reuteri CNCM 1-2452, S. thermophilus CNCM 1-5167, S. thermophilus CNCM 1-5168 and L. delbrueckii sbsp. delbrueckii CNCM 1-5166 show differences in some of the residues in motifs II and IV forming the substrate-binding site. Similarly to GtfC and GtfD enzymes, the subsite +1 Asn residue (N1029 in L. reuteri Gtfl80 GS) is replaced by His in these five GtfB proteins. The correspondence between the L. reuteri Gtfl80 GS numbering and numbering in other enzyme sequences for the residues in motifs I to IV is shown in Figure 1. For example, residue 1029 according to L. reuteri Gtfl80 GS numbering is residue 683 in the GtfB of L. reuteri CNCM 1-2451, residue 646 in the GtfB of L. reuteri CNCM 1-2452, residue 1039 in the GtfB of S. thermophilus CNCM I- 5167, residue 1039 in the GtfB of S. thermophilus CNCM 1-5168 and residue 678 in the GtfB of L. delbrueckii sbsp. delbrueckii CNCM 1-5166. For the GtfB proteins of L. reuteri CNCM 1-2451 and L. reuteri CNCM 1-2452 the amino acids at positions 1137 and 1140 following the putative transition state stabilizer (Gtfl80 L. reuteri 180 numbering), are Ser and Ala, instead of the Gin and Lys residues typically found in most GtfB-and GtfC-like 4,6-a-GTases. For the GtfB proteins of L. delbrueckii sbsp. delbrueckii CNCM 1-5166 the amino acid at position 1140 following the putative transition state stabilizer (Gtfl80 L. reuteri 180 numbering) is also Ala.

An embodiment of the present invention provides a method of producing an a-glucan with a ratio of branching of at least 8% comprising contacting a polysaccharide or oligosaccharide substrate comprising at its non-reducing end at least two (al->4) linked D-glucose units with a GtfB enzyme comprising an amino acid sequence with histidine at residue 1029 and/or serine at residue 1137 and/or alanine at reside 1140, following Gtfl80 Lactobacillus reuteri 180 numbering. The GtfB enzyme according to the method of the invention may comprise an amino acid sequence with a tyrosine residue at position 1065 and a histidine residue at position 1029 (Gtfl80 L. reuteri 180 numbering). The GtfB enzyme according to the method of the invention may comprise an amino acid sequence with a tyrosine residue at position 1065, a histidine residue at position 1029 and an alanine residue at position 1140 (Gtfl80 L. reuteri 180 numbering). The GtfB enzyme according to the method of the invention may comprise an amino acid sequence with a tyrosine residue at position 1065, a histidine residue at position 1029 and/or a serine residue at position 1137 and/or an alanine residue at position 1140, following Gtfl80 Lactobacillus reuteri 180 numbering. The invention may provide a method of producing an α-glucan with a ratio of branching of at least 8% comprising contacting a polysaccharide or oligosaccharide substrate comprising at its non-reducing end at least two (al->4) linked D-glucose units with an a-glucanotransferase enzyme capable of cleaving (al->4) glucosidic linkages and making new (al->6) glucosidic linkages without forming consecutive (al->6) glucosidic linkages, to form a glucose polymer having linear segments of (al->4) linked D-glucose units interspersed with (al->6) glucosidic linkages and having (al->4,6) branching points, wherein said a-glucanotransferase is a GtfB type of enzyme comprising an amino acid sequence with a histidine residue at position 1029 and/or a serine residue at position 1137 and/or an alanine reside at position 1140, following Gtfl80 Lactobacillus reuteri 180 numbering.

The substrate in the method of the invention may have a degree of polymerization of at least four, for example it may comprise at least four D-glucose units. The degree of polymerization is the number of monomeric units in a polymer or oligomer molecule. For example, the substrate in the method of the invention may have a degree of polymerization of at least five, for example it may comprise at least five D-glucose units. The substrate in the method of the invention may be selected from the group consisting of starch (for example waxy starch or high amylose starch), starch derivatives, malto- oligosaccharides, amylose, amylopectin, maltodextrins, (al->4) glucans and combinations thereof. Starch derivatives are prepared by physically, enzymatically, or chemically treating native starch to change its properties. The substrate in the method of the invention may be comprised within another material, for example the substrate may be starch provided in the form of flour. It is advantageous to be able to convert polysaccharides or oligosaccharides comprised within food ingredients into a-glucans with lower digestibility, for example branched a-glucans. Such a conversion may increase the fibre content of the ingredients and/or may aid in reducing the calorie content of the ingredients. The method of the invention may be performed as part of a food processing operation, for example the a-glucanotransferase enzyme may be applied to food ingredients during a process to produce a food product. The substrate may be comprised within a material which already has a positive nutritional profile, for example the substrate may be comprised within wholegrain flour.

The extent to which the polysaccharide or oligosaccharide substrate may be converted by the α-glucanotransferase enzyme in the method of the invention can be adjusted by limiting the time of reaction. Partially converted substrates will provide different physical properties. The production of a- glucan in the method of the invention may be stopped before the reaction between the substrate and the α-glucanotransferase enzyme has reached completion, for example it may be stopped by denaturing (e.g. by heat) or removing the enzyme. The α-glucanotransferase enzyme in the method of the invention may be immobilized, for example immobilized before contacting the polysaccharide or oligosaccharide substrate. Such immobilization techniques are well known in the art. Removal of the enzyme (discussed above) may be facilitated by immobilization of the enzyme. Immobilization techniques may be selected from the group consisting of covalent binding, entrapment, physical adsorption, cross-linking and combinations of these. In immobilization by covalent binding, enzymes are covalently linked to a support through the functional groups in the enzymes that are not essential for the catalytic activity. Oxide materials such as alumina, silica, and silicated alumina can be used for covalent binding of the enzyme. In immobilization by entrapment the enzyme is localized within the lattice of a polymer matrix or membrane. Entrapment methods are classified into five major types: lattice, microcapsule, liposome, membrane, and reverse micelle. The enzyme is entrapped in the matrix of various synthetic or natural polymers. Alginate, a naturally occurring polysaccharide that forms gels by ionotropic gelation is one such immobilization matrix. Immobilization by physical adsorption is the simplest and the oldest method of immobilizing enzymes onto carriers. Immobilization by adsorption is based on the physical interactions between the enzymes and the carrier, such as hydrogen bonding, hydrophobic interactions, van der Waals force, and their combinations. Adsorption is generally less disruptive to the enzymes than chemical means of attachment. Immobilization by cross-linking utilizes bi- or multifunctional compounds, which serve as the reagent for intermolecular cross-linking of the enzymes. Cross-linking may be used in combination with other immobilization methods such as adsorption or entrapment.

The polysaccharide or oligosaccharide substrate may be contacted with an a-glucanotransferase enzyme in the method of the invention at a temperature of between 10 °C and 75 °C (for example between 20 °C and 70 °C, for example between 30 °C and 65 °C, for example between 35 °C and 45 °C) and a pH of between 4.0 and 9.0 (for example between 4.8 and 8.0, for example between 5.0 and 6.0). The L. reuteri CNCM 1-2452 GtfB enzyme is active at high pH values which is useful for applications in alkali environments.

In a further embodiment the present invention pertains to an α-glucan comprising linear segments of (al->4) linked D-glucose units interspersed with (al->6) glucosidic linkages and having (al->4,6) branching points wherein the α-glucan has a ratio of branching of at least 8% (for example at least 12%, for further example at least 15%); comprises less than 1 wt.% consecutive (al->6) linkages; has an average molecular mass between 1 x 10 ³ Da and 5 x 10 ⁴ Da (for example, an average molecular mass between 2 x 10 ³ Da and 2 x 10 ⁴ Da, for example, an average molecular mass between 5 x 10 ³ Da and 1 x 10 ⁴ Da); and at least 85 wt.% (for example at least 90 wt.%, for further example at least 95 wt.%) of the α-glucan comprises (al->4) linked D-glucose units having a degree of polymerisation from 2 to 7. The percentage of the α-glucan comprising (al->4) linked D-glucose units having a degree of polymerisation from 2 to 7 may for example be measured by digestion of the α-glucan with pullulanase and evaluating the resulting mixture with TLC and/or HPAEC. The α-glucan according to the invention may comprise between 55 and 65 percent consecutive (al->4) glucosidic linkages, between 8 and 15 percent single (al->6) glucosidic linkages interspersed between linear (al->4) linked D-glucose units and between 10 and 20 percent (al->4,6) branching points, for example between 14 and 18 percent (al->4,6) branching points. The α-glucan according to the invention may have less than 1 % consecutive (al->6) glucosidic linkages, for example it may have less than 0.5 % consecutive (al->6) glucosidic linkages, for further example it may have no consecutive (al->6) glucosidic linkages. The α-glucan of the invention is similar to the low molecular mass a-glucan synthesized by the P. beijingensis GtfD from starch (co-pending application PCT/EP2016/071474), but has almost no (al->4) linked D-glucose units having a degree of polymerisation greater than 7. This is beneficial as an increase in shorter chain fractions has been linked to a reduced digestion rate in starches [Xingfeng Li et al., Food Chemistry, 164, 502-509 (2014)]. In a further aspect, the invention provides an a-glucan obtainable (for example obtained) by contacting a polysaccharide or oligosaccharide substrate comprising at its non-reducing end at least two (al->4) linked D-glucose units with an a-glucanotransferase enzyme comprising an amino acid sequence having at least 90% identity to SEQ ID NO:l, for example the L. reuteri CNCM 1-2451 GtfB enzyme or the L. reuteri CNCM 1-2452 GtfB enzyme.

The α-glucan of the invention can be regarded as a dietary fiber. Due to its highly branched structure, the α-glucan will resist enzymatic degradation in the upper gastrointestinal tract and end up in the large intestine where it can be fully fermented by the colonic microflora. In addition, such dietary fibres enhance satiety in humans or animals. Blood sugar levels rise after a meal. As the a-glucans of the invention display reduced digestibility compared to materials such as starch, meals prepared containing them will cause a reduced blood glucose response compared to the equivalent meal with starch, and will provoke a lower insulin response. A composition comprising the α-glucan of the invention may be for use in the control of postprandial blood glucose and insulin levels in a subject. The subject may be a human or a pet. A composition comprising the α-glucan of the invention may be for use in the treatment or prevention of a disorder linked to an increase in postprandial blood glucose and insulin levels in a subject. The disorder may be selected from the group consisting of diabetes, for example gestational diabetes; impairment of glucose metabolism; hyperinsulinemia or insulin resistance. The subject may be a diabetic or pre-diabetic human or pet.

Typically, postprandial hyper-insulinemia may promote the development of insulin resistance, metabolic syndrome, glucose intolerance and type-2 diabetes [Kopp W., Metabolism. 2003, Jul; 52(7):840-844]. Lowering the insulin demand after a meal however, can reduce on one hand the deterioration of the glycemic control in type-2 diabetes and on the other hand reduce the risk of developing type-2 diabetes in predisposed subjects.

A "pre-diabetic patient" is a subject showing insulin resistance or impaired glucose metabolism and is predisposed, for example by family history, lifestyle or genetics, for developing diabetes later in life. Reducing insulin secretion reduces the risk of the pancreas becoming exhausted in the long term, and so is beneficial for management of the pancreas in pre-diabetes or patients with metabolic disorders. The use of a composition comprising the a-glucan of the invention would consequently reduce the risk and/or the development of diabetes, impaired glucose metabolism, hyperinsulinemia or insulin resistance in those subjects.

Prevalence of diabetes, insulin resistance or glucose intolerance is mostly observed in adult humans. However, more and more children are affected, or predisposed or at risk of developing such a disorder later in life. Hence, advantageously, prevention and/or treatment of those disorders is started already in young age. Alternatively, and similarly as observed with humans; diabetes, hyperinsulinemia or insulin resistance are more and more widespread among animals, particularly with animals kept as pet animals. Hence, the invention also pertains to cats and dogs. A composition comprising the α-glucan of the invention may be for non-therapeutic use to decrease plasma postprandial glucose and insulin levels. It is advantageous that a composition comprising the α-glucan of the invention can also be administered to subjects, for example healthy subjects, which may be at risk of developing diabetes type-2, insulin resistance or glucose intolerance at some later time. A composition comprising the α-glucan of the invention, as disclosed herein, provides a reduced insulin level after consumption. Many healthy people desire to lose weight. Consuming meals which contain dietary fibre can increase satiety and therefore help people consume fewer digestible calories. A composition comprising the α-glucan of the invention may be for non-therapeutic use to lose weight.

Another aspect of the invention relates to a food composition comprising the α-glucan of the invention. The food composition may for example comprise between 1 and 20 wt.% of the α-glucan of the invention. The food composition may be a beverage, for example a powdered beverage mix or a beverage creamer; a potato product, for example instant mashed potato; a breakfast cereal, for example extruded cereal or porridge; a pet food product; a baked dough product, for example a bread, a pizza or a filled savoury turnover; or a confectionery product. The confectionery product may be a frozen confectionery product such as an ice-cream; a baked confectionery product such as a biscuit, for example a filled biscuit or wafer; a chocolate confectionery product; or a sugar-style confectionery product such as a gum, a jelly, a hard-boiled sweet or a chewy sweet. The term "sugar-style confectionery product" or "sugar-style candy" refers to confectionery products which would traditionally have been based on sugar, but may be manufactured with alternative sweeteners and/or sugar substitutes. In a further embodiment, the invention provides for the use of an a-glucanotransferase enzyme for reducing the digestible carbohydrates of a food material, for example a starch-based food material, wherein the α-glucanotransferase enzyme comprises (for example consists of) an amino acid sequence having at least 85% identity to SEQ ID NO:l (for example at least 90, 95, 96, 97, 98, or 99% identity to SEQ ID NO:l), or has an amino acid sequence of SEQ ID NO:l. In the scope of the current invention, digestible carbohydrates correspond to the fraction of the total carbohydrates that is digestible and available to provide energy to body cells.

The invention further provides for the use of a GtfB α-glucanotransferase enzyme for reducing the digestible carbohydrates of a food material, for example a starch-based food material, wherein the α-glucanotransferase GtfB enzyme comprises an amino acid sequence with histidine at residue 1029 and/or serine at residue 1137 and/or alanine at reside 1140 following Gtfl80 Lactobacillus reuteri 180 numbering. The invention further provides for the use of a GtfB α-glucanotransferase enzyme for reducing the digestible carbohydrates of a food material, for example a starch-based food material, wherein the α-glucanotransferase GtfB enzyme is from a bacterium selected from the group consisting of L reuteri CNCM 1-2451, L reuteri CNCM 1-2452, S. thermophilus CNCM 1-5167, S. thermophilus CNCM 1-5168 and L. delbrueckii sbsp. delbrueckii CNCM 1-5166, for example L. reuteri CNCM 1-2452. In an embodiment, the invention provides for the use of an α-glucanotransferase enzyme for reducing the glycemic index of a food material, for example a starch-based food material, wherein the α-glucanotransferase enzyme comprises (for example consists of) an amino acid sequence having at least 85% identity to SEQ ID NO:l (for example at least 90, 95, 96, 97, 98, or 99% identity to SEQ ID NO:l), or has an amino acid sequence of SEQ ID NO:l. The glycemic index is a number associated with a particular type of food that indicates the food's effect on a person's blood glucose (also called blood sugar) level. A value of 100 represents the standard, an equivalent amount of pure glucose.

The invention further provides for the use of a GtfB α-glucanotransferase enzyme for reducing the glycemic index of a food material, for example a starch-based food material, wherein the α-glucanotransferase GtfB enzyme comprises an amino acid sequence with histidine at residue 1029 and/or serine at residue 1137 and/or alanine at reside 1140 following Gtfl80 Lactobacillus reuteri 180 numbering.

One aspect of the invention provides a bacteria comprising a nucleic acid sequence having at least 95 % identity to SEQ ID NO:l (for example at least 96, 97, 98, or 99% identity to SEQ ID NO:l), for example a Lactobacillus reuteri bacteria. In another aspect, the invention provides a bacteria comprising a nucleic acid sequence having at least 95 % identity to SEQ ID NO:4 (for example at least 96, 97, 98, or 99% identity to SEQ ID NO:4) for example S. thermophilus CNCM 1-5167 or S. thermophilus CNCM 1-5168 bacteria. In another aspect, the invention provides a bacteria comprising a nucleic acid sequence having at least 95 % identity to SEQ ID NO:5 (for example at least 96, 97, 98, or 99% identity to SEQ ID NO:5), for example a Lactobacillus delbrueckii subsp. delbrueckii bacteria CNCM 1-5166. An aspect of the invention is a bacteria selected from the group consisting of L. reuteri CNCM 1-2451, L. reuteri CNCM 1-2452, S. thermophilus CNCM 1-5167, S. thermophilus CNCM 1-5168 and L delbrueckii sbsp. delbrueckii CNCM 1-5166. For example a bacteria selected from the group consisting of S. thermophilus CNCM 1-5167, S. thermophilus CNCM 1-5168 and L. delbrueckii sbsp. delbrueckii CNCM I- 5166.

A further aspect of the invention is an a-glucanotransferase enzyme comprising (for example comprising) an amino acid sequence having at least 95 % identity to SEQ ID NO:l (for example at least 96, 97, 98, or 99% identity to SEQ ID NO:l). A further aspect of the invention is an a-glucanotransferase enzyme comprising (for example comprising) an amino acid sequence having at least 95 % identity to SEQ ID NO:4 (for example at least 96, 97, 98, or 99% identity to SEQ ID NO:4). A further aspect of the invention is an α-glucanotransferase enzyme comprising (for example comprising) an amino acid sequence having at least 95 % identity to SEQ ID NO:5 (for example at least 96, 97, 98, or 99% identity to SEQ ID NO:5). The α-glucanotransferase enzyme may be for example a GtfB enzyme. A still further aspect of the invention is an expression vector comprising a nucleic acid sequence encoding a polypeptide possessing at least 95 % sequence identity to SEQ ID NO:l (for example at least 96, 97, 98, or 99% identity to SEQ ID NO:l). Another aspect of the invention is an expression vector comprising a nucleic acid sequence encoding a polypeptide possessing at least 95 % sequence identity to SEQ ID NO:4 (for example at least 96, 97, 98, or 99% identity to SEQ ID NO:4). Another aspect of the invention is an expression vector comprising a nucleic acid sequence encoding a polypeptide possessing at least 95 % sequence identity to SEQ ID NO:5 (for example at least 96, 97, 98, or 99% identity to SEQ ID NO:5).

EXPERIMENTAL

Materials and methods

Annotation of the GH70 family enzymes present in the NCC genome database was performed using the dbCAN database for automated Carbohydrate-active enzyme Annotation [Y. Yin et al., dbCAN: a web resource for automated carbohydrate-active enzyme annotation Nucleic Acids Res. 40 (2012) W445-51.] Hits having an E-Value below 1E-5 and a bit score above 350 were considered. As a result 788 protein sequences were retrieved and used together with the L. reuteri 121 GtfB (Accession number: AAU08014.2), Leuconostoc citreum NRRL B-1299 branching sucrase (Accession number: CDX66820.1) and L. reuteri 180 Gtfl80 GS (accession number: AAU08001.1) protein sequences for the construction of multiple sequence alignments with Jalview 2 desktop application using the MUSCLE algorithm [A.M. Waterhouse et al., Jalview Version 2-a multiple sequence alignment editor and analysis workbench, Bioinformatics 25 (2009) 1189-1191.] Sequences were only considered to be putative starch-acting GH70 enzymes if they possessed an aromatic Tyr (Y1055 L. reuteri 121 GtfB numbering) replacing the conserved Trp typically present in GSs, resulting in a set of 106 GtfB-like gene products. Branching sucrases were distinguished by the presence of a Gly residue at this position in the alignments. For further analysis, the set of GtfB proteins identified within the NCC genome database was expanded with characterized GH70 proteins indexed in CAZy (http://www.cazy.org/) and aligned by MUSCLE, using default parameters. Phylogenetic relationships were determined by the Maximum Likelihood method based on the JTT matrix model using MEGA6 [K. Tamura, G. Stecher, D. Peterson, A. Filipski, S. Kumar, MEGA6: Molecular Evolutionary Genetics Analysis version 6.0, Mol. Biol. Evol. 30 (2013) 2725-2729.] The analysis involved 167 amino acid sequences. Partial deletion of the positions containing alignment gaps and missing data was conducted. Statistical confidence of the inferred phylogenetic relationships was assessed by performing 1,000 bootstrap replicates.

Analysis of the L reuteri CNCM 1-2452 GtfB protein sequence Multiple amino acid sequence alignments were generated with Clustal Omega (http://www.ebi.ac.uk/Tools/msa/clustalo/) and visualized by using the Jalview 2 desktop application. Subcellular localization of the L. reuteri GtfB protein was predicted using CELLO v.2.5: subCELIular Localization predictor (http://cello.life.nctu.edu.tw/) and its theoretical M _w (molecular weight) was predicted by ExPASy Compute pl//W _w (http://web.expasy.org/compute_pi/). Structural modelling of the L. reuteri CNCM 1-2452 GtfB protein

A three-dimensional model of the L. reuteri CNCM 1-2452 GtfB was constructed with Phyre [Kelley et al., Nat. Protoc. 10 (2015) 845-858] using the recently determined three-dimensional structure of L. reuteri 121 GtfB 4,6-a-GTase (PDB entry: 5JBD); [Bai et al., Structure 25 (2016) 231-242] as a template for one-to-one threading of the full-length sequence, with default settings. For comparison of binding sites, also the crystal structures of L. reuteri 121 GtfB 4,6- a-GTase complexed with maltopentaose or isomalto-maltopentasaccharide (PDB entries: 5JBE, 5JBF) were used.

Cloning of the L. reuteri gtfB gene

The gtfB gene fragment encoding for an N-terminally truncated variant of the GtfB protein (GtfB-ΔΝ) was amplified from L. reuteri CNCM 1-2452 genomic DNA with Phusion DNA polymerase (Finnzyme, Helsinki, Finland) and cloned into a modified pET15b vector by ligation-independent cloning (LIC) [D. Bonsor et al., Org. Biomol. Chem. 4 (2006) 1252-1260]. The primers used contained LIC-compatible extensions (underlined), and were: Forward CAGGGACCCGGTGGG C ATTTACTTG G A A ATC and Reverse CGAGGAGAAGCCCGGTTAATCGTCTTCAATATTAGC. The Kpnl-digested vector and the generated PCR product were purified from gel, and subsequently treated with T4 DNA polymerase in the presence of dATP and dTTP, respectively. The two reaction products were mixed together in a 1:4 molar ratio, and the mixture was used to transform chemical-competent Escherichia coli DH5a cells (Phabagen), yielding pETlSb/gi/B- l/V. This vector encodes the GtfB-ΔΝ (amino acids 417 to 1281) fused with an N- terminal His6-tag cleavable by a 3C protease. The constructed expression vector pET15b/gf/B--l/\/ was verified by nucleotide sequencing (GATC, Cologne, Germany), and transformed into E. coli BL21 Star (DE3).

Expression and purification of the L. reuteri CNCM 1-2452 GtfB protein

Fresh Luria Broth medium supplemented with ampicillin (100 μg ml ^"1 ) was inoculated with 1% (v/v ¹ ) of an overnight culture of E. coli BL21 Star (DE3) harboring the pET15b/gf/B--l/\/ plasmid, and cultivated at 37°C and 160 rpm. Protein expression was induced at an Οϋεοο of 0.7 by adding isopropyl- -d-l- thiogalactopyranoside to 0.1 mM, and cultivation was continued for 20 h at 16 °C. Cells were harvested by centrifugation (10,000 ^χ g, 20 min). The GtfB-ΔΝ enzyme was purified by Ni ²⁺ -nitrilotriacetic acid (NTA) affinity chromatography (Sigma Aldrich, St. Louis, USA) as described previously [Gangoiti et al., Biochim Biophys Acta 1860 (2016) 1224-1236]. Purity was assessed by SDS-PAGE analysis, and protein concentrations were determined by measuring the absorbance at 280 nm, using a NanoDrop 2000 spectrophotometer (Isogen Life Science, De Meern, The Netherlands).

Enzyme activity assays

The initial total activity of the L. reuteri CNCM 1-2452 Gtf Β-ΔΝ enzyme was determined by the amylose- iodine staining method using 0.125% (w v ^"1 ) amylose V (AVEBE, Foxhol, The Netherlands) as described before [19, 29]. Routinely, enzymatic assays were performed with 2 μg ml ¹ of enzyme in 25 mM sodium acetate (pH 5.5) and 1 mM CaC . The decrease in absorbance of the a-glucan-iodine complex resulting from transglycosylation and/or hydrolytic activity was monitored at 660 nm for 8 min at 40 °C. One unit of activity was defined as the amount of enzyme converting 1 mg of substrate per min. The pH profile and optimum pH were determined at 40 °C by varying the pH between 3.0 and 10.0. Sodium citrate buffer (25 mM) was used at pH 3.0-7.0, sodium phosphate buffer (25 mM) at pH 7.0- 8.0, Tris-HCI (25 mM) at pH 8.0-9.0, and sodium bicarbonate (25 mM) at pH 9.0-10.0. Substrate specificity of the L. reuteri CNCM 1-2452 GtfB enzyme

The substrate specificity of the L. reuteri CNCM 1-2452 GtfB enzyme was investigated by incubating 40 μg ml ¹ of purified enzyme with either 25 mM sucrose (Acros), nigerose (Sigma-Aldrich), panose (Sigma- Aldrich), isomaltose (Sigma-Aldrich), isomaltotriose (Sigma-Aldrich), isomaltopentaose (Carbosynth), malto-oligosaccharides (MOS) with degrees of polymerization (DP) 2-7 (Sigma-Aldrich), or with 0.6% (w v ^"1 ) amylose V (AVEBE, Foxhol, The Netherlands), potato starch (Sigma-Aldrich) or amylopectin (Sigma-Aldrich). Potato starch was pregelatinized by autoclaving (15 min, 120°C). Amylose V (1%, w/v) was prepared as a stock solution in sodium hydroxide (1 M). Prior to use, the stock solution was neutralized with 7 M HCI and diluted to a concentration of 0.85% (w v ^"1 ). Incubations were carried out in 25 mM sodium acetate buffer, pH 5.5 with 1 mM CaC at 37°C for 24 h. Reactions were stopped by heating the samples to 100 °C for 8 min. The progress of the reactions was analysed by thin-layer chromatography (TLC) and/or high-performance-anion-exchange chromatography (HPAEC).

Thin Layer Chromatography and High Performance Anion Exchange Chromatography with pulsed amperometric detection analysis

Carbohydrate samples were spotted in 1-cm lines on a TLC silica gel 60F254 sheet (Merck, Darmstadt, Germany). The TLC plate was run for 6 h in butanohacetic acid:water (2:1:1, v v ^"1 ), and products were visualized with orcinol/sulfuric acid staining. A mixture of glucose and malto-oligosaccharides (DP2 to DP7) was used as standard. HPAEC-PAD analysis was performed using an ICS3000 workstation (Thermo Scientific, Amsterdam, The Netherlands), equipped with a CarboPac PA-1 column (Thermo Scientific; 250 x 2 mm) and an ICS3000 electrochemical detection module. Prior to analysis the carbohydrate samples were diluted 1:300 in DMSO and the oligosaccharides were separated at a 0.25 ml min ¹ flow rate by using a sodium acetate gradient (10 to 240 mM) in 100 mM NaOH over 57 min. The injection volume of each sample was 5 μΙ. The identity of the peaks was determined using commercial oligosaccharide standards and a mixture of MOS of DPs from 2 to 30.

HPSEC analysis

Molecular mass distribution of the product mixtures was determined using a size exclusion chromatography system (Agilent Technologies 1260 Infinity) equipped with a multi angle laser light scattering detector (SLD 7000 PSS, Mainz), a viscometer (ETA-2010 PSS, Mainz) and a differential refractive index detector (G1362A 1260 RID Agilent Technologies), as described before [20, 29]. Briefly, samples were dissolved at a concentration of 4 mg ml ¹ in DMSO-LiBr (0.05 M) and separation was carried out by using three PFG-SEC columns with porosities of 100, 300 and 4000 A, coupled with a PFG guard column. The eluent was DMSO-LiBr (0.05 M) at a flow rate of 0.5 ml min ^"1 . The system was calibrated and validated using a standard pullulan kit (PSS, Mainz, Germany) with M _w ranging from 342 to 805 000 Da. The specific Rl increment value (dn/dc) was also measured by PSS and was 0.072 ml g ^"1 (private communication with PSS). The multiangle laser light scattering signal was used to determine the molecular masses of amylose V and the high molecular mass (HMM) polysaccharides generated by the A. chroococcum and P. beijingensis GtfD enzymes. The dn/dc va\ues for these polysaccharides were taken to be the same as for pullulan. The molecular masses of the L. reuteri CNCM 1-2452 GtfB, L. reuteri 121 GtfB and P. beijingensis GtfD low molecular mass (LMM) polymers were determined by universal calibration method. Measurements were performed in duplicate.

Production, isolation and structural analysis of the products from amylose V incubation with L. reuteri CNCM 1-2452 GtfB

Incubations of amylose V (0.6% w v ^"1 ) and GtfB-ΔΝ (0.2 mg) were performed under the conditions described in "Substrate specificity of the L. reuteri CNCM 1-2452 GtfB". After incubation for 24 h at 37°C, the reaction was stopped by transfer to 100 °C for 10 min. The polysaccharide was separated from trace amounts of small oligosaccharides (DP<5) also present in the product mixture by size- exclusion chromatography on a Biogel P2 column (2.5 x 50 cm; Bio-Rad, Veenendaal, The Netherlands) using 10 mM NH ₄ HC0 ₃ as eluent at a flow rate of 48 ml h ¹ .

NMR spectroscopy Resolution-enhanced 1D/2D ¹ H and ¹³ C NMR spectra were recorded in D2O on a Varian lnova-500 spectrometer (NMR center, University of Groningen, The Netherlands) at a probe temperature of 298 K. Samples were exchanged twice in D2O (99.9 at% D, Cambridge Isotope Laboratories, Inc., Andover, MA) with intermediate lyophilization, and then dissolved in 0.6 ml of D2O. One-dimensional 500-MHz ¹ H NMR spectra were recorded at a 4000 Hz spectral width and 16k complex points, using a WET1D pulse to suppress the HOD signal. Two-dimensional H spectra (COSY, TOCSY MLEV17 30, 50, and 150 ms, and ROESY 300 ms) were recorded with 4000 Hz spectral width, collecting 200 increments. In case of TOCSY spectra 2000 complex data points were collected, for COSY and ROESY spectra 4000 complex data points were used. 2D ^C^H NMR spectra were recorded in 128 increments of 2000 complex points with 4000 Hz spectral width in t2 and 10 000 Hz in tl. The data were processed using MestReNova 5.3 (Mestrelabs Research SL, Santiago de Compostella, Spain). Manual phase correction and Whittacker smoother baseline correction were applied to all spectra. Chemical shifts (<5) are expressed in ppm with reference to internal acetone (<5 2.225 for ¹ H and δ 31.08 for ¹³ C).

Methylation analysis

Polysaccharide samples (~5 mg) were per-methylated using CH3I and solid NaOH in DMSO, as described before [S.S. van Leeuwen et al., Carbohydr. Res. 343 (2008) 1237-1250]. After hydrolysis with 2 M trifluoroacetic acid (2 h, 120 °C), the partially methylated monosaccharides generated were reduced with NaBD ₄ (2 h, room temperature, aqueous solution), and the solution was neutralized with acetic acid. Subsequently, boric acid was removed by co-evaporation with methanol. The resulting partially methylated alditols were per-acetylated using pyridine:acetic anhydride (1:1 v/v) at 120°C yielding mixtures of partially-methylated alditol acetates, which were analyzed by GLC-EI-MS as described.

Enzymatic treatments with a-amylase, dextranase and pullulanase

The a-glucan samples (5 mg) were dissolved in 500 μΙ of sodium acetate buffer (50 mM pH 5.0), and incubated separately with excess amounts of a-amylase [Aspergillus oryzae; Megazyme), dextranase (Chaetomium erraticum; Sigma-Aldrich), and pullulanase Ml (Klebsiella planticola; Megazyme) at 37°C. After 48 h, the degree of hydrolysis was evaluated by TLC and/or HPAEC. Starch, dextran and pullulan, were used as positive controls for the a-amylase, dextranase and pullulanase treatments, respectively, obtaining fully hydrolyzed products under these conditions.

Characterization of GtfB-treated wheat flour: in vitro digestion Samples of refined wheat flour were treated with different amounts of L. reuteri CNCM 1-2452 GtfB enzyme and an in vitro method was used to evaluate digestibility.

Enzyme V(enzyme) V(H20) V(CaCI ₂ 50 mM) Enzyme [μΙ] [ml] [μΙ] per 100 mg Starch]

Reference - 16.283 333 667

LrGtfB 95 16.188 333 667

LrGtfB (-) 48 16.235 333 333.5

LrGtfB (+) 190 16.093 333 1334 Samples were prepared by adding 143 mg of pregelatinized refined wheat flour into a 50 ml falcon tubes and adding the required quantity of milli-Q H2O (see Table). Vortex mixing and stirring with a magnet was applied until homogenization (almost 30 min). 333 μΙ of 50 mM CaCI ₂ solution was added. Tubes were equilibrated at 37 °C in an oven at 45 rpm on a roller mixer at 60 rpm. The required quantity of enzyme was added and allowed to incubate for 24 h. Enzymes were inactivated by putting the tubes in boiling water for 6 min. The solutions were freeze-dried Phosphate buffer solution (PBS) (10 mM) was prepared in a 1000 mL volumetric flask by dissolving 0.26 g of KH ₂ P0 ₄ , 1.44 g Na ₂ HP0 ₄ *2H ₂ 0 and 8.71 g NaCI with 800 mL mQ H ₂ 0. The pH was adjusted to 6.9 with HCI (1M) and brought to the mark with mQ H ₂ 0. For the preparation of 100 mg/mL enzyme, 1.5 g of pancreatin (P) (Sigma Aldrich) or rat intestinal powder (RIP) (Sigma Aldrich) was mixed with 15 mL PBS (10 mM) in a centrifugation tube. The solution was Vortexed and sonicated on ice for 7 min. The tubes were centrifuged at 10Ό00 x g for 30 min at 4 °C. The supernatant was transferred to a plastic bottle.

The sample and reference contained 1 % (w/V) of total glucose in PBS-buffer and were stirred magnetically for 2 h before the start of the digestion. For each time (0, 15, 30, 60, 120 and 180 min), a set of 5 mL Eppendorf was prepared, one for the blank, one for the reference, and one for each sample. The blank contained PBS buffer only and the reference pregelatinized refined wheat flour, treated in the same way as the samples. For each time set, 300 μί (V _sa mpie) of the required solution were added to the 5 mL Eppendorf tubes (PBS, reference or sample). Pancreatin and RIP solution were equilibrated at 37 °C for 5 min in a water bath and the 5 mL Eppendorf of the time set were equilibrated at 37 °C in a thermomixer. 200 U/mg of pancreatin (V _p ) U) of RIP were added to each tube. One U corresponds to the amount of protein that releases ΙμιτιοΙ of glucose per min. The tubes were mixed (1000 rpm) and incubated at 37 °C, 450 rpm for the corresponding time (15, 30, 60, 120 and 180 min). After incubation, a 500 μί aliquot of the sample was added into 1.5 mL ethanol (EtOH) into 2 mL Eppendorf tubes were prepared before and stored at 4 °C. The tubes were centrifuged for 10 min at 10Ό00 x g. For time 0, the enzymes were replaced with 10 mM PBS and a 500 μί aliquot taken into 1.5 mL of EtOH and centrifuged under the same conditions as for the other points.

Free glucose was measured with the Wako glucose kit using glucose standards of 0, 0.125, 0.25, 0.5, 0.75, 1.0, 1.5 and 2.0 mg/ml. Total glucose release (total [Gl]) is determined as in Equation 1 where a and b are the slope and intercept of the standard curve, [Gl] _B iank is the blank sample with PBS buffer only, and F _d n is the dilution factor. Percentage of glucose release corresponds to the total glucose release divided by the mass of glucose in the sample (ITIGI total) multiplied by 100, Equation . total[Gl]

Equation g ] = [Gl] _Btanfc J · (V _sample + V(P) + V(R)) ^■ F _c

total[Gl]

Gl release[%] = — - ^■ 100

Equation 2 ^m Gi total

RESULTS AND DISCUSSION Identification of novel starch active GH70 enzymes within the NCC genome database

The NCC genome database was screened for novel GtfB-like enzymes. Among the GtfB enzymes identified were L reuteri CNCM 1-2451, L reuteri CNCM 1-2452, S. thermophilus CNCM 1-5167, S. thermophilus CNCM 1-5168, L. delbrueckii sbsp. delbrueckii CNCM 1-5166 and L. fermentum CNCM I- 5068 (a 4,3-a-GTase described in co-pending application EP16172606.2). The conserved motifs I to IV of these GtfB proteins were analyzed in detail (Fig. 1). Motifs I to IV of the GtfB enzymes identified in the NCC displayed clear similarity with those corresponding to previously characterized 4,6-a-GTases, and were easily identified. The order of these conserved regions I to IV in the GtfB sequences is ll-lll- IV-I, reflecting their circularly permutated domain organization. Six residues, conserved among these GH70 motifs, including the catalytic residues (D1015, E1053, D1125; L. reuteri 121 GtfB numbering) and residues involved in the formation of subsite -1 ( 1013, H1124, D1479 were found in all the identified GtfB protein sequences.

Regarding other functionally important positions in motifs III and IV, a unique sequence feature is the replacement of the W1065 (L reuteri 180 Gtfl80 numbering) residue of motif III forming a stacking interaction with the acceptor substrate in glucansucrases, by a tyrosine in the GtfB type of enzymes. Interestingly, a Tyr residue is also present in the L. fermentum CNCM 1-5068 GtfB 4,3-a-GTase, and it is strictly conserved throughout the GtfC and GtfD subfamilies as well. Thus, in this study this "sequence fingerprint" was used as a criterion to select only those GH70 enzymes active on starch. Second, in motif IV, previously characterizedGtfB 4,6-a-GTases have an invariant motif QRK downstream the transition state stabilizer (note that the alignment depicted in Fig. 1 predicts a one amino acid gap), whereas GSs and previously described GTFB 3,4 a-GTase show variations in this region. Previous mutational studies combined with structural data revealed that this region, and more specifically, residues 1137 and 1140 (first and fourth residue downstream the transition state stabilizer in L. reuteri 180 Gtfl80 GS) contribute to glycosidic linkage specificity in GSs. In case of the reuteran- like polymer synthesizing GtfD 4,6-a-GTase enzymes, the Gin residue at position 1137 is also conserved, whereas the Lys residue at position 1140 is substituted by a His and has been proposed to define this novel product specificity. In contrast to these differences, GtfB 4,6-a-GTases and GSs share high conservation in the subsite +1 Asn residue in motif II (N1019 in L. reuteri GfB), differing from GtfC and GtfD enzymes that contain a His residue at this position. This subsite +1 Asn residue is critical for the activity and linkage specificity of the Gtfl80 GSs.

Interestingly, the GtfB protein sequences of L. reuteri CNCM 1-2451, L. reuteri CNCM 1-2452, S. thermophilus CNCM 1-5167, S. thermophilus CNCM 1-5168 and L. delbrueckii sbsp. delbrueckii CNCM I- 5166 show differences in some of the residues in motifs II and IV forming the substrate-binding site. Similarly to GtfC and GtfD enzymes, the subsite +1 Asn residue (N1029 in L. reuteri Gtfl80 GS) is replaced by His in these five GtfB proteins. For the GtfB proteins of L. reuteri CNCM 1-2451 and L. reuteri CNCM 1-2452 the amino acids at positions 1137 and 1140 following the putative transition state stabilizer (Gtfl80 L. reuteri 180 numbering), are Ser and Ala, instead of the Gin and Lys residues typically found in most GtfB-and GtfC-like 4,6-a-GTases. For the GtfB proteins of L. delbrueckii sbsp. delbrueckii CNCM 1-5166 the amino acid at position 1140 following the putative transition state stabilizer (Gtfl80 L. reuteri 180 numbering) is also Ala. It is noteworthy that the L. fermentum CNCM I- 5068 GtfB, which shares high identity with L. reuteri 121 GtfB but displays 4,3-a-GTase activity, also contains unique variations in residues 1029, 1137 and 1140, providing support for these being "hot- spot" positions for product specificity in GtfB enzymes.

Amino acid sequence analysis and structure modelling of the GtfB enzyme of L. reuteri CNCM 1-2452

L. reuteri CNCM 1-2452 genome contains a single gene coding for a GH70 enzyme with a theoretical molecular mass of 145 kDa. As reported for other GH70 family proteins, the L. reuteri CNCM 1-2452 GH70 enzyme is predicted to function as an extracellular protein. Alignment of its amino acid sequence with biochemically characterized GH70 enzymes shows highest sequence identity with the L. fermentum GtfB 4,3-a-GTase (83% identity). The characterized GtfB 4,6-a-GTase enzymes of L. reuteri 121, Lactobacillus reuteri ML1 and Lactobacillus reuteri DSM 20016 also share significant amino acid identity (76%, 75% and 66% identity) with the L. reuteri CNCM 1-2452 GH70 enzyme, further indicating that this protein belongs to the GtfB subfamily of GH70 enzymes. The obtained 3D model of L. reuteri CNCM 1-2452 GH70 enzyme (Fig. 2), based on the L. reuteri 121 GtfB-ΔΝΔν 4,6-a-GTase crystal structure [Bai et al., (2016)], comprises domains A, B, C and IV; it reflects the high sequence similarity between the two enzymes (79% identity for these domains). The same "U-fold" domain organization is observed, with a circularly permuted catalytic (β/α)8 barrel in domain A, characteristic of GSs and GtfB type of enzymes. Sequence comparison revealed that the L. reuteri CNCM 1-2452 GH70 enzyme, like the L. reuteri 121 GtfB enzyme, also has an N-terminal variable domain (residues 1-446) and lacks a C-terminal variable domain. In its catalytic domain (A), the spatial arrangement of the catalytic residues in the active center is similar to that of L. reuteri 121 Gtf Β-ΔΝΔν. On the other hand, notable structural differences were observed between the substrate binding sites of the two enzymes. Most importantly, whereas the L. reuteri 121 enzyme features a tunnel extending beyond the active site formed by the 13-residue loop Al and the 20-residue loop B, the corresponding loops in the L. reuteri CNCM 1-2452 GH70 enzyme are only 6 and 4 residues long (802-807 and 590- 593, respectively; Fig. 2). As a consequence, these loops do not form a tunnel covering donor substrate binding subsites, rendering the binding groove fully accessible, like in a-amylases. Superposition with the maltopentaose-bound L. reuteri 121 GtfB-ΔΝΔν structure shows that residues in the highly similar loop A2 of the L. reuteri CNCM 1-2452 GH70 enzyme likely interact with bound substrates, and so may residue Y592 from loop B (Fig. 2). Also its tyrosine residue (Y1177, corresponding to Y1521 of the L. reuteri 121 GtfB enzyme) at subsite -6 is conserved (not shown) to provide an aromatic stacking interaction. Other notable differences are the presence of a histidine residue (H683) in motif II replacing the asparragine present in 4,6-a-GTases (N1019 in L. reuteri 121 GtfB), and the three residues following the transition state stabilizer in motif IV (SRA replacing QRK). On the other hand, its tyrosine residue near subsite +1 (Y719, motif III) is conserved with 4,6-a-GTases. Residues from these motifs are known to contribute to the product specificity of GH70 enzymes. These structural differences observed in the architecture of the active site of the 3D model of the L. reuteri CNCM 1-2452 GH70 enzyme prompted us to study the reaction and product specificity of this enzyme.

Purification and biochemical properties of the L. reuteri CNCM 1-2452 GH70 enzyme. Previous work showed that truncation of the N-terminal variable region of the L. reuteri 121 GtfB did not affect the enzyme catalytic properties, but facilitated protein expression [Y. Bai et al., Environ. Microbiol. 81 (2015) 7223-7232]. Thus, the L. reuteri CNCM 1-2452 gene encoding a GtfB enzyme was cloned and expressed in E. coli (DE3) BL21 star without its N-terminal variable region (amino acids 417 to 1281). Under the conditions used, high protein expression levels were observed in the soluble fraction, and following His tag affinity purification a total of ~50 mg of pure protein per liter of culture was obtained. SDS-PAGE analysis revealed a single protein band with an apparent molecular weight of ~100 kDa, which fits the predicted molecular mass deduced from its amino acid sequence (98 kDa). The purified L. reuteri CNCM 1-2452 GH70 enzyme was inactive with sucrose but active with maltodextrins/starch, confirming its identity as a GtfB-ΔΝ enzyme. In order to determine the best conditions for subsequent reactions, the effects of pH on its enzyme activity were determined by using amylose V as substrate. This GtfB-ΔΝ enzyme showed its maximal activity at pH 5.5, but exhibited a broad pH tolerance, retaining more than 80% of this activity over a pH from 4 to 9. This pH profile significantly differs from those reported for other GtfB enzymes, which showed significantly lower activities at basic pH values. The specific total activity value of the purified L. reuteri CNCM 1-2452 GtfB- ΔΝ on 0.125% (w v ^{" 1} ) amylose in 25 mM citrate phosphate buffer, pH 5.5, containing 1 mM CaC at 40 °C was 24 ± 0.6 U/mg. This value is similar to the one reported for the L. fermentum GtfB-ΔΝ 4,3-a- GTase (22 U/mg), but remarkably higher than that determined for the L. reuteri 121 GtfB 4,6-a-GTase, namely 2.8 U mg ¹ (at 40 °C and pH 5.5 and 5.0, respectively).

Substrate and product specificity of the L. reuteri CNCM 1-2452 GtfB enzyme

The L. reuteri CNCM 1-2452 GtfB-ΔΝ was incubated with different carbohydrate substrates at 37°C for 24 h, and its activity was compared with that of the L. reuteri 121 4,6-a-GTase GtfB. As shown by TLC (Fig. 3), both GtfB enzymes displayed hydrolysis and transglycosylase (disproportionation) activity on MOS with DP4 to DP7, as revealed by the formation of a range of shorter and longer oligosaccharide products. Both enzymes also accumulated polymeric material from MOS. In the case of the L. reuteri CNCM 1-2452 GtfB-ΔΝ, polymer accumulation was detected when using maltopentaose (DP5) and longer MOS substrates, whereas the L. reuteri 121 GtfB already formed polymer from maltotetraose (DP4). Note that for the L reuteri 121 GtfB, glucose clearly accumulated from the different MOS substrates. Instead, the L. reuteri CNCM 1-2452 GtfB-ΔΝ accumulates maltose and some low molecular mass oligosaccharides, but not glucose as a side product of its hydrolase/transglycosidase activity. This observation suggests that these two GtfB enzymes differ in their mode of action. Incubation of amylose V, potato starch and amylopectin with the L. reuteri CNCM 1-2452 GtfB-ΔΝ enzyme resulted in the appearance of some low molecular mass products that were not clearly detectable by TLC, but that were indicating that, similar to the L. reuteri 121 GtfB, this enzyme is also active on these polymeric substrates. The L. reuteri CNCM 1-2452 GtfB enzyme was inactive on sucrose, panose, nigerose, pullulan, dextran and isomalto-oligosaccharides with DP2, DP3, and DP5 (data not shown).

To study the product specificity of the L. reuteri CNCM 1-2452 GtfB-ΔΝ in more detail, the products synthesized from amylose V were analysed by one-dimensional ¹ H NM spectroscopy. As shown in Fig. 4A, this ¹ H NMR analysis revealed the presence of two broad anomeric signals indicative of (al->4) linkages (δ ~ 5.40-5.35) and (al->6) linkages (δ ~ 4.97); thus L. reuteri CNCM 1-2452 GtfB-ΔΝ also acts as a 4,6-a-GTase. Small signals corresponding to free glucose units (Get H-l, δ 5.225; H-l, δ 4.637) and 4-substituted reducing-end glucose residues (Rot H-l, δ 5.225; R H-l, δ 4.652) were detected as well, indicating that trace amounts of glucose, maltose and other small oligosaccharides were also present in this product mixture. This ^X H NMR spectrum was highly similar to those of the reuteran-like polymers synthesized by the A. chroococcum GtfD and P. beijingensis GtfD, as indicated by the presence of extra signals strongly overlapping in the (al->4) anomeric region (Fig.4A). Note that these signals are not present in the NMR spectrum of the IMMP generated by L. reuteri 121 GtfB (Fig. 4A).

The amylose-derived products from L. reuteri CNCM 1-2452 GtfB-ΔΝ were also analyzed by HPSEC with multidetection. The HPSEC profile of the original amylose V substrate consisted of a single peak eluting at ~21 ml with an average M _w of 200 ^χ 10 ³ Da. As shown in Fig. 4B, the action of the L reuteri CNCM I- 2452 GtfB-ΔΝ on amylose V resulted in the formation of a peak at a higher elution volume (~ 29 ml) corresponding to a low molecular mass a-glucan with an average M _w of 7 ^χ 10 ³ Da, together with a small shoulder peak corresponding to maltose. This HPSEC profile significantly differs from those reported for other 4,6-a-GTases [EP2427565, PCT/EP2016/071474] producing higher molecular mass polymers from amylose V. For example, the M _w value of the a-glucan generated by L. reuteri CNCM I- 2452 GtfB-ΔΝ is half that of the IMMP products of L reuteri 121 GtfB (15 x 10 ³ Da). On the other hand, it is much smaller than the HMM polysaccharide synthesized by the A. chroococcum GtfD, which had an average M _w of 13 ^χ 10 ^s Da. The L. reuteri NCC2613 GtfB-ΔΝ product profile is also different from that of P. beijingensis GtfD which showed a bimodal polymer distribution containing both HMM (27 ^χ 10 ^s Da) and LMM (19 x 10 ³ Da) polymers (Fig. 4B).

Structural characterization of the L. reuteri CNCM 1-2452 GtfB-ΔΝ LMM polysaccharide

To further explore the structural characteristics of the L reuteri CNCM 1-2452 GtfB-ΔΝ LMM polysaccharide, the amylose-derived reaction mixture was subjected to Bio-Gel P-2 size-exclusion chromatography. ID NMR analysis of this polysaccharide showed a linkage ratio (al->4):(al->6) = 75:25. The typical chemical shift values corresponding to consecutive (al->6) linkages were not identified in the 2D NMR spectra of this L. reuteri CNCM 1-2452 GtfB-ΔΝ polymer (Fig. 5). Methylation analysis revealed the presence of terminal, 4-substituted, 6-substituted, and 4,6- disubstituted glucopyranose residues in a molar percentage of 15, 59, 10, and 16%, which is in agreement with the linkage ratios determined by ^X H NMR. Taken together, these data confirm that similar to GtfD type of enzymes, the L. reuteri CNCM 1-2452 GtfB-ΔΝ synthesizes a reuteran-like a- glucan, providing the first evidence of this product specificity within the GtfB-like GH70 subfamily. The structural characteristics of the different amylose-derived reuteran type of polymers are summarized in Table 1. Regarding its size and (al->4):(al->6) linkage ratio, the a-glucan synthesized by the L. reuteri CNCM 1-2452 GtfB-ΔΝ resembles mostly the LMM P. beijingesis GtfD polymer, however, it contains higher amounts of alternating (al->4)/(al->6) glycosidic linkages as indicated by the increased amount of 6-substituted glucopyranosyl units (i.e. 10% rather than 5%).

TABLE 1. Structural characterization of the polysaccharide formed upon incubation of amylose V with the L. reuteri CNCM 1-2452 GtfB-ΔΝ enzyme. For comparison the characteristics of the polymer produced by the A. chroococcum and P. beijingensis GtfD 4,6-a-GTases are included as well.

P. beijingensis P. beijingensis L. reuteri CNCM 1-

Type of glucosyl A. chroococcum

Parameter GtfD HMM GtfD LMM 2452 GtfB-ΔΝ units GtfD polymer

polymer polymer polymer

Glcp(l- 19 17 15 15

Methylation - 4)-Glcp-(l- 45 54 62 59 analysis (%) - 6)-Glcp-(l- 18 11 5 10

- 4,6)-Glcp-(l- 18 18 18 16

NMR chemical (al- 4) 68 71 77 75 shift (%) ^a (al- 6) 32 29 23 25

Molecular

mass 13 10 ³ 27 10 ³ 19 7

(10 ³ Da) ^b

a The data represent the ratios of integration of the surface areas of the (al->6) linkage signal at 4.97 ppm and the ( 1->4) linkage signal at 5.36 ppm in the ¹ H NMR spectra of the polysaccharides. The average molecular mass of polysaccharide was determined in duplicate.

Oligosaccharides formed from maltoheptaose in time by the L. reuteri CNCM 1-2452 GtfB-ΔΝ To gain a better understanding of the L. reuteri NCC 2623 GtfB-ΔΝ reuteran-like product formation, the oligosaccharides formed from maltoheptaose in time were analyzed by HPAEC (Fig. 6). From maltoheptaose (slightly contaminated with maltohexaose and maltopentaose), the L. reuteri CNCM I- 2452 GtfB-ΔΝ released maltose, maltotriose and maltopentaose as the main hydrolysis products, at the early stage of the reaction. Together with these first clear hydrolysis products, a significant number of peaks eluted at higher elution times than the maltooctaose standard (elution time = 45.7 min) with products resulting from its transglycosylating activity. After 24 h of reaction, the maltoheptaose substrate was completely depleted, whereas some MOS of low DP and oligosaccharides of unknown structure remained in the reaction mixture. Notably, only trace amounts of glucose were detected during the 24 h of reaction. The formation of maltose and maltotriose as main hydrolysis products, combined with the appearance of peaks corresponding to oligosaccharides with DP higher than 8, suggests that the L. reuteri CNCM 1-2452 GtfB enzyme preferentially transfers MOS of different DP (instead of glucose) to another glucan chain to form a reuteran-like polymer. This mechanism of polymerization differs from the one observed for GSs, which only transfer a single glucosyl unit per reaction cycle. Instead, the mode of action of the L. reuteri CNCM 1-2452 GtfB-ΔΝ resembles that of GtfD 4,6-a-GTase which also produces reuteran-type polymers [PCT/EP2016/071474]. Similarly, the L. fermentum GtfB 4,3-a-GTase converts amylose into a polymer containing alternating (al->3)/(al->4) linkages and (al->3,4) branching points by transferring MOS of different DPs [co-pending application EP16172606.2]. Whereas in GSs the active site is blocked beyond subsite -1, the GH70 starch-active enzymes appear to have more than one donor substrate binding subsite, allowing the elongation process to occur by successive transfer of MOS units coming from starch.

L reuteri CNCM 1-2452 GtfB-ΔΝ acceptor substrate reaction studies The acceptor substrate specificity of the L. reuteri CNCM 1-2452 GtfB-ΔΝ enzyme and L. reuteri 121 GtfB were compared by incubating the enzymes in the presence or absence of maltose and isomaltose as acceptor substrates for 24 h. As depicted in Fig. 7 A, maltose can serve as acceptor substrate for L. reuteri CNCM 1-2452 GtfB-ΔΝ in the presence of amylose V as donor substrate, resulting in formation of panose, maltotriose, maltotetraose and maltopentaose, and significant amounts of other unidentified oligosaccharides. The L. reuteri 121 GtfB displayed a different product distribution, but this enzyme was also able to use maltose as an acceptor substrate, yielding panose, maltotriose and maltotetraose, together with a series of elongated products with successive (al->6) linkages and increasing degrees of polymerization (Fig. 7B). These results suggest that both GtfB enzymes can elongate maltose forming either a new (al->4) or (al->6) linkage. Acceptor reactions with isomaltose more clearly reflected the different modes of action of these GtfB enzymes. The L. reuteri 121 GtfB enzyme clearly preferred isomaltose as acceptor substrate over maltose, as indicated by the detection of significant amounts of isomaltotriose, isomaltotetraose and isomaltopentaose (resulting from the elongation of the isomaltose by successive (al->6) linkages). In contrast, no significant change in oligosaccharide formation was observed when amylose V was incubated with L. reuteri CNCM 1-2452 GtfB-ΔΝ in the presence or absence of isomaltose. Thus, whereas the L. reuteri 121 GtfB preferentially elongates oligosaccharides with al->6 linked non-reducing ends, the L. reuteri CNCM 1-2452 GtfB-ΔΝ is unable to recognize isomaltose as an acceptor substrate, similar to A. chroococcum GtfD [PCT/EP2016/071474]. In agreement with these observations, the L. reuteri CNCM 1-2452 GtfB-ΔΝ and L. reuteri 121 GtfB products differ by the absence or presence of consecutive (al->6) linkages in their structures, respectively. Enzymatic hydrolysis of the L. reuteri CNCM 1-2452 GtfB-ΔΝ reuteran-like polysaccharide

The reuteran-like structure of the a-glucan produced by L. reuteri CNCM 1-2452 GtfB-ΔΝ was further confirmed by treating this α-glucan with excess amounts of different hydrolytic enzymes: a-amylase, dextranase and pullulanase. For comparison, the IMMP synthesized by the L. reuteri 121 GtfB 4,6-a- GTase and the reuteran-like polymers produced by the A. chroococcum and P. beijingensis Gtf D 4,6-a- GTases were subjected in parallel to the same enzymatic treatments. As shown in Fig. 8, the L. reuteri CNCM 1-2452 GtfB-ΔΝ polymer was quite resistant to the endo-(l->4) hydrolase activity of the a- amylase. Compared to the amylose control that was completely degraded, only trace amounts of HMM oligosaccharides and maltose were formed when this polymer was incubated with the a-amylase. Similar hydrolytic patterns were obtained for the reuteran-like polymers synthesized by the A. chroococcum and P. beijingensis GtfD 4,6-a-GTases, whereas these small amounts of maltose or other oligosaccharides were not detected in the case of the IMMP digestion. The L. reuteri CNCM 1-2452 GtfB-ΔΝ polymer was also subjected to dextranase and pullulanase Ml enzymatic hydrolysis, which catalyses the hydrolysis of (l->6) glycosidic linkages. Whereas dextranase specifically attacks linear sequences of (al->6)-linked D-glucopyranosyl repeating units, pullulanase is specific for al->6 linkages in the backbone chains of pullulan and at branching points of starch molecules. For the dextranase and pullulanase enzymatic treatments, dextran and pullulan were used as positive controls, respectively. As expected, the L. reuteri CNCM 1-2452 GtfB-ΔΝ polymer and the A chroococcum and P. beijingensis GtfD polymers were not degraded by the action of dextranase, and instead these polymers were hydrolyzed by pullulanase. In contrast, the IMMP product was resistant to the pullulanase treatment, but it was digested by the endo-(al->6)-hydrolase activity of dextranase. These results are in agreement with the presence of only successive (al->6) linkages in the L. reuteri 121 GtfB polymer and their absence in the L. reuteri CNCM 1-2452 GtfB-ΔΝ polymer. Similar to the reuteran type of polymers synthesized by GtfA GS and GtfD 4,6-a-GTases, and differing from the IMMP, this L. reuteri CNCM 1-2452 GtfB-ΔΝ polymer appears to contain alternating (al->6)/(al->4) linkages and (al->4,6) branching points.

Further information about the structure of the L. reuteri CNCM 1-2452 GtfB-ΔΝ polymer was obtained by the identification of the reaction products that resulted from the pullulanase treatment by HPAEC. As shown in Fig. 9A, the pullulanase digested the L. reuteri CNCM 1-2452 GtfB-ΔΝ polymer, yielding mainly glucose, and a mixture of MOS from DP2 to 7. This finding indicates that this polymer is formed by maltose, maltotriose, maltotetraose, maltopentaose, maltohexaose and maltoheptaose elements linked by single (al->6) linkages. These structural elements are also present in the LMM P. beijingensis GtfD polymer, however with longer linear (al->4) sequences (from DP8 to DP13) also being detected (Fig. 9B). Pullulanase degraded the HMM reuteran polymers synthesized by P. beijingensis GtfD and A. chroococcum GtfD enzymes into MOS units up to DP6 and DP5, respectively (Fig. 9C and 9D). Overall, these HPAEC profiles suggest that the 4,6-a-GTases characterized so far have a preference for transferring different lengths of (al->4) glucan chains, yielding as a result, reuteran polymers with unique structures. Figure 10 shows composite models for the reuteran-like polymers produced by the L. reuteri NCC2613 GtfB-ΔΝ and the previously characterized GtfD type of enzymes. The L. reuteri NCC2613 GtfB-ΔΝ enlarges the variety of reuteran-like a-glucans that can be easily synthesized using GH70 enzymes from amylose.

Characterization of GtfB-treated wheat flour: in vitro digestion Wheat flour samples were treated with different concentrations of L. reuteri CNCM 1-2452 GtfB enzyme as described above. First, the percentage of glucose released by the samples was analyzed by in-vitro digestion. This measurement was set-up to mimic human digestion and gives the percentage of glucose released by the sample compared to a reference.

Glucose released from refioned pregelatinized wheat flour modified with different concentration (333.5 (LrGtfB (-)), 667 (LrGtfB) and 1334 (LrGtfB (+)) μg/100 mg starch) of L reuteri CNCM 1-2452 GtfB were compared with the reference (Fig. 11). The reference pregelatinized refined wheat flour was rapidly digested by more than 70 % after 15 min and reached a plateau (ca. 85 %) after 180 min. For wheat flours treated with different concentrations of L. reuteri CNCM 1-2452 GtfB, the higher the enzyme concentration, the lower the digestibility.