YEB

Recent Publications in this Series QIAO SHI Synthesis and Structural Characterization of Glucooligosaccharides and Dextran from

4/2016 Hany S.M. EL Sayed Bashandy Flavonoid Metabolomics in Gerbera hybrida and Elucidation of Complexity in the Flavonoid Biosynthetic Pathway 5/2016 Erja Koivunen Home-Grown Grain Legumes in Poultry Diets 6/2016 Paul Mathijssen dissertationes schola doctoralis scientiae circumiectalis, Holocene Carbon Dynamics and Atmospheric Radiative Forcing of Different Types of Peatlands alimentariae, biologicae. universitatis helsinkiensis 21/2016 in Finland 7/2016 Seyed Abdollah Mousavi Revised Taxonomy of the Family Rhizobiaceae, and Phylogeny of Mesorhizobia Nodulating Glycyrrhiza spp. 8/2016 Sedeer El-Showk Auxin and Cytokinin Interactions Regulate Primary Vascular Patterning During Root QIAO SHI Development in Arabidopsis thaliana 9/2016 Satu Olkkola Antimicrobial Resistance and Its Mechanisms among Campylobacter coli and Campylobacter Synthesis and Structural Characterization of upsaliensis with a Special Focus on Streptomycin 10/2016 Windi Indra Muziasari Glucooligosaccharides and Dextran from Impact of Fish Farming on Antibiotic Resistome and Mobile Elements in Baltic Sea Sediment Weissella confusa Dextransucrases 11/2016 Kari Kylä-Nikkilä Genetic Engineering of Lactic Acid Bacteria to Produce Optically Pure Lactic Acid and to Develop a Novel Cell Immobilization Method Suitable for Industrial Fermentations 12/2016 Jane Etegeneng Besong epse Ndika Molecular Insights into a Putative Potyvirus RNA Encapsidation Pathway and Potyvirus Particles as Enzyme Nano-Carriers 13/2016 Lijuan Yan Bacterial Community Dynamics and Perennial Crop Growth in Motor Oil-Contaminated Soil in a Boreal Climate 14/2016 Pia Rasinkangas New Insights into the Biogenesis of Lactobacillus rhamnosus GG Pili and the in vivo Effects of Pili 15/2016 Johanna Rytioja Enzymatic Plant Cell Wall Degradation by the White Rot Fungus Dichomitus squalens 16/2016 Elli Koskela Genetic and Environmental Control of Flowering in Wild and Cultivated Strawberries 17/2016 Riikka Kylväjä

Staphylococcus aureus and Lactobacillus crispatus: Adhesive Characteristics of Two Gram- Weissella confusa Dextransucrases Positive Bacterial Species 18/2016 Hanna Aarnos Photochemical Transformation of Dissolved Organic Matter in Aquatic Environment – From a Boreal Lake and Baltic Sea to Global Coastal Ocean 19/2016 Riikka Puntila Trophic Interactions and Impacts of Non-Indigenous Species in the Baltic Sea Coastal Ecosystems DEPARTMENT OF FOOD AND ENVIRONMENTAL SCIENCES 20/2016 Jaana Kuuskeri FACULTY OF AGRICULTURE AND FORESTRY Genomics and Systematics of the White-Rot Fungus Phlebia radiata: DOCTORAL PROGRAMME IN FOOD CHAIN AND HEALTH Special Emphasis on Wood-Promoted Transcriptome and Proteome UNIVERSITY OF HELSINKI 21/2016

Helsinki 2016 ISSN 2342-5423 ISBN 978-951-51-2451-7 Department of Food and Environmental Sciences Faculty of Agriculture and Forestry University of Helsinki

Doctoral Programme in Food Chain and Health Doctoral School in Environmental, Food and Biological Sciences

Synthesis and structural characterization of glucooligosaccharides and dextran from Weissella confusa dextransucrases

Qiao Shi

Doctoral Thesis

To be presented, with the permission of the Faculty of Agriculture and Forestry, University of Helsinki, for public examination in Auditorium 2, Infocenter Korona (Viikinkaari 11), on September 29th, 2016, at 12 o’clock noon.

Helsinki 2016 Custos Professor Maija Tenkanen Department of Food and Environmental Sciences University of Helsinki, Finland

Supervisor Professor Maija Tenkanen Department of Food and Environmental Sciences University of Helsinki, Finland

Reviewers Dr. Gregory L. Côté Agricultural Research Service United States Department of Agriculture, USA

Professor Markus Linder School of Chemical Technology Aalto University, Finland

Opponent Professor Henk Schols Laboratory of Food Chemistry Wageningen University, The Netherlands

Cover picture: Representation of a dextransucrase (based on Protein Data Bank no. 3HZ3) and its oligosaccharide products, made by Qiao Shi.

ISBN 978-951-51-2451-7 (Paperback) ISBN 978-951-51-2452-4 (PDF) ISSN 2342-5423 (Print) ISSN 2342-5431 (Online)

Unigrafia Helsinki 2016 ABSTRACT

Many lactic acid bacteria are able to synthesize dextrans from sucrose by dextransucrases. Weissella confusa strains have attracted increasing attention due to their production of texture-modifying dextrans during food fermentation. Potential prebiotic oligosaccharides have also been produced by dextransucrases in the presence of sucrose and acceptor sugars. However, only few W. confusa dextransucrases have been previously studied, and the reactions of Weissella dextransucrases to synthesize oligosaccharides have not been investigated. In this thesis, two W. confusa dextransucrases from efficient dextran-producers were studied, and their products were structurally characterized because this information is essential for a better usability of these enzymes and corresponding bacteria in food and health applications.

The biochemical and kinetic properties of one of the W. confusa dextransucrases were studied. Two activity assays were compared to determine the kinetic parameters. A sucrose radioisotope assay gave a KM of 14.7 mM and a Vmax of 8.2 µmol/(mg∙min), whereas a Nelson-Somogyi assay gave values of 13.0 mM and 19.9 µmol/(mg∙min), respectively. The dextrans from the two W. confusa dextransucrases were found by, e.g., NMR analysis to consist of mainly α-(1→6) linkages and 3% α-(1→3) branches, of which some were elongated. A high-performance size-exclusion chromatography analysis of the dextrans revealed high molar masses of 107‒108 g/mol.

Weissella dextransucrases were also studied with acceptor sugars for the synthesis of glucooligosaccharides. The most efficient acceptor was maltose, followed by isomaltose, maltotriose, and nigerose, which formed series of glucooligosaccharides by the further elongation of intermediate acceptor products. The products derived from maltose formed two homologous series, with one series being predominant and the other being minor. The major maltose product series were linear isomaltooligosaccharides (IMOs) with reducing-end maltose units, as identified by multistage mass spectrometry (MSn) analysis. The minor maltose series were revealed by an NMR analysis of the isolated minor trisaccharide product to bear a novel branched structure. These products contained an α-(1→2)-linked glucosyl residue on the reducing residue of the linear IMOs. These structures have not been previously obtained by a dextransucrase. They probably formed by the attachment of a single-unit branch to linear IMOs.

For the acceptor analogs lactose and cellobiose, their main acceptor products were identified by NMR analysis to be branched trisaccharides, with a glucosyl residue α- (1→2) linked to the acceptor’s reducing end. Surprisingly, a side product, isomelezitose

1 (6Fru-α-Glcp-sucrose), was produced when using lactose as an acceptor. The synthesis of this nonreducing trisaccharide by a dextransucrase was reported here for the first time.

Linear IMOs with a degree of polymerization ≥3.3 serve as prebiotics for their resistance to digestion. However, industrial IMO production often leads to a high portion of unwanted digestible sugars. The dextransucrase reaction in the presence of the efficient acceptor maltose was demonstrated here as a promising alternative synthesis process to control IMO size distribution by varying the sucrose/maltose ratio. The effects of substrate concentrations (0.15–1 M) and dextransucrase dosage (1–10 U/g sucrose) on the IMO yield and profile were modeled. High sucrose (1 M) and medium maltose (0.5 M) concentrations were found to be optimal for the synthesis of long-chain IMOs, with 366 g/L of total IMOs attained.

2 ACKNOWLEDGEMENTS

This thesis was carried out at the Department of Food and Environmental Sciences, Faculty of Agriculture and Forestry, University of Helsinki. The work was financially supported by the Finnish Graduate School on Applied Bioscience: Bioengineering, Food & Nutrition, Environment (ABS Graduate School), the Academy of Finland via project “Molecular and functional characterization of dextran production in Weissella spp. ‒ Superior dextran producers for cereal applications (WISEDextran)” and the Finnish Food and Drink Industries' Federation (ETL).

I am deeply indebted to my supervisor Professor Maija Tenkanen. She provided me with the great opportunity to pursue a doctoral degree in her research group and has offered me insightful scientific guidance, advice, and encouragement throughout the years. Her open-mindedness, curiosity, and enthusiasm towards science have inspired me. It was also a privilege for me to work with Docent Liisa Virkki, who introduced me to NMR spectral interpretation. I feel grateful to Professor Vieno Piironen and Research Professor Kristiina Kruus for participating in my follow-up group during my thesis work. I also want to express my sincere gratitude to Dr. Gregory L. Côté and Professor Markus Linder for their pre-examination of this thesis.

I owe my gratitude to Docent Hannu Maaheimo, Assistant Professor Kati Katina, Docent Päivi Tuomainen, Minna Juvonen, Dr. Ndegwa Henry Maina, Kaj-Roger Hurme, Dr. Sanna Koutaniemi, and Dr. Leena Pitkänen, who gave me invaluable help in their fields of expertise, as well as Yaxi Hou, who assisted me with laboratory work. I also thank the project collaborators Dr. Riikka Juvonen, Ilkka Kajala, and Dr. Antti Nyyssölä from the VTT Technical Research Centre of Finland Ltd. and Professor Arun Goyal and Shraddha Shukla from the Indian Institute of Technology Guwahati (IITG). This thesis could not have been completed without their fruitful collaborations.

I am fortunate to have been surrounded by friendly and supportive colleagues along the journey, including Sun-Li Chong, Taru Rautavesi, Miikka Olin, Kirsi Mikkonen, Susanna Heikkinen, Kirsti Parikka, Minna Malmberg, Laura Huikko, Paula Hirsilä, Abdul Ghafar, Suvi Alakalhunmaa, Bhawani Chamlagain, Yan Xu, and Yu Wang, to name a few. I truly appreciate the joyful time I shared with all of the nice people from the department, especially those in D building.

Last but not least, my warmest gratitude goes to my dearest family, Xiaoguang Shi, Wenxiu Zhu, and Ling Zou, for being my rock and having faith in me! I cannot thank them enough!

3 LIST OF ORIGINAL PUBLICATIONS

This thesis is based on the following publications:

I. Kajala I, Shi Q, Nyyssölä A, Maina NH, Hou Y, Katina K, Tenkanen M, Juvonen R. 2015. Cloning and characterization of a Weissella confusa dextransucrase and its application in high fibre baking. PLoS ONE 10:e0116418.

II. Shukla S, Shi Q, Maina NH, Juvonen M, Tenkanen M, Goyal A. 2014. Weissella confusa Cab3 dextransucrase: properties and in vitro synthesis of dextran and glucooligosaccharides. Carbohydr. Polym. 101:554-64.

III. Shi Q, Juvonen M, Hou Y, Kajala I, Nyyssölä A, Maina NH, Maaheimo H, Virkki L, Tenkanen M. 2016. Lactose- and cellobiose-derived branched trisaccharides and a sucrose-containing trisaccharide produced by acceptor reactions of Weissella confusa dextransucrase. Food Chem. 190:226-36.

IV. Shi Q, Hou Y, Juvonen M, Tuomainen P, Kajala I, Shukla S, Goyal A, Maaheimo H, Katina K, Tenkanen M. 2016. Optimization of isomaltooligosaccharide size distribution by acceptor reaction of Weissella confusa dextransucrase and characterization of novel α-(1→2)-branched isomaltooligosaccharides. J. Agric. Food Chem. 64:3276-86.

The author’s contribution

I. Shi Q planned and performed the experiments related to dextransucrase and dextran characterization. She interpreted the results and prepared that part of the manuscript.

II. Shi Q was responsible for producing and characterizing glucooligosaccharides. She participated in planning and performing the related experiments, interpreted the results, and prepared the manuscript, including the introduction and parts related to glucooligosaccharide work and most of the dextran characterization.

III. Shi Q participated in planning the study and was responsible for all of the experimental work and result interpretation, except for the MS analysis. She prepared the manuscript except for the MS part and was the corresponding author.

4 IV. Shi Q was co-responsible for the experimental work and results interpretation related to optimization. She planned and performed most of the experiments and interpreted the results regarding novel oligosaccharides. She prepared the manuscript and was the corresponding author.

Other related publications of the author:

Juvonen R, Honkapää K, Maina NH, Shi Q, Viljanen K, Maaheimo H, Virkki L, Tenkanen M, Lantto R. 2015. The impact of fermentation with exopolysaccharide producing lactic acid bacteria on rheological, chemical and sensory properties of pureed carrots (Daucus carota L.). Int. J. Food Microbiol. 207:109-18.

Kajala I, Mäkelä J, Coda R, Shukla S, Shi Q, Maina NH, Juvonen R, Ekholm P, Goyal A, Tenkanen M, Katina K. 2016. Rye bran as fermentation matrix boosts in situ dextran production by Weissella confusa compared to wheat bran. Appl. Microbiol. Biotechnol. 100:3499-510.

5 ABBREVIATIONS

BIMi (i≥3) maltose-derived branched isomaltooligosaccharides of DPs≥3 Cel3 cellobiose-derived trisaccharide COSY homonuclear correlation spectroscopy

D2O deuterium oxide DMSO dimethyl sulfoxide DNS 3,5-dinitrosalicylic acid DP degree of polymerization DQFCOSY double-quantum-filtered COSY DSR-S dextransucrase from Leuconostoc mesenteroides NRRL B- 512F EPS exopolysaccharides ESI-IT-MSn electrospray ionization multistage ion trap mass spectrometry GH glycoside hydrolase GLOSs glucooligosaccharides HePS heteropolysaccharides HMBC heteronuclear multiple-bond correlation spectroscopy HoPS homopolysaccharides HPAEC-PAD high-performance anion-exchange chromatography with pulsed amperometric detection HPSEC high-performance size-exclusion chromatography HSQC heteronuclear single-quantum coherence spectroscopy IMOs isomaltooligosaccharides LAB lactic acid bacteria Lac3 lactose-derived trisaccharide LIMi (i≥4) maltose-derived linear isomaltooligosaccharides of DPs≥4 NMR nuclear magnetic resonance NR3 nonreducing sucrose-containing trisaccharide PEG polyethylene glycol Q2 the coefficient of prediction in regression analysis R2 the coefficient of determination in regression analysis TLC thin-layer chromatography TOCSY total correlation spectroscopy WcCab3-DSR dextransucrase from W. confusa Cab3 WcE392-rDSR recombinant dextransucrase from W. confusa VTT E-90392

6 TABLE OF CONTENTS

Abstract ...... 1 Acknowledgements ...... 3 List of original publications ...... 4 Abbreviations ...... 6 1. Introduction ...... 9 2. Review of the literature ...... 11 2.1 Carbohydrates from lactic acid bacteria ...... 11 2.2 Glucansucrases ...... 12 2.2.1 Catalytic mechanism of glucansucrases ...... 12 2.2.2 Production and characterization of dextransucrases ...... 15 2.3 Carbohydrates synthesized by dextransucrases ...... 16 2.3.1 Substrate and product specificity of dextransucrases ...... 16 2.3.2 Functional oligosaccharides from dextransucrases ...... 19 2.3.3 Characterization of dextransucrase-synthesized carbohydrates ...... 21 2.4 Weissella spp. and Weissella dextransucrases ...... 23 3. Aim of the study ...... 26 4. Materials and methods ...... 27 4.1 Dextransucrase preparations and commercial enzymes ...... 27 4.2 Biochemical characterization of WcE392-rDSR (I) ...... 27 4.3 Analysis of monosaccharides and oligosaccharides ...... 28 4.4 Synthesis, isolation, and molar mass characterization of dextrans (I and II) ..... 29 4.5 Acceptor reactions and isolation of oligosaccharides (II‒ IV) ...... 29 4.6 Structural analysis of dextrans and oligosaccharides ...... 31 4.6.1 Enzymatic hydrolysis of dextrans and oligosaccharides (II and III) ...... 31 4.6.2 NMR spectroscopy analysis of dextrans and oligosaccharides (I‒IV) ...... 31 4.6.3 MSn analysis of oligosaccharides (II‒IV) ...... 32 4.7 Experimental design for IMO synthesis (IV) ...... 32 5. Results ...... 34 5.1 Characterization of WcE392-rDSR (I) ...... 34

7 5.2 Characterization of Weissella dextrans (I and II) ...... 35 5.3 Characterization of the oligosaccharides produced by dextransucrases ...... 38 5.3.1 Acceptor selectivity (III) ...... 38 5.3.2 Principal maltose product series (II) ...... 38 5.3.3 Minor maltose product series (IV) ...... 41 5.3.4 Lactose- and cellobiose-derived branched trisaccharides (III) ...... 42 5.3.5 Novel sucrose-containing trisaccharide (III) ...... 46 5.3.6 Applicability of MSn analysis in the characterization of dextransucrase- synthesized GLOSs (III and IV) ...... 48 5.4 Optimization of IMO synthesis by response surface modeling (IV) ...... 52 6. Discussion ...... 57 6.1 Properties of Weissella dextransucrases and dextrans (I and II) ...... 57 6.2 Specificity of Weissella dextransucrases in acceptor reactions ...... 58 6.2.1 Comparison with commercial Lc. mesenteroides dextransucrase (III and IV) ...... 58 6.2.2 Specificity to form an α-(1→2) branch (III and IV) ...... 60 6.2.3 Novel sucrose-containing trisaccharide (III) ...... 61 6.3 Potential of dextransucrases in prebiotic GLOS synthesis ...... 62 6.3.1 Synthesis of prebiotic IMOs from maltose (IV) ...... 62 6.3.2 Synthesis of other GLOSs (III) ...... 63 7. Conclusion ...... 65 References ...... 67

8 1. INTRODUCTION

The lactic acid fermentation of food has been practiced worldwide for thousands of years for improved preservability and flavors. Moreover, the oligo- and polysaccharides secreted by lactic acid bacteria (LAB) can enhance the textural and nutritional properties of the fermented foods (Seo DM et al., 2007; Welman 2009; Di Cagno et al., 2013; Cho et al., 2014). The exopolysaccharides (EPS) produced in situ by LAB can avoid being declared on food labels, because the LAB with a history of safe use have, for example, GRAS (generally recognized as safe) or QPS (qualified presumption of safety) status in the United States and the European Union, respectively (Waldherr and Vogel 2009). Such EPS offer an attractive solution for producing additive-free foods, which are increasingly pursued by today’s consumers. EPS-producing cultures have been applied in fermented milks, vegetables, and fruits and in cheese and sourdoughs, where EPS exert technological advantages, such as controlling syneresis; enhancing the thickness, creaminess, and smoothness of yoghurt; and improving the dough rheology, bread volume, and shelf life of sourdough breads (Di Cagno et al., 2006; Schwab et al., 2008; Welman 2009; Katina et al., 2009; Galle et al., 2012). In addition to sensory benefits, there is accumulated evidence on the health benefits of oligo- and polysaccharides from LAB, including prebiotic, cholesterol-lowing, immunomodulatory, and anti- carcinogenic effects (Schwab et al., 2008; Welman 2009; Ryan et al., 2015).

In addition to the in situ synthesis of oligo- and polysaccharides in fermented foods, LAB have been used to produce α-glucans of industrial interest (Zannini et al., 2015; Miao et al., 2016). For example, dextran produced by Leuconostoc mesenteroides was one of the first biopolymers produced on an industrial scale and has various applications in food, medicine, and separation technology as a food thickener, blood plasma expander, and size-exclusion chromatography matrix, respectively. (Waldherr and Vogel 2009; Zannini et al., 2015). Dextran was approved by the European Commission for use as a novel food ingredient in bakery products for improved softness, crumb texture, and loaf volume (Zannini et al., 2015). LAB produce α-glucans extracellularly via sucrose-acting α-transglucosylases, e.g., dextransucrases. These enzymes are promising biosynthetic tools because, unlike Leloir type-glycosyltransferases, they do not require energy-rich sugar nucleotides as glycosyl donor substrates and can readily synthesize α-glucans from a sole substrate sucrose (Monsan et al., 2010). In addition to polymers, these enzymes can also synthesize, e.g., α-glucooligosaccharides (GLOSs) in the presence of a wide variety of acceptor sugars. These enzymes of LAB have high regio- and stereospecificities compared to those of chemical approaches and have been used to synthesize structurally diverse carbohydrates (Leemhuis et al., 2013; Vuillemin et al., 2016). For example, Leuconostoc dextransucrases have been used to produce various

9 prebiotic GLOSs, e.g., isomaltooligosaccharides (IMOs) (Goffin et al., 2011; Ruiz- Matute et al., 2011; García-Cayuela et al., 2014).

Weissella spp. are LAB frequently found in plant materials and fermented foods (Fusco et al., 2015). They have drawn attention in recent years as efficient dextran producers for food applications. In situ-produced Weissella dextrans improve the shelf life, volume, and softness of sourdough breads and the texture of fermented vegetable foods (Katina et al., 2009; Galle et al., 2010; Juvonen et al., 2015). Weissella spp. have an advantage over many Leuconostoc starters in that they do not lead to over-acidification in foods. To facilitate the future use of this genus and its dextransucrases, it is important to understand the structures and possible functions of its products. To date, only few dextransucrases from W. confusa and W. cibaria have been characterized (Bounaix et al., 2010; Amari et al., 2012; Bejar et al., 2013; Rao and Goyal 2013), and no Weissella dextransucrase has been examined as a synthetic tool for, e.g., potential prebiotic GLOSs. In this thesis, the dextransucrases from two efficient dextran-producing W. confusa strains were studied, with one dextransucrase biochemically characterized (article I). Their dextran (articles I and II) and GLOS products (articles II‒IV) were structurally analyzed, focusing on GLOSs produced in the presence of various acceptor sugars. The reaction with the acceptor maltose was explored to synthesize prebiotic IMOs (article IV).

10 2. REVIEW OF THE LITERATURE

2.1 Carbohydrates from lactic acid bacteria

Several LAB produce polysaccharides that are components of the cell wall or are released from the cell. The latter can be divided into capsular polysaccharides (CPS) and EPS (Zannini et al., 2015). Whereas CPS are permanently attached to the cell surface in the form of capsules, EPS are secreted into the environment (Ruas-Madiedo et al., 2009). EPS are involved in biofilm formation and have various physiological functions, such as in surface adhesion and the protection of bacterial cells from environmental stress (Zannini et al., 2015). EPS can be divided into two groups based on the mechanism of biosynthesis and the precursors required (Boels et al., 2001). One group comprises homopolysaccharides (HoPS), namely α-glucans and β-fructans, which are synthesized extracellularly by glucansucrases (glycoside hydrolase family 70) and fructansucrases (glycoside hydrolase family 68), respectively. Both enzymes use sucrose as a substrate, while the latter can also use, e.g., raffinose to synthesize β-fructan in the absence of sucrose. The other group includes β-glucans (HoPS) and heteropolysaccharides (HePS). Their synthesis requires sugar nucleotide precursors and several enzymes that are involved in central carbon metabolism and in the synthesis of some cell-wall components (Zannini et al., 2015).

Sucrose-derived HoPS are synthesized extracellularly by the action of glycansucrases of certain LAB, such as Leuconostoc, Lactobacillus, Streptococcus, and Weissella (Zannini et al., 2015). One bacterium may harbor multiple glucansucrase and fructansucrase genes (Olvera et al., 2007; Malik et al., 2009; Malang et al., 2015). When a strain secretes multiple glucansucrases, these enzymes may act jointly to synthesize a hybrid polymer that is different from that produced by purified enzyme (Côté and Skory 2015). The hydrolysis of the glycosidic linkage of sucrose provides these glycansucrases with the energy to form a new glycosidic linkage in corresponding HoPS (Zannini et al., 2015). Because the biosynthesis of these HoPS is less energy demanding than that of HePS, they can be produced in larger amounts. According to the type of glucosidic linkages, α- glucans are classified into dextrans, mutans, alternans, and reuterans (Leemhuis et al., 2013). Dextrans contain mainly α-(1→6) linkages and α-(1→2), α-(1→3), or α-(1→4) branch linkages, mutans mainly α-(1→3) linkages, reuterans mainly α-(1→4) linkages, and alternans contain alternating α-(1→6) and α-(1→3) linkages (Naessens et al., 2005; Bounaix et al., 2009). β-Fructans are categorized into levans, with mainly β-(2→6) linkages, and inulins, with mainly β-(2→1) linkages (Zannini et al., 2015). In addition to polymeric glucans and fructans, LAB can also synthesize oligosaccharides, such as

11 isomaltooligosaccharides and fructooligosaccharides. The size and yield of carbohydrate products depend on the strain and fermentation conditions (Jeanes et al., 1954; Schwab et al., 2008; Anwar et al., 2010; Malang et al., 2015).

2.2 Glucansucrases

Because of the high interest in the carbohydrates of LAB origin for food and non-food applications, their synthesis machineries have been extensively explored, and a large number of synthesizing enzymes have been identified and characterized. Glucansucrases (EC 2.4.1.5) are produced extracellularly by LAB to synthesize α-glucans (Leemhuis et al., 2013). They are named based on the α-glucans that they synthesize, i.e., dextransucrases, mutansucrases, alternansucrases and reuteransucrases. Although glucansucrases are transferases with respect to their activity, they are structurally close to glycoside hydrolases and are classified as part of the glycoside hydrolase family 70 (GH70), which contains mainly glucansucrases (Leemhuis et al., 2013). The recently characterized 4,6-α-glucanotransferases, which are inactive on sucrose but convert starch or starch hydrolysates into isomalto/maltopolysaccharides, constitute a new subfamily of GH70 (Bai et al., 2015). Moreover, by sequencing the genomes of the producers of original α-glucans, a cluster of branching sucrases specialized in dextran grafting through α-(1→2) or α-(1→3) linkages have been recently discovered (Passerini et al., 2015; Vuillemin et al., 2016). GH70 enzymes have been found almost exclusively in LAB according to the entries in the Carbohydrate-Active Enzymes (CAZy) database (Lombard et al., 2014). So far, the encoding genes of more than three hundred GH70 enzymes have been cloned and sequenced, and the three-dimensional structures of four truncated enzymes have been solved (Vujicic-Zagar et al., 2010; Ito et al., 2011; Brison et al., 2012; Pijning et al., 2012). The GH70 enzymes are mechanistically, structurally, and evolutionary closely related to GH13 (α-amylases) and GH77 (amylomaltases) enzymes. The three families containing a catalytic (β/α)8-barrel domain form the GH-H clan (Stam et al., 2006).

2.2.1 Catalytic mechanism of glucansucrases The catalysis of glucansucrases is proposed to be a two-step process, starting with the cleavage of the substrate sucrose and followed by the formation of a covalent β-glucosyl- enzyme intermediate and the release of a free fructose (Figure 1). In the second step, the glucosyl moiety is transferred to an acceptor, retaining the α-anomeric configuration (Leemhuis et al., 2013). Hydrolysis takes place when a water molecule acts as the acceptor. In the presence of suitable low-molar-mass molecules, e.g., maltose,

12 glucansucrases catalyze so-called acceptor reactions, where the glucosyl residue from sucrose is diverted from α-glucan formation and transferred onto the small molecule instead (Figure 1). Di- and oligosaccharides are formed when carbohydrates act as the acceptor molecules, whereas glucoconjugates are formed in the presence of non- carbohydrate compounds (Monsan et al., 2010).

H2O hydrolysis

fructose R-OH acceptor reaction covalent glucosyl- enzyme intermediate

dextran synthesis

Figure 1. Types of reactions catalyzed by dextransucrases.

The three-dimensional structures of truncated glucansucrases revealed that they exhibit a U-fold shape, organized into five (A, B, C, IV, and V) domains (Vujicic-Zagar et al., 2010; Ito et al., 2011; Brison et al., 2012; Pijning et al., 2012). Except for domain C at the base of the U-fold, other domains are formed by discontiguous N- and C-terminal segments of the polypeptide, as illustrated by the glucansucrase of Lactobacillus reuteri 180 with its N-terminal end (~740 residues) truncated (Figure 2). Domains A, B, and C resemble family GH13 enzymes, whereas domains IV and V are unique in GH70 enzymes. The catalytic domain A adopts a (β/α)8-barrel fold and harbors a catalytic triad, which is composed of two aspartates (acting separately as a nucleophile and a transition- state stabilizer) and one glutamate (acting as an acid/base catalyst) (Leemhuis et al., 2013). The acceptor binding site is shaped by residues from domains A and B. Mutagenesis studies have shown that the interplay of these residues determines how a growing glucan chain interacts with the acceptor binding site (which hydroxyl group of a glucosyl residue is capable of attacking the glucosyl-enzyme intermediate to form a glycosidic linkage) and thus the proportion of specific linkages in the glucan synthesized

13 (Irague et al., 2013; Leemhuis et al., 2013; Meng et al., 2015a; Meng et al., 2015c). Domain V is part of the glucan binding domain consisting of N- and C-terminal segments, while domain IV may provide a hinge that brings the glucan-bound domain V toward or away from the catalytic site to facilitate the synthesis (Leemhuis et al., 2013; Meng et al., 2015b).

Figure 2. Crystal structure of an N-terminus-truncated dextransucrase from Lactobacillus reuteri 180 and schematic presentation of the U-fold shape of the polypeptide chain (adapted from Vujicic-Zagar et al., 2010).

Moulis and others (2006b) proposed a semi-processive mechanism for the polymerization of glucans. There is a lag-phase before glucan synthesis starts, when sucrose and glucose (produced by hydrolysis) act as initiators of polymerization. Glucose is the preferred initiator when present in sufficient amounts. Elongation occurs through the non-processive (the product is released after each elongation step of the acceptor) attachment of glucosyl moieties to the nonreducing end of the initially formed products. The longer oligosaccharides are better acceptors than are sucrose, fructose, and glucose, ensuring that the glucosyl moieties are used to synthesize glucans instead of large amounts of oligosaccharides. Once a glucan chain reaches a minimal size, it may remain attached to the enzyme, and the elongation becomes processive. The glucan- binding domain is involved in the processive mechanism, as shown by a gradual loss of elongation efficiency (a decrease in the ratio of high-/low-molar-mass population) as the domain is progressively deleted (Moulis et al., 2006b).

14 2.2.2 Production and characterization of dextransucrases Among LAB, the expression of the dextransucrase-encoding gene is usually sucrose- inducible in wild-type strains of Leuconostoc spp. but constitutive in, e.g., Streptococcus spp. and some Lactobacillus strains (Naessens et al., 2005; Bounaix et al., 2010). The constitutive expression of the glucansucrase-encoding gene is suggested to be advantageous in ecosystems with fluctuating sucrose concentrations, such as the oral cavity (Leemhuis et al., 2013). Dextransucrase production is affected by temperature, aeration and medium composition, and methods such as response surface methodology have been used to optimize these culture conditions (Purama and Goyal 2008a). Native dextransucrases from different LAB have been purified by, e.g., salt, alcohol precipitation, polyethylene glycol (PEG) fractionation, ultrafiltration, and chromatography (Naessens et al., 2005; Purama and Goyal 2008b). PEG fractionation can separate dextransucrases from contaminating enzymes, such as levansucrase; however, dextran remains in the preparation (Majumder et al., 2007). Because the induced production of dextransucrase leads to the concomitant production of dextran, which can envelop the enzyme and hinder the enzyme characterization (Moulis et al., 2006a), it is desired to produce dextransucrases by the constitutive mutants or heterologous hosts, where the expression of the dextransucrase-encoding gene does not require sucrose induction and therefore no dextran is produced concomitantly. The dextran present, however, stabilizes dextransucrase, and so does PEG and Tween 80 (Majumder et al., 2007). The presence of Ca2+ is essential for dextransucrase activity (Naessens et al., 2005). The Ca2+-binding site is located at the interface between domains A and B (Figure 2). The absence of Ca2+ may lead to a compromised substrate binding and thus a decreased activity (Vujicic-Zagar et al., 2010). Dextransucrases normally have a pH optimum of 5.0‒5.5 and a Michaelis-Menten constant (KM) for sucrose of approximately 12‒16 mM (Naessens et al., 2005).

Several assays have been used for dextransucrase activity measurement (Vettori et al., 2011). They can be divided into two groups. One group is based on the measurement of the fructose released, which represents sucrose consumption (from both hydrolytic and transferase reactions). The hydrolytic activity, which is usually low, can be determined by quantifying the free glucose released. Reducing-value assays are based on an increase in the reducing value in the reaction mixture (mostly resulting from the fructose released in dextran formation) and are normally used for dextransucrase activity assays, although the dextran product itself and small amounts of side products, i.e., leucrose (α-D-Glcp- (1→5)-D-Frup, an acceptor product of fructose), glucose and fructose resulting from hydrolysis, also contribute to the increased reducing value. The commonly used reducing-value assays include the 3,5-dinitrosalicylic acid (DNS) method (Sumner and

15 Howell 1935) and the Nelson-Somogyi method (Nelson 1944; Somogyi 1945). The other assay group is based on the quantification of dextran product by, e.g., its weight. The 14C-sucrose radioisotope method has also been used (Vettori et al., 2011; Côté and Skory 2012) and is based on the incorporation of 14C-labeled glucose from 14C-U- labeled sucrose to the growing dextran chain. The second group provides a more accurate measurement of dextransucrase activity (Germaine et al., 1974; Vettori et al., 2011).

2.3 Carbohydrates synthesized by dextransucrases

Glucansucrases are useful synthetic tools for various carbohydrates and glucoconjugates because of their high regioselectivity in glucosyl transfer reactions, relaxed specificity toward a variety of acceptors, and low cost of the donor substrate sucrose (Seibel et al., 2010). Moreover, glucansucrases have been used to improve the flavor, water solubility, and oxidative stability of various compounds for food, cosmetic, and therapeutic applications (Côté et al., 2009; Kim et al., 2012; Liang et al., 2016).

2.3.1 Substrate and product specificity of dextransucrases Although glucansucrases are highly specific towards sucrose, certain other carbohydrates can also serve as glycosyl donors (Jung and Mayer 1981; Seto et al., 2008; Timm et al., 2013). Especially, the transfer of a different glycosyl residue than a glucosyl one indicates broader applications of glucansucrases in carbohydrate bioconversion. For example, carbonyl-group-containing dextran was synthesized using keto-sucrose at C-2 by dextransucrase (Seto et al., 2008). Similarly, a reuteransucrase was used to transfer an allosyl (differs in C3 configuration from glucosyl) group to several acceptors with α- D-allopyranosyl β-D-fructofuranoside as a donor substrate (Timm et al., 2013).

The binding kinetics between Lc. mesenteroides dextransucrase and immobilized glucose, maltose, and dextran were determined using a quartz crystal microbalance (Nihira et al., 2011; Fang et al., 2015). The dissociation constants for glucose, maltose, and dextran were 64, 35, and 18 nM, respectively, indicating an increasing order of specificity between the enzyme and these acceptors. The effectiveness of different acceptors was compared for Lc. mesenteroides dextransucrases at a sucrose/acceptor ratio of 1:1 (Robyt and Eklund 1983; Kothari and Goyal 2013). Among the acceptor sugars tested in both studies, maltose and, to a lesser extent, isomaltose were the best acceptors; D-glucose, lactose, and cellobiose were moderately effective; and raffinose, D-galactose, and D-xylose were poor acceptors. The more effective acceptors can

16 generally be further glucosylated into a series of oligosaccharides and suppress dextran formation to a higher degree (Robyt and Eklund 1983; Kothari and Goyal 2013).

Dextransucrases are able to form mainly α-(1→6) linkages and, to a lesser extent, α- (1→2), α-(1→3), or α-(1→4) linkages (Leemhuis et al., 2013). It has been proposed that the linkage specificity depends on the orientation, in which the acceptor is guided toward the reaction center and is therefore determined by the conformation of the acceptor binding cleft (Leemhuis et al., 2013). Amino acid residues forming acceptor binding subsites affect dextransucrase specificity, as mutations of targeted residues were shown to alter the linkage composition of the product (Brison et al., 2012; Leemhuis et al., 2013; Meng et al., 2015c). Many dextransucrases catalyze the formation of more than one type of glucosidic linkage, requiring different binding modes of an acceptor. Thus, the acceptor binding cleft needs to be wide enough to accommodate the acceptor in different orientations. For example, the structure of a truncated dextransucrase from Lactobacillus reuteri 180, which forms both α-(1→6) and α-(1→3) linkages, shows a rather wide binding cleft. However, a unique branching sucrase derived from Leuconostoc citreum NRRL B-1299 (formerly known as Leuconostoc mesenteroides NRRL B-1299) dextransucrase only has α-(1→2) branching activity, and correspondingly, its binding cleft is partially blocked, allowing only one acceptor binding mode (Brison et al., 2012; Leemhuis et al., 2013).

In acceptor reactions, the product pattern for a given acceptor depends on dextransucrase specificity. Maltose is the most studied acceptor due to its high effectiveness, and the patterns of oligosaccharide products of maltose acceptor reactions corresponded to the structure of the dextran produced by the same strain (Côté and Leathers 2005). For example, the typical dextransucrases that predominantly synthesize α-(1→6) linkages, represented by the commercially used Leuconostoc mesenteroides NRRL B-512F dextransucrase (DSR-S), produce homologous oligosaccharides in the presence of maltose (Robyt and Eklund 1983). The initial product, panose (α-D-Glcp-(1→6)-α-D- Glcp-(1→4)-D-Glcp), is elongated by the successive attachment of glucosyl residues to the 6-OH of the nonreducing end glucose, forming linear isomaltooligosaccharides (IMOs), as illustrated by series A in Figure 3. For Lc. citreum NRRL B-1299 dextransucrase synthesizing α-(1→2)-branched dextran, however, the maltose acceptor products are three homologous IMO series, with the major series constituting linear IMOs of series A (Figure 3). The two minor series (marked B and C in Figure 3) have α-(1→6)-linked main chains with a single-unit branch α-(1→2) attached to the terminal and the penultimate α-(1→6)-linked residue of the nonreducing end, respectively (Dols et al., 1998).

17 Figure 3. Oligosaccharide series (named A, B, and C) produced in the maltose acceptor reaction of dextransucrases. Series A are produced by Lc. mesenteroides NRRL B-512F and Lc. citreum NRRL B-1299 dextransucrases, whereas series B and C are only produced by B-1299 dextransucrase (adapted from Dols et al., 1998).

In addition to this thesis work, GLOS products from different acceptors (Table 1) have only been characterized for DSR-S from Lc. mesenteroides B-512F (Yamauchi and Ohwada 1969; Robyt 1995; Demuth et al., 2002; Côté et al., 2009; Díez-Municio et al., 2012a; Barea-Alvarez et al., 2014; Côté et al., 2014). The type of linkage that is formed is a function of the acceptor structure. Surprisingly, α-(1→2)-linkage is formed on lactose and cellobiose, even though this enzyme is known to synthesize α-(1→6) and α- (1→3) linkages in dextran. In addition to 2Glc-α-D-Glcp-lactose, an unknown trisaccharide has been reported for the lactose acceptor reaction (Díez-Municio et al., 2012b). Controversial results have been reported regarding the main cellobiose product. Instead of 21-α-D-Glcp-cellobiose, the enzyme from a dextransucrase-hyper-producing Lc. mesenteroides B-512F constitutive mutant was reported to produce 22-α-D-Glcp- cellobiose (Kim et al., 2010). However, both the wild-type and mutant enzymes formed a minor trisaccharide: linear 62-α-D-Glcp-cellobiose (Argüello Morales et al., 2001; Kim et al., 2010).

18 Table 1. Acceptor products formed by DSR-S from Lc. mesenteroides NRRL B-512F. Acceptor First acceptor producta Reference Maltose (α-D-Glcp-(1→4)-D-Glcp) Panose* Robyt 1995 Isomaltose (α-D-Glcp-(1→6)-D-Glcp) Isomaltotriose* " Nigerose (α-D-Glcp-(1→3)-D-Glcp) 62-α-D-Glcp-nigerose* " Methyl α-D-glucopyranoside Methyl α-isomaltoside* " D-Glucose Isomaltose* " Lactose (β-D-Galp-(1→4)-D-Glcp) 2Glc-α-D-Glcp-lactose " Cellobiose (β-D-Glcp-(1→4)-D-Glcp) 21-α-D-Glcp-cellobiose* " D-Fructose Leucrose " (α-D-Glcp-(1→5)-D-Frup) Isomaltulose (α-D-Glcp-(1→6)-D-Fruf) D-Galactose α-D-Glcp-(1→1)-β-D-Galf " Maltotriose 63-α-D-Glcp-maltotriose* " 61-α-D-Glcp-maltotriose Maltotetraose 64-α-D-Glcp-maltotetraose* " 61-α-D-Glcp-maltotetraose Gentiobiose (β-D-Glcp-(1→6)-D-Glcp) 62-α-D-Glcp-gentiobiose Yamauchi and Ohwada 1969 D-Mannose α-D-Glcp-(1→6)-D-Manp Côté et al., 2014 α-D-Glcp-(1→1)-β-D- Robyt 1995 Manp Raffinose 4Gal-α-D-Glcp-raffinose Côté et al., 2009 (α-D-Galp-(1→6)-α-D-Glcp- (1→2)-β-D-Fruf) Isomaltulose (α-D-Glcp-(1→6)-D-Fruf) 62-α-D-Glcp-isomaltulose* Barea-Alvarez et al., 2014 Lactulose (β-D-Galp-(1→4)-D-Fruf) β-D-Galp-(1→4)-β-D-Fruf- Díez-Municio et al., 2012b (2→1)-α-D-Glcp Maltulose (α-D-Glcp-(1→4)-D-Fruf) α-D-Glcp-(1→4)-β-D-Fruf- Demuth et al., 2002 (2→1)-α-D-Glcp a The starred product is the first member of a homologous series with α-isomalto- oligosaccharides attached to the acceptor.

2.3.2 Functional oligosaccharides from dextransucrases Glucansucrase-catalyzed synthesis provides access to a large range of bioactive glucooligosaccharides (GLOSs), particularly non-digestible oligosaccharides with prebiotic effects (Monsan et al., 2010). The use of prebiotics is gaining popularity because, compared to probiotics, they are resistant to the aggressive gastro-intestinal environment and cheaper and easier to incorporate into the diet (Goffin et al., 2011; Al-

19 Sheraji et al., 2013). Prebiotics are currently defined as the dietary components that selectively stimulate the growth and/or activity(ies) of one or a limited number of microbial genus(era)/species in the gut microbiota and thus confer health benefits to the host (Roberfroid et al., 2010). The acceptor reactions of glucansucrases, especially dextransucrases, from Leuconostoc spp. have been used to synthesize functional GLOSs (Chung and Day 2002; Seo DM et al., 2007; Holt et al., 2008; Goffin et al., 2011; Kothari and Goyal 2015). The GLOSs derived from maltose, lactose, lactulose, and cellobiose by DSR-S from Lc. mesenteroides NRRL B-512F have shown prebiotic potential in vitro and for maltose products also in vivo (Argüello Morales et al., 2001; Ruiz-Matute et al., 2011; Díez-Municio et al., 2012b; García-Cayuela et al., 2014). GLOS products of Leuconostoc dextransucrases have also shown other health benefits, e.g., preventing the installation of type 2 diabetes, preventing dental caries, and inhibiting cancer cells (Kim et al., 2010; Monsan et al., 2010; Kothari and Goyal 2015).

Among the GLOSs from dextransucrases, maltose-derived isomaltooligosaccharides (IMOs) represent the most commercialized prebiotic products. In particular, the IMOs containing α-(1→2)-linkages from Lc. citreum NRRL B-1299 dextransucrase are marketed as prebiotics for their low digestibility and produced on an industrial scale of 40 metric tons/year (Plou et al., 2002). The linear IMOs with higher degrees of polymerization (DPs) also demonstrated prebiotic effects in humans and animals (Goffin et al., 2011; Yen et al., 2011; Ketabi et al., 2011). In fact, IMOs of DP 3.3 were fermented in the large intestine of rats, while IMOs of DP 2.7 were mainly digested and absorbed in the small intestine, without reaching the large intestine (Iwaya et al., 2012). However, commercial IMOs usually contain a high proportion of digestible carbohydrates, which decreases their functionality as prebiotics (Hu et al., 2013). The industrial IMO production includes the enzymatic hydrolysis of starch into high maltose syrup, followed by the action of α-transglucosidase, which catalyzes the transfer of the nonreducing glucosyl residue of maltose to the 6-OH group of glucose (released by hydrolysis), the nonreducing glucosyl unit of maltose or any α-glucooligosaccharide present in the solution (Goffin et al., 2011). Isomaltose, panose and isomaltotriose are the main components in the IMOs thus produced. The dextransucrase acceptor reaction holds promise for the size-controlled synthesis of IMOs by varying concentrations of sucrose (donor) and maltose (acceptor). With an increase in the sucrose/maltose ratio, the DP of both the major and the longest products increased (Su and Robyt 1993; Lee et al., 1997; Lee et al., 2008). However, detailed studies are needed to select the optimal synthesis conditions for IMOs of higher DP.

20 2.3.3 Characterization of dextransucrase-synthesized carbohydrates Knowledge of the structure of dextransucrase-synthesized products is essential for the application of enzymes and the dextran and oligosaccharide products. Typical strategies for the structural characterization of dextrans and oligosaccharides are presented in Figure 4. Extracellular water-soluble glucans can simply be purified by the solvent (normally ethanol) precipitation of the cell-free culture and dialysis against water when no severe interfering compound, e.g., protein is present (Ruas-Madiedo and de los Reyes-Gavilán 2005). The molar mass of dextrans can be determined by, e.g., ultracentrifugation, high-performance size-exclusion chromatography (HPSEC), and asymmetric flow field-flow fractionation (AsFlFFF) (Irague et al., 2012; Maina et al., 2014). HPSEC is widely used to analyze the molar mass distribution based on the separation of molecules by hydrodynamic radius (Pitkänen 2011). Weight-average molar mass (Mw) can be determined absolutely by HPSEC using universal calibration (applying refractive index and viscosity detectors) or static light scattering detection. A 6 9 wide range of Mw (10 ‒10 g/mol) has been reported for native dextrans; however, discrepancy can arise from the different analysis methods applied (Maina 2012). DMSO was reported to be the solvent of choice for dextran Mw determination by HPSEC, as aggregate formation was less prevalent in DMSO than in aqueous solution. Moreover, the degradation of dextrans due to shear forces during HPSEC analysis was not significant in DMSO (Maina et al., 2014). High-molar-mass dextrans with low-branched structures have applications in food industries as hydrocolloids (Lacaze et al., 2007; Waldherr and Vogel 2009).

The monosaccharide compositions of polysaccharides can be analyzed after total depolymerization by acid methanolysis or acid hydrolysis (Leemhuis et al., 2013). The monomers released are analyzed by gas chromatography (GC) or high-performance liquid chromatography (HPLC). High-performance anion-exchange chromatography with pulsed amperometric detection (HPAEC-PAD) is one of the most frequently used techniques for mono- and oligosaccharide analysis. It exempts derivatization and affords sensitive detection. Carbohydrates are readily separated by HPAEC under alkaline conditions, where retention increases with decreasing pKa (Corradini et al., 2012). For structural elucidation, large α-glucans often need to be degraded into the constituting oligosaccharides by partial acid hydrolysis or specific enzymes. The fragments are then analyzed by, e.g., thin-layer chromatography, HPLC, or NMR spectroscopy, and the structure and abundance of these building blocks are used to build a composite polymer structure (Remaud-Slmeon et al., 1994; van Leeuwen et al., 2008b; Vettori et al., 2012). Enzyme-aided methods are useful for fingerprinting the type of branch linkages in different dextrans (Sawai et al., 1978; Maina 2012). One method involves the specific

21 hydrolysis of dextran with a mixture of dextranase and α-glucosidase. The chromatographic profiles of enzyme-resistant oligosaccharides vary depending on the branch points of dextran and can therefore be used as structural markers (Maina 2012).

Glucansucrase or LAB

Partial hydrolysis Glucan Oligosaccharides

Mw determination Profiling by e.g. by e.g. HPSEC HPAEC-PAD

Detailed information Isolated by NMR spectroscopy oligosaccharides

Monosaccharide Linkage sequence Linkage ratio by composition by ESI-MSn methylation Figure 4. Commonly used strategies and techniques for the structural analysis of products from glucansucrases.

The position of linkages between monosaccharide units in oligo- and polysaccharide can be determined by methylation analysis. In this method, all of the free hydroxyl groups in a carbohydrate are tagged by methylation prior to hydrolysis, after which the monomers are reduced and derivatized for gas chromatography-mass spectrometry (GC- MS) analysis (Naessens et al., 2005). Nuclear magnetic resonance (NMR) spectroscopy is the most powerful technique for the structure elucidation of oligosaccharides and has been intensively used for characterizing glucansucrase-synthesized carbohydrates (Côté and Skory 2012; Díez-Municio et al., 2012a; Barea-Alvarez et al., 2014). It can provide all of the information about the type of constituent monosaccharides, including ring size and anomeric configuration, and the sequence of glycosidic linkages (Bubb 2003). Combinations of two-dimensional 1H and 13C techniques have been developed, such as homonuclear (1H-1H) COSY and TOCSY experiments for proton chemical shifts of each residue and heteronuclear (1H-13C) HSQC and HMBC experiments for substitution information (Bubb 2003). The primary structures of α-glucans have been characterized based on the 1H NMR structural-reporter-group concept (van Leeuwen et al., 2008a; van Leeuwen et al., 2008b).

22 Multistage mass spectrometry (MSn) has proven valuable in the sequential analysis of the glycosidic linkages in oligosaccharides (Usui et al., 2009; Mischnick 2012; Maina et al., 2013). Matrix-assisted laser desorption/ionization (MALDI) and electrospray ionization (ESI) are the main ionization techniques used, with the latter coupling MS with liquid chromatography (Harvey 2012). In negative ion MSn, C-ions (Figure 5) are predominantly produced due to glycosidic cleavage, whereas isomeric C- and Y-ions are produced in positive ion MSn. Negative ion MSn is preferably used for the structural analysis of an underivatized linear oligosaccharide (Maina et al., 2013). In each MSn stage, the cross-ring cleavages at the reducing end residue of the isolated C-ion are diagnostic of the reducing end linkage. The linkage sequence of an unknown oligosaccharide can be deduced by comparing its fragmentation behaviors in MSn against a library of model compounds (Mischnick 2012). Unlike NMR spectroscopy analysis, which requires purified carbohydrates, MS coupled with liquid chromatography provides an integrated strategy for the rapid characterization of oligosaccharides in mixtures (Wuhrer et al., 2009).

Figure 5. Nomenclature of fragment ions of cellotriose according to Domon and Costello (1988).

2.4 Weissella spp. and Weissella dextransucrases

The genus Weissella was proposed in 1993 after the reclassification of some Leuconostoc-like bacteria (Collins et al., 1993). Weissella are Gram-positive, catalase- negative, non-endospore-forming, coccoid or rod-shaped heterofermentative LAB from the family Leuconostocaceae (Fusco et al., 2015). Dextran production is a phenotypic criterion for the identification of bacteria in this genus. Weissella strains have been found in a wide range of habitats, such as plants and vegetables and the gastrointestinal tract of animals and humans (Fusco et al., 2015). Some strains of W. confusa and W. cibaria species have been suggested to be probiotics (Ayeni et al., 2011; Lee et al., 2012). Although W. confusa has rarely been described as a human opportunistic pathogen, infections usually occur in patients with immunosuppression or underlying diseases (Fusco et al., 2015).

23 Weissella spp. are among the most frequently occurring LAB in raw and spontaneously fermented cereals, vegetables, and fruits. They have adapted to the specific plant matrix and usually lead to desired and predictable changes in the shelf life, nutritional and sensory properties of the matrix (Di Cagno et al., 2013). In a study to modify the texture of pureed carrots by fermentation with EPS-producing LAB, dextran-producing W. confusa VTT E-90392, which was initially isolated from a sour carrot mash, conferred the highest viscosity and desirable sensory profile (Juvonen et al., 2015). The highly nutritious barley malt extract was fermented by W. cibaria MG1, resulting in a functional alcohol-free beverage with desirable textural properties (Zannini et al., 2013).

Sourdough is a cocktail of LAB and yeasts originating from wheat or rye flour and is used traditionally as a leavening agent in baking (Di Cagno et al., 2006). As common sourdough isolates, the closely related W. confusa and W. cibaria are promising in starting sourdoughs (Di Cagno et al., 2006; Katina et al., 2009; Galle et al., 2010). In wheat sourdough, W. confusa VTT E-90392 synthesized a high yield of dextran, whereas the commercial dextran-producing Leuconostoc mesenteroides NRRL B-512F formed almost none under the same conditions. Unlike many Leuconostoc spp., Weissella spp. produce dextran without extensive acidification, as they do not convert fructose into mannitol, with concomitant acetate production (Galle et al., 2010). Therefore, the technological benefits of Weissella dextran are not overridden by high acidity in sourdough bread baking. The in situ-produced dextran improved the moisture retention, dough rheology, bread volume and crumb softness and retarded staling (Di Cagno et al., 2006; Schwab et al., 2008; Katina et al., 2009; Galle et al., 2010; Galle et al., 2012). Moreover, W. confusa and W. cibaria also produced a significant amount of IMOs in sourdoughs and thus provided extra health benefits (Schwab et al., 2008; Katina et al., 2009; Galle et al., 2010). The amounts of IMOs and dextran in a W. confusa VTT E- 90392 sourdough were estimated to be 4.0% and 1.8% (both dry weight), respectively, produced from 10% (dry weight) added sucrose (Maina 2012). The supplementation of sucrose for Weissella dextran production also leads to free fructose production in sourdough (5% fructose produced from 10% added sucrose), which does not comply to the general diet advised to reduce sugar consumption. However, fructose leads to a lower blood glucose increase compared to sucrose or glucose (European Food Safety Authority (EFSA) 2011) and thus has a relatively small health implication.

Dextransucrases from Weissella spp. may also find various uses, e.g., to modify food structure without concomitant acidification or for safety considerations compared to whole-cell fermentation. The crude dextransucrase from pear-derived Weissella sp. TN610 induced the solidification of sucrose-supplemented milk and was suggested as a promising, safe food additive (Bejar et al., 2013). Several W. cibaria and W. confusa

24 isolates from sourdoughs were found to produce constitutive soluble dextransucrases, as the enzyme activities were always higher when they grew on glucose than on sucrose (Bounaix et al., 2010). However, for W. confusa Cab3 isolated from fermented cabbage, a strain related to this thesis work, dextransucrase production was boosted with sucrose (Shukla and Goyal 2011). In addition to this thesis, the biochemical and kinetic properties of dextransucrases from two other Weissella strains have been recently studied, with one being a native enzyme and the other being a recombinant one (Amari et al., 2012; Rao and Goyal 2013). The gene encoding the latter (of strain LBAE C39-2) was cloned and expressed in E. coli. An amino acid sequence alignment indicated that Weissella dextransucrases form a distinct phylogenetic group within glucansucrases (Amari et al., 2012). The two enzymes exhibited similar molar masses (approximately 180 kDa), pH optima (pH 5.4), thermal stabilities (inactivated rapidly above 35 °C), and

Michaelis-Menten kinetic parameters (KM and Vmax values of 8.6–13 mM and 20–27.5 µmol/(mg∙min), respectively), which were also comparable to the properties reported for Leuconostoc dextransucrases (Naessens et al., 2005; Côté and Skory 2012). Of note, the activity and thus the kinetic values of both enzymes were measured by conventional reducing-value assays.

Dextrans from Weissella spp. share highly similar structures, normally consisting of predominantly α-(1→6) linkages and only a few α-(1→3) branch linkages (<5%) (Bounaix et al., 2009; Maina et al., 2011; Ahmed et al., 2012; Bejar et al., 2013). Weissella dextrans show high molar masses of >106 g/mol (Ahmed et al., 2012; Maina et al., 2014). In fact, the technological benefits of dextrans in sourdoughs are attributed to their low degree of branching and high molar mass (Lacaze et al., 2007). Other than the GLOSs formed in the presence of maltose (Katina et al., 2009), the acceptor products of Weissella dextransucrases have not yet been investigated. Weissella spp. may form, simultaneously with dextrans, byproducts from, e.g., lactose acceptor reaction when endogenous sugars are present in food matrices. Furthermore, the enzymes may provide a new opportunity for the biosynthesis of GLOSs of industrial interest via acceptor reactions.

25 3. AIM OF THE STUDY

The overall aim of the thesis was to study the function and applicability of Weissella dextransucrases, especially in the synthesis of glucooligosaccharides. Dextransucrases from two efficient dextran-producing W. confusa strains were studied, namely, a native enzyme from W. confusa Cab3 (WcCab3-DSR) and a recombinant one from W. confusa VTT E-90392 (WcE392-rDSR), as no W. confusa dextransucrase had been studied in detail by the time that this work started.

The specific objectives were to:

· study the biochemical and kinetic properties of WcE392-rDSR and compare a reducing-value assay with a 14C-sucrose radioisotope assay to determine these parameters (article I)

· structurally characterize the dextrans synthesized by the enzymes (articles I and II)

· explore the selectivity of Weissella dextransucrase towards various acceptor sugars and structurally characterize the selected GLOSs (articles II‒IV)

· apply Weissella dextransucrase for the synthesis of prebiotic IMOs (article IV)

26 4. MATERIALS AND METHODS

4.1 Dextransucrase preparations and commercial enzymes

Sources of Weissella dextransucrases and other enzymes used in this work are presented in Table 2. The Weissella preparations are described in articles I and II. Briefly, the gene encoding dextransucrase of W. confusa VTT E-90392 was cloned into Lactococcus lactis NZ9800. The resultant recombinant enzyme WcE392-rDSR was purified by nickel affinity chromatography or partially purified (the crude form) by ultrafiltering the culture supernatant (article I). W. confusa Cab3 was previously isolated from fermented cabbage, and WcCab3-DSR was partially purified by PEG fractionation (article II).

Table 2. Dextransucrase preparations and commercial enzymes used in the study. Type Name Source Article Dextran- Purified recombinant dextransucrase from VTT Technical Research I sucrase W. confusa E-90392 (WcE392-rDSR) Centre of Finland Ltd. A crude preparation of WcE392-rDSR VTT I, III, IV Native dextransucrase from W. confusa Indian Institute of II, IV Cab3 (WcCab3-DSR) Technology Guwahati Commercial dextransucrase from Sigma-Aldrich, MO IV Leuconostoc mesenteroides Hydrolytic Chaetomium erraticum dextranase Sigma-Aldrich II enzyme Aspergillus niger α-glucosidase Megazyme, Ireland II, III (E-TRNGL) Bacillus stearothermophilus α- Megazyme II glucosidase (E-TSAGL) A. niger β-galactosidase Megazyme III Agrobacterium sp. β-glucosidase Megazyme III Yeast invertase (β-fructosidase) Megazyme III

4.2 Biochemical characterization of WcE392-rDSR (I)

The protein concentration of the WcE392-rDSR preparation was determined by the DC Protein Assay Kit (Bio-Rad, CA) with bovine serum albumin as the standard. The activity of WcE392-rDSR was assayed by the 14C-sucrose radioisotope method and the Nelson-Somogyi method (Nelson 1944; Somogyi 1945; Côté and Skory 2012). One activity unit was defined as the amount of enzyme that catalyzed the incorporation of 1

27 µmol of D-glucose into dextran and the formation of 1 µmol of reducing sugar in 1 min, respectively.

The Nelson-Somogyi method was used to determine the pH optimum, the effect of temperature on the activity, and the thermal stability of WcE392-rDSR. Kinetic parameters were determined by both the 14C-sucrose radioisotope and Nelson-Somogyi methods, with the sucrose concentration ranging from 3 to 146 mM. The initial velocity data were analyzed by nonlinear regression using GraphPad Prism 6 (GraphPad, CA).

4.3 Analysis of monosaccharides and oligosaccharides

HPAEC-PAD was used to analyze mono- and oligosaccharides. Prior to analysis, the samples were filtered through a 10 kDa Amicon Ultra-0.5 centrifugal filter (Millipore, MA) or a 0.45 μm membrane (Pall Corporation, NY). The instrument (unless otherwise specified) was equipped with a CarboPac PA-100 column (250 × 4 mm, i.d, Dionex, CA), a Decade detector (Antec Leyden, The Netherlands), a Waters 717 autosampler, and two Waters 515 pumps. For a qualitative analysis of the oligosaccharide profiles, the elution was performed with 100 mM NaOH (15 min), followed by a gradient to 120 mM Na-acetate in 100 mM NaOH (20 min) with a flow rate of 1 ml/min.

For the quantification of mono- and oligosaccharides, the same system was run by another elution method, starting with 75 mM of NaOH (8 min) and followed by a gradient to 100 mM of Na-acetate in 75 mM of NaOH (27 min) with a flow rate of 1 ml/min. In the case of the co-elution of lactose and sucrose, another HPAEC-PAD system was used in article III. Glucose, fructose, galactose, maltose, sucrose, lactose, cellobiose (Merck, Germany), leucrose (Santa Cruz Biotechnology, CA), isomaltose, isomaltotriose, panose (TCI Europe nv, Belgium) and kojibiose (Carbosynth, UK) were used as standards. Panose was also used as a standard for the quantification of total IMO products (by integrating the total peak area) and individual IMOs of DP4‒6 in article IV, as no reference compounds were available. Maltooligosaccharides of DP3‒6 (Sigma– Aldrich, MO) were used in article IV to study the effect of DP on the elution time and detector response factor for members of a homologous series. Oligosaccharides in isolated fractions in article III were also separated by thin-layer chromatography (TLC) (Côté and Leathers 2005) to confirm that the peaks detected by HPAEC-PAD consisted of single compounds.

28 The total carbohydrate concentration in oligosaccharide and dextran solutions (in articles III and IV) was determined by the phenol-sulfuric acid method (Dubois et al., 1956).

4.4 Synthesis, isolation, and molar mass characterization of dextrans (I and II)

Dextrans were synthesized in vitro by incubating 1.06 U/ml (Nelson-Somogyi units unless otherwise specified) of WcE392-rDSR or WcCab3-DSR in 20 mM Na-acetate buffer, pH 5.4, containing 2 mM CaCl2 and 146 mM sucrose at 35 °C for 24 h. After the addition of two or three volumes of ethanol, the solution was centrifuged at 10,000 g and 4 °C for 15 min. The re-dissolved dextran pellet was precipitated and re-dissolved again. Dextrans were produced on agars by the native host W. confusa VTT E-90392 or the heterologous host L. lactis and extracted according to Maina et al. (2008). The dextrans were freeze-dried for characterization. The selective enzymatic hydrolysis and NMR spectroscopy analysis of dextrans are described in following sections.

The molar mass of dextrans was determined with a high-performance size-exclusion chromatography (HPSEC) multidetector system (Maina et al., 2014). Dextran solutions of 2.0 mg/mL were prepared in dimethylsulfoxide (DMSO) with 10 mM LiBr. The solutions were kept at room temperature for at least four days for dissolution and then filtered using a 0.45 μm membrane filter before analysis. The system was equipped with two analytical columns (exclusion limit >2 × 106) and a guard column. The HPSEC data were processed with the Viscotek OmniSEC 4.5 software. A dη/dc value of 72 µL/g for dextran in DMSO-based eluent was used (Basedow and Ebert 1979).

4.5 Acceptor reactions and isolation of oligosaccharides (II‒ IV)

To compare the specificity of WcE392-rDSR, WcCab3-DSR, and the commercial Lc. mesenteroides dextransucrase, maltose acceptor reactions were carried out under similar conditions (Table 3). WcE392-rDSR was also tested with several other di- and trisaccharides as acceptors (Table 3). The reactions were conducted at 30 °C in 0.2 ml of a 20 mM Na-acetate buffer (pH 5.4) containing 2 mM of CaCl2 for 24 h. The enzyme action was terminated by incubation in a boiling water bath for 10 min. A sucrose:acceptor molar ratio of 4 was used, except for the lactose reaction, where a second major product was promoted with an increasing sucrose:lactose ratio. In article IV, panose and a mixture of maltose trisaccharide products of a total carbohydrate

29 concentration of 1 mM were studied as acceptors in reactions using 5 mM of sucrose and 5.5 U/g sucrose of WcE392-rDSR at room temperature for 24 h.

Table 3. Acceptor information and the initial concentrations of sucrose (Suc) and the acceptor (Acc) in WcE392-rDSR-catalyzed (10 U/g of sucrose) reactions. Acceptor Structure Supplier Acc Suc:Acc (mM) molar ratio Maltose α-D-Glcp-(1→4)-D-Glcp Merck, Germany 150 4 Lactose β-D-Galp-(1→4)-D-Glcp " 150 6.7 Cellobiose β-D-Glcp-(1→4)-D-Glcp " 50 4 Laminaribiose β-D-Glcp-(1→3)-D-Glcp Megazyme, Ireland 50 4 Nigerose α-D-Glcp-(1→3)-D-Glcp Sigma-Aldrich, MO 50 4 Melibiose α-D-Galp-(1→6)-D-Glcp " 50 4 Isomaltose α-D-Glcp-(1→6)-D-Glcp TCI Europe nv, Belgium 25 4 Maltotriose α-D-Glcp-(1→4)-α-D- Sigma-Aldrich 50 4 Glcp-(1→4)-D-Glcp Forkosea α-D-Glcp-(1→4)-[α-D- Carbosynth, UK 50 4 Glcp-(1→6)]-D-Glcp a Name according to Hansen et al. (2008).

A common series of oligosaccharides were synthesized by WcE392-rDSR and WcCab3- DSR in the presence of maltose, as exhibited by their similar HPAEC-PAD profiles. Thus, only the products from WcCab3-DSR were taken for structural characterization. The synthesis was carried out in 100 ml using 292 mM sucrose, 146 mM maltose, and a WcCab3-DSR dosage of 10 U/g sucrose. Moreover, to characterize the two main products (named Lac3 and NR3) observed in the WcE392-rDSR lactose acceptor reaction, the synthesis was carried out in 50 ml. The major product of the cellobiose acceptor reaction (named Cel3) was also structurally characterized for comparison with the product derived from analogous lactose. The cellobiose acceptor reaction was scaled up to 50 ml.

After the reaction was stopped, two or three volumes of ethanol (Etax A, Altia Corporation, Finland) were added to the reaction mixture, which was then centrifuged to remove polymeric dextran. The supernatant containing fructose, residual acceptor, and acceptor products was concentrated using a rotatory vacuum evaporator (Heidolph Laborota 4001 Digital, Sigma-Aldrich, MO). Prior to injection into a Biogel P2 column (5 × 95 cm; Bio-Rad, Hercules, CA), the concentrate was filtered through a 0.45 μm membrane filter. Water was the eluent at a flow rate ≤0.7 ml/min. The oligosaccharides

30 in the collected fractions were analyzed by HPAEC-PAD. The fractions containing the highest proportion of products of interest were used for a subsequent structural analysis.

4.6 Structural analysis of dextrans and oligosaccharides

4.6.1 Enzymatic hydrolysis of dextrans and oligosaccharides (II and III) Dextransucrase products were selectively digested with specific enzymes for structural information. The hydrolysis reactions are presented in Table 4. After enzyme inactivation in a boiling water bath for 10 min, the digests were analyzed by HPAEC- PAD and compared to controls without enzyme treatment.

Table 4. Enzymatic hydrolysis conditions for dextran and oligosaccharide characterization. Analyte Enzyme dosagesa Conditionb Dextran 0.6 U/mg of C. erraticum dextranase and 30 °C, 48 h 0.06 U/mg of A. niger α-glucosidase Maltose-derived 0.3 U/mg of C. erraticum dextranase or " product series 0.04 U/mg of B. stearothermophilus α-glucosidase Lactose-derived 25 U/mg of A. niger α-glucosidase or overnight at trisaccharide (Lac3) 50 U/mg of A. niger β-galactosidase 30 °C Cellobiose-derived 25 U/mg of A. niger α-glucosidase or " trisaccharide (Cel3) 75 U/mg of Agrobacterium sp. β-glucosidase Nonreducing 25 U/mg of A. niger α-glucosidase, " trisaccharide (NR3) 200 U/mg of dextranase, or 100 U/mg of yeast invertase a Enzyme dosages reference the amount of dextran, total carbohydrate in dextran-removed maltose acceptor reaction mixture, or total carbohydrate in isolated product fractions for Lac3, Cel3 and NR3. b Reactions were performed in Na-acetate or Na-citrate buffer at 20‒50 mM and pH 5.0‒5.5.

4.6.2 NMR spectroscopy analysis of dextrans and oligosaccharides (I‒IV) NMR spectra were recorded with a 600 MHz Bruker Avance III NMR spectrometer (Bruker BioSpin, Germany). The dextran (10 mg/ml) and isolated trisaccharides (>0.3 mg/ml) were exchanged with D2O (MagniSolv, Merck, Germany) and placed in NMR tubes (Wilmad NMR tubes, 5 mm, ultra-imperial grade, Sigma-Aldrich, MO). The 1H and the corresponding 2D (multiplicity-edited HSQC, DQFCOSY, TOCSY, and HMBC) NMR experiments were performed at a temperature at which the residual water signal did not affect the signals of the analyte. Topspin v.3.2 (Bruker, Germany) software was

31 1 13 used to process and analyze the NMR spectra. The chemical shifts of H (δH) and C

(δC) were referenced to acetone (δH 2.225 and δC 31.55, respectively).

4.6.3 MSn analysis of oligosaccharides (II‒IV) Electrospray ionization multistage ion trap mass spectrometry (ESI-IT-MSn) was used for the molar mass determination and characterization of the oligosaccharides isolated by gel filtration. The analysis was carried out in negative or positive mode ([M+Cl]- and [M+Li]+, respectively) using the Agilent XCT Plus Ion Trap mass spectrometer with an electrospray ion source (Agilent Technologies, CA) or a Bruker Esquire LC quadrupole ion trap mass spectrometer with an electrospray ion source (Bruker Daltonik GmbH, Germany). A 1–10 µl sample (~0.5 mg/ml) was diluted in 200–400 µl of methanol- water-formic acid (50:49:1, v:v:v), and 0.5–1 µl of 10 mg/ml ammonium chloride or lithium acetate was added for adduct ion formation. The samples were introduced to the ion source with a direct infusion syringe pump at a flow rate of 5 µl/min. MS spectra were processed with LC/MSD Trap Software (Bruker Daltonik GmbH). Cellotriose (Megazyme) and panose (TCI Europe nv) were used as model compounds for the characterization of Cel3 and BIM3 (the minor trisaccharide obtained in the maltose acceptor reaction), respectively.

13 glc The maltose and lactose acceptor reactions were further carried out using [UL- C6 ] 13 glc sucrose (Omicron Biochemicals Inc. IN) so that the GLOS products were C6 -labeled for ESI-IT-MSn analysis. The maltose reaction was conducted by incubating 0.3 U of commercial Lc. mesenteroides dextransucrase in 0.5 ml of 20 mM Na-acetate buffer, pH 13 glc 5.4, containing 0.1 M C6 sucrose and 0.2 M maltose at 30 °C for 48 h. The GLOS products were isolated by gel filtration. According to TLC analysis, the fraction containing solely DP5 was selected for MSn analysis as [M+Cl]- and [M+Li]+ ions. 13 glc C6 sucrose was also used to synthesize the novel trisaccharides (Lac3 and NR3) identified in the lactose reaction mixture. The product mixture obtained was analyzed by MSn to characterize the fragmentation patterns of the two labeled products (article III).

4.7 Experimental design for IMO synthesis (IV)

The concentrations of sucrose, maltose, and WcE392-rDSR were selected as three independent variables in the experimental design. Preliminary experiments were carried out to select the following ranges of factors: sucrose (0.15–1 M), maltose (0.15–1 M), and dextransucrase in relation to sucrose concentration (1–10 U/g sucrose). An

32 experimental series consisting of 18 experiments was designed to model the effects of the factors on IMO synthesis. The reactions were conducted in 20 mM Na-acetate buffer, pH 5.4, containing 2 mM CaCl2 at 30 °C for 24 h.

The yields of individual IMOs of DP3‒6, the yield of total IMOs, and the percent of consumed maltose were each modelled by a quadratic regression model. Modde 9.0 (Umetrics AB, Sweden) was used to analyze the results by multiple linear regression (MLR) or partial least squares (PLS) method and to generate contour plots. The validity of the regression models was evaluated by the coefficient of determination (R2) and the coefficient of prediction (Q2).

33 5. RESULTS

The dextransucrases from two efficient dextran-producing Weissella confusa strains (W. confusa Cab3 isolated at IITG and W. confusa E-90392 from VTT Culture Collection) were the subjects of this study. In the beginning, the dextransucrase from native W. confusa E-90392 was produced and partially purified by PEG fractionation. However, the yield was low and subsequent purification attempts failed to remove the concomitantly produced dextran from the enzyme preparation (unpublished). Because the presence of dextran is known to affect the enzyme characteristics, dextran-free enzyme was obtained by cloning and expressing the dextransucrase gene in Lactococcus lactis with a nisin-inducible expression system (others’ work). The recombinant W. confusa E-90392 dextransucrase (WcE392-rDSR) was used in biochemical and kinetic studies. Moreover, the dextransucrase from the other strain, W. confusa Cab3 (WcCab3- DSR), was obtained with better yield by PEG fractionation and was used to synthesize dextran and GLOSs in parallel with WcE392-rDSR to compare their product specificity.

5.1 Characterization of WcE392-rDSR (I)

The specific activity of the nickel-affinity-purified WcE392-rDSR was 7.2 ± 0.3 U/mg and 18.9 ± 0.6 U/mg, as determined by the 14C-sucrose radioisotope assay and the Nelson-Somogyi assay, respectively. According to the HPAEC-PAD analysis of the reaction mixture after sucrose depletion, glucose and leucrose accounted for only 2% and 5% (w/w) of the total mono- and disaccharides. Fructose represented the remaining 93%, indicating that the transferase reaction synthesizing dextran comprises the main proportion of the WcE392-rDSR activity.

The Nelson-Somogyi assay was used to determine the effects of pH and temperature on the activity. The optimal pH for WcE392-rDSR activity was observed at pH 5.4 in 20 M Na-acetate buffer (Figure 6A). The rate of dextran synthesis increased with higher temperatures up to 35 °C, after which it decreased (Figure 6A), as shown in the activity assay (15 min) conducted at different temperatures. The thermal stability of WcE392- rDSR was studied at 25, 35 and 40 °C in the presence or absence of glycerol up to 24 h (Figure 6B). The enzyme stability decreased sharply when the temperature increased from 35 °C (half-life of 26.3 h) to 40 °C (half-life of 7.4 h). Glycerol enhanced the stability at each of the three temperatures; for example, it increased the enzyme half-life at 35 °C from 26.3 to 37.5 h.

34 WcE392-rDSR exhibited Michaelis-Menten kinetics with a KM of 14.7 mM, a Vmax of

8.2 µmol/(mg∙min) and a kcat of 23.2 1/s, determined using the sucrose radioisotope assay.

The sucrose radioisotope assay and the Nelson-Somogyi assay gave similar KM values, whereas values approximately 2.5 times higher were obtained for Vmax, kcat and kcat/KM with the Nelson-Somogyi method (Table 5).

Table 5. Specific activities and kinetics of WcE392-rDSR (Table 5 in article I).

Assay Activity KM Vmax kcat kcat/KM (U/mg) (mM) [µmol/(mg∙min)] (1/s) [1/(M∙s)]

14C-Sucrose 7.2 ± 0.3 14.7 ± 2.3 8.2 ± 0.4 23.2 ± 1.1 (1.6 ± 0.3) × 103 radioisotope Nelson-Somogyi 18.9 ± 0.6 13.0 ± 1.1 19.9 ± 0.6 56.4 ± 1.7 (4.3 ± 0.5) × 103

A B

Figure 6. (A) Effects of pH and temperature on WcE392-rDSR activity. The effect of pH was assayed in 20 mM Na-acetate buffer (pH 4‒6) or in 20 mM Tris-HCl buffer (pH 6.5‒7.0), and the effect of temperature in 20 mM Na-acetate buffer, pH 5.4. Error bars represent the standard errors of the mean (n=3). (B) The thermal stability of WcE392-rDSR in the presence and absence of 0.5% (v/v) glycerol after incubation for 1, 2, 5 and 24 h. (Adapted from Figures 3 and 4 in article I.)

5.2 Characterization of Weissella dextrans (I and II)

The dextrans synthesized in vitro by the two Weissella dextransucrases were isolated by ethanol precipitation and lyophilized for enzyme-aided branch-type fingerprinting, NMR spectroscopy and HPSEC analysis. For the fingerprinting of branch type, the

35 dextrans were hydrolyzed with a mixture of dextranase and α-glucosidase. The resulting resistant oligosaccharides exhibited a similar HPAEC-PAD profile (Figure 7A for WcCab3-DSR dextran, whereas that for WcE392-rDSR dextran is not shown) to those of dextran extracted from W. confusa VTT E-90392 culture (Maina et al., 2011). Based on the structure identified earlier for the in vivo-synthesized dextran (Maina et al., 2011), the enzymatically produced dextrans were shown to contain α-(1→3)-linked branches, with some branches elongated.

WcCab3-DSR dextran (Figure 7B) had a similar 1H NMR spectrum to that of the dextrans produced by WcE392-rDSR and its heterologous host L. lactis (Figure S3 in article I). Based on the relative intensities of the anomeric proton signals at 4.97 and 5.32 ppm, the ratio of α-(1→6) to α-(1→3) linkages was 97:3. The remaining proton resonances at 3.97, 3.91, 3.77, 3.73, 3.58, and 3.52 ppm were assigned to H-6b, H-5, H- 6a, H-3, H-2, and H-4 of the α-(1→6)-linked glucopyranosyl residues, respectively. In the edited HSQC NMR spectrum of WcCab3-DSR dextran (Figure 7C), a cross peak of 4.18→71.6 ppm was assigned to H-5→C-5 of the elongated branch point residue [- (1→6)-α-D-Glcp-(1→3)-] (Maina et al., 2011), confirming the presence of branches with more than one glucosyl unit.

The dextrans that were enzymatically synthesized and those produced in vivo by native W. confusa VTT E-90392 and heterologous host L. lactis were analyzed by HPSEC. The molar masses were obtained using DMSO with 10 mM LiBr as the solvent because dextrans are prone to aggregation in aqueous solutions (Maina et al., 2014). One major peak was observed with both refractive index (RI) and light scattering detectors (HPSEC chromatograms not shown). The molar masses of dextrans were in the range of 1.8 to 8.5 × 107 g/mol.

36 Figure 7. (A) HPAEC-PAD chromatograms showing resistant oligosaccharides from WcCab3- DSR dextran after dextranase and α-glucosidase treatment. 1H (B) and edited HSQC (C) NMR spectra of WcCab3-DSR dextran. The spectra were recorded at 600 MHz in D2O at 50 °C. Peaks in the NMR spectra were referenced to internal acetone (δH 2.225 ppm and δC 31.55 ppm). (Adapted from Figures 3 and 4 in article II.)

37 5.3 Characterization of the oligosaccharides produced by dextransucrases

5.3.1 Acceptor selectivity (III) The action of WcE392-rDSR on several glucose-containing di- and trisaccharides was evaluated. In all of the reactions, a large amount of fructose was released, and dextran was produced. Judging by the product peak area in the HPAEC-PAD chromatograms (Figure S1 in article III), the more effective acceptors among those tested were maltose, isomaltose, maltotriose, and nigerose. A series of products were detected in these reaction mixtures, indicating that the acceptor products could be further elongated. Lactose and cellobiose showed moderate effectiveness (section 5.3.4), whereas melibiose, laminaribiose, and forkose were poor acceptors (data not shown) for WcE392-rDSR.

5.3.2 Principal maltose product series (II) The HPAEC-PAD profile of the glucooligosaccharides (GLOSs) produced in the presence of maltose was similar between WcCab3-DSR, WcE392-rDSR, and the commercial Lc. mesenteroides dextransucrase. According to the HPAEC-PAD chromatogram of WcCab3-DSR products (Figure 8A, those of WcE392-rDSR and Lc. mesenteroides dextransucrase are not shown), a number of GLOSs were produced, with some being predominant and others being minor. After treatment with endo-acting dextranase, most GLOSs eluting after panose were degraded into their building blocks, i.e., glucose, isomaltose, isomaltotriose, and panose (Figure 8B), indicating that they have consecutive α-(1→6) linked glucosyl residues attached to the nonreducing end of the maltose and therefore belong to isomaltooligosaccharides (IMOs). As expected, the GLOSs were resistant to the hydrolysis of an α-glucosidase exo-acting on α-(1→4) linkages from the nonreducing end (article II).

The GLOSs that were produced in the WcCab3-DSR maltose reaction were isolated according to DP by gel filtration (Figures 8C‒F). A HPAEC-PAD and MS analysis of the fractions showed that for each principal IMO product, there was a minor product of the same DP. The minor products eluted earlier than did the corresponding major products (Figures 8C‒F) and seemed to constitute another series. The HPAEC retention times of the two product series were plotted versus their DP (Figure S4 in article IV). The plot for the minor series exhibited a similar slope to that of the principal IMOs but distinct from that of the α-(1→4)-linked maltooligosaccharides. For GLOSs growing via the same linkage, the slope (representing the peak interval between adjacent members

38 in the series) appeared to depend on the type of linkage. The HPAEC mobility result indicated that the two series observed in the maltose reaction mixture were IMOs. This was confirmed by the susceptibility of higher DP products of both series to dextranase (Figure 8B). Because a minor DP4 product coeluted with panose in HPAEC, the dextranase hydrolysate was further fractionated by gel filtration (data not shown). An HPAEC-PAD analysis of the isolated DP4 fractions showed that the minor DP4 remained after dextranase hydrolysis. The members in principal series, supposedly linear, were named after panose, LIM4 (DP4), LIM5 (DP5), and so on, while the minor IMOs were suspected to be branched and named BIM3 (DP3), BIM4 (DP4), and so on.

Fru DP4‒DP12 Mal Pan Glc A

IM2

B IM3

LIM6

BIM6 C

LIM5 D BIM5 m/z 863

LIM4 E BIM4 m/z 701

Pan

BIM3 F m/z 539

0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 min

Figure 8. HPAEC-PAD profiles of (A and B) acceptor products synthesized by WcCab3-DSR from 5% (w/v) maltose and 10% (w/v) sucrose: (A) unhydrolyzed; (B) hydrolyzed by dextranase; and (C‒F) selected fractions of isolated products by gel filtration. The m/z of DP3‒5 as [M+Cl]- ions are shown. The principal product series are named Pan and LIM4‒6 and the minor series BIM3‒6. Labeled peaks are: Glc, glucose; Fru, fructose; IM2, isomaltose; IM3, isomaltotriose; Mal, maltose; and Pan, panose. (Adapted from Figure 4 and Figure 2 in articles I and IV, respectively.)

To confirm the proposed structures for the principal series, negative ion MSn was used to analyze isolated principal products of DP3‒5 ([M+Cl]- ions of m/z at 539, 701, and 863, respectively). The diagnostic fragment ions at m/z 425 and 383 (loss of 78 and 120 Da, respectively) and m/z 281, 251, and 221 (loss of 60, 90 and 120 Da, respectively) in

39 the MS2 spectrum of DP3 (not shown) confirmed the α-(1→4) linkage at its reducing end and the α-(1→6) linkage at its nonreducing end, i.e., panose (Maina et al., 2013). Similarly, in the MS2 spectrum of the [M+Cl]- ion of DP4 (Figure 9A), fragment ions at m/z 587 and 545 (loss of 78 Da and 120 Da, respectively) indicated the presence of an α-(1→4) linkage at the reducing end. The ions at m/z 443, 413, and 383 (loss of 60, 90, and 120 Da, respectively) in the MS3 spectrum, starting with the ion at m/z 503, indicated the subsequent α-(1→6) linkage (Figure 9B). The α-(1→6) linkage at the nonreducing end was confirmed with the MS3 spectrum of the ion at m/z 341 (Figure 9C), which contained fragment ions at m/z 281, 251, and 221 (loss of 60, 90, and 120 Da, respectively). The MSn spectra of DP5 showed a similar trend (Figure 3S in article I): an α-(1→4) linkage at its reducing end and the remaining linkages being α-(1→6). These findings confirmed that the principal maltose products are homologous IMOs that contain maltose at the reducing end.

Intens. C x10 6 2 3 A DP4 -MS 503

1.0 0,4 A3 0.8 383 [M-H]- 665

0.6 C2 0,2 A3 341 0,2 443 A4-H2O 0.4 587 - 0,3 2,4 [M+Cl] A3 B A4 0.2 3 701 413 485 545 0.0

B -MS3 m/z 503 443 8000

383 6000

4000 413 485

2000 179 503 281 323 0 5000 C -MS3 m/z 341

4000

C1 3000 179 C2 341 B2 2000 0,2A 323 0,4A 2 2 281 221 B 0,3 1000 1 A2 161 251 0 100 200 300 400 500 600 700 m/z Figure 9. MS2 (A) and MS3 (B and C) spectra of the [M+Cl]- ion of the principal maltose product of DP4 (LIM4). The MS3 spectra in B and C are for the fragment ions at m/z 503 and 341, respectively. (Adapted from Figure 5 in article I.)

40 5.3.3 Minor maltose product series (IV) Because BIM3 of the minor series eluted slightly earlier than did panose in gel filtration, few fractions collected before the elution of panose contained enriched BIM3 and were selected for NMR spectroscopy and ESI-IT-MSn (section 5.3.6) analysis. The structure of BIM3 was determined by a set of NMR experiments using a fraction containing an approximately equimolar mixture of panose and BIM3. Pure panose was analyzed in parallel so that the signals from panose could be subtracted from those of the mixture, and the remaining signals were interpreted. Compared to the 1D 1H spectrum of panose, five doublets in the anomeric proton region were found for BIM3 (Figure 10). The integrals of the anomeric protons indicated that the trisaccharide was an anomeric mixture, with the resonances of one nonreducing residue affected by mutarotation equilibrium. The proton signals of each spin system were assigned using TOCSY and

DQFCOSY spectra (Figure S5 in article IV). The signals at δH 5.45 (d J1-2 3.4 Hz) and

δH 4.81 (d J1-2 8.0 Hz) were assigned to the α and β anomeric protons of the reducing glucopyranosyl residue, respectively. The signal at δH 5.49 (d J1-2 3.5 Hz) corresponded to the anomeric proton of the α-(1→4)-linked glucopyranosyl residue in maltose. The other two anomeric signals at δH 5.10 (d J1-2 3.6 Hz) and δH 5.41 (d J1-2 3.9 Hz) were assigned to a terminal α-(1→2)-linked glucopyranosyl residue in α and β anomers, respectively (Figure 10).

Figure 10. 1D 1H NMR spectra of (A) pure panose and (B) a mixture containing equimolar amounts of panose and BIM3. The spectra were measured at 600 MHz in D2O at 11 °C. The structure of BIM3 is illustrated. (Adapted from Figure 4 in article IV.)

41 The downfield shift of C2 signals of the reducing residue in the edited HSQC spectrum (Figure 4C and Table S3 in article IV) indicated that the third glucosyl residue was attached to the C2 of maltose-reducing end, forming a branched structure. Direct evidence for the α-(1→2) linkage was observed in the HMBC (Figure 4D in article IV) I III correlation peaks: between Glc -C2 (δC 76.76) and Glc -H1 (δH 5.10) and between III I I Glc -C1 (δC 97.41) and Glc -H2 (δH 3.67) for the α-anomer and between Glc -C2 (δC III III I 79.52) and Glc -H1(δH 5.41) and between Glc -C1 (δC 98.83) and Glc -H2 (δH 3.41) for the β-anomer. Thus, BIM3 was identified as α-D-Glcp-(1→2)-[α-D-Glcp-(1→4)]- D-Glcp, also known as centose.

To understand the structures and synthesis mechanism of the higher homologues of the minor series, BIM3 (in a fraction containing an equal molarity of BIM3 and panose) was tested as an acceptor. The HPAEC-PAD analysis before and after the reaction showed that BIM3 did not act as an acceptor (data not shown). Thus, the higher DP minor products contained a single-unit α-(1→2)-branch at the reducing end, and additional α- (1→6)-linked residues in the main chain, as the α-(1→2)-linked branch was not able to elongate further. Moreover, the minor series were proposed to be synthesized through direct branch formation on an elongated linear IMO rather than through the elongation of the main chain after branch formation. This mechanism was supported by the product pattern observed using panose as a sole acceptor. A trace of BIM5 was formed (Figure S6 in article IV) most probably by branch formation on LIM4. The branch may be formed on panose as well, although the resulting BIM4 could not be directly detected in the HPAEC analysis (as it was not well separated from panose).

5.3.4 Lactose- and cellobiose-derived branched trisaccharides (III) The major products observed in the reaction mixtures of lactose and cellobiose (Figure 11A and 11B) were selected for structural analysis to determine whether the same type of linkage was formed on the two analogs by Weissella dextransucrase. The products derived from lactose and cellobiose were referred to as Lac3 and Cel3, respectively, as they were later found to be trisaccharides. Moreover, in the lactose reactions with varying sucrose and enzyme concentrations (Figure S2 in article III), a second product was promoted when both concentrations increased. This product was also selected for characterization and named NR3 (later found to be a nonreducing trisaccharide, Figure 11A). The two products from the WcE392-rDSR lactose reaction mixture and one from the cellobiose reaction were isolated by gel filtration (Figure 11C). The selected Lac3 and Cel3 fractions contained negligible amounts of impurities, whereas the NR3 fraction contained a minor portion of Lac3, in addition to other impurities. The three product

42 fractions were subjected to an ESI-IT-MS2 and NMR spectroscopy analysis and enzymatic hydrolysis.

C Lac3 Fraction Lac3

Isolated from lactose NR3 product mixture Fraction NR3

Cel3

Fraction Cel3

DP3#

Isolated from cellobiose product mixture

Cel

5.00 10.00 15.00 20.00 25.00 30.00 35.00 min Figure 11. HPAEC-PAD profiles of (A and B) acceptor product mixtures of WcE392-rDSR (10 U/g sucrose) with (A) 150 mM lactose and 1 M sucrose and (B) 50 mM cellobiose and 200 mM sucrose and those of (C) selected product fractions isolated by gel filtration in order of elution. The insets show the zoomed-in chromatograms of the product areas. Labeled peaks are: Glc, glucose; Fru, fructose; Leu, leucrose; Lac, lactose; Cel, cellobiose. The peak DP3# was presumed to be NR3. (Modified from Figure 1 in article III.)

43 The three isolated products were preliminarily analyzed by negative ion MS2. The three products ([M+Cl]- ions) exhibited an m/z value of 539 and were therefore trisaccharides. The MS2 spectra of [M+Cl]- ions of Lac3 and Cel3 were similar (Figure 12A), indicating similar structures and thus fragmentation patterns. The relative abundance of the fragment ions resulting from the cross-ring cleavage of the reducing-end residue was different from that of cellotriose (data not shown) and other linear oligosaccharides containing a β-(1→4)-linked reducing-end hexopyranose (Maina et al., 2013). This suggests that Lac3 and Cel3 probably have a rare branched structure. The absence of cross-ring fragment ions in the MS2 spectrum of the [M+Cl]- ion of NR3 (Figure 12A, the ion at m/z 425 probably resulted from Lac3 present in the fraction) indicated that NR3 is possibly a nonreducing trisaccharide.

The structures of Lac3 and Cel3 were determined by a set of NMR experiments. Their anomeric proton regions in 1D 1H spectra (Figure 12B) were similar. Both showed six anomeric proton signals representing two sets of resonances due to the effect of mutarotation equilibrium on the two other residues. For example, the signals of Cel3 at

δH 5.44 (d J1-2 3.5 Hz) and δH 4.81 (d J1-2 8.0 Hz) corresponded to α- and β-anomeric protons of the reducing glucopyranosyl residue, respectively. The signals at δH 4.52 (d

J1-2 7.8 Hz) and δH 4.51 (d J1-2 7.9 Hz) corresponded to the anomeric proton of the β-

(1→4)-linked glucopyranosyl residue. The last two anomeric signals at δH 5.10 and δH

5.37 (both d J1-2 3.8 Hz) were from an α-(1→2)-linked glucopyranosyl residue. Evidence for α-(1→2)-linkage in Cel3 was observed in the HMBC (Figure S3 in article III) I III correlation peaks: between Glc -C2 (δC 76.84) and Glc -H1 (δH 5.10) and between III I I Glc -C1 (δC 97.70) and Glc -H2 (δH 3.67) for the α-anomer and between Glc -C2 (δC III III I 79.60) and Glc -H1(δH 5.37) and between Glc -C1 (δC 99.20) and Glc -H2 (δH 3.41) for the β-anomer. Thus, Cel3 was identified as α-D-Glcp-(1→2)-[β-D-Glcp-(1→4)]-D- Glcp (21-α-Glcp-cellobiose). Similarly, Lac3 was identified as α-D-Glcp-(1→2)-[β-D- Galp-(1→4)]-D-Glcp (2Glc-α-Glcp-lactose). Thus, both Lac3 and Cel3 contain a glucosyl residue that is α-(1→2) linked to the reducing end of the acceptor disaccharide (lactose and cellobiose, respectively).

44 A B Intens. 6 x10 6 425

4 β-GlcII(β) Glc II Glc I Cel3 [M-H]- α-GlcIII(β) β-GlcII(α) 2 Hex 503 [M+Cl]- I α-GlcIII(α) β-GlcI 179 485 539 α-Glc 263 341 467 221 III 0 Glc x106

425 1.5 β-Gal(β) α-GlcII(β) β-Gal(α) Lac3 1.0 [M-H] - Gal Glc I 503 α-GlcI α-GlcII(α) β-GlcI Hex - 0.5 [M+Cl] 179 263 341 485 539 221 383 467

x107 Glc II 3 [M-H]- 503

I II 2 α-Glc α-Glc Glc I NR3 Glc II 1 [M+Cl] - 539 425

100 150 200 250 300 350 400 450 500 m/z Fru Figure 12. MS2 spectra of [M+Cl]- ions (A) and 1D 1H NMR spectra (B) of isolated Cel3, Lac3, and NR3. The NMR spectra were measured at 600 MHz in

D2O at 16 ºC. The β-anomeric signals of the reducing-end residue of Lac3 and Cel3 were partially suppressed with the residual water signal. (Adapted from Figures 2 and S6‒8 in article III.)

45 Glc Lac3 Lac Fraction Lac3

Glc NR3 Fraction NR3 + α-glucosidase Suc Fru Lac3

Glc Cel3 Fraction Cel3 Cel

Lac3 Gal Fraction Lac3 + β-galactosidase Glc Koj

Cel3 Fraction Cel3 + β-glucosidase Glc Koj

15.00 25.00 35.00 45.00 min Figure 13. HPAEC-PAD profiles of isolated product fractions before (dashed lines) and after (solid lines) specific enzymatic digestion. The labeled peaks were identified based on their retention time against standards: Glc, glucose; Fru, fructose; Lac, lactose; Cel, cellobiose; Gal, galactose; Suc, sucrose; and Koj, kojibiose. (Adapted from Figure 3 in article III.)

The product fractions of Lac3 and Cel3 were selectively digested with specific enzymes and then analyzed by HPAEC-PAD (Figure 13). The α-(1→2) linkage was hydrolyzed by α-glucosidase, generating glucose and the corresponding acceptor (lactose and cellobiose, respectively). The β-(1→4) linkage in Lac3 was hydrolyzed by β- galactosidase, leading to the formation of galactose and kojibiose (α-D-Glcp-(1→2)-D- Glcp), as expected. The glucose that formed concomitantly was due to the side reaction of the β-galactosidase preparation on the α-(1→2)-linked glucosyl residue, as manifested by the reaction of the β-galactosidase preparation on Cel3 and pure kojibiose (data not shown). Although Cel3 was not very susceptible to hydrolysis by the β- glucosidase used, a small amount of it was converted to glucose and kojibiose. The above results further support the notion that Lac3 and Cel3 contain a glucosyl unit that is α-(1→2) linked to the reducing end of their acceptors.

5.3.5 Novel sucrose-containing trisaccharide (III) Under the conditions selected for the lactose acceptor reaction, a second product denoted by NR3 (Figure 11A) was obtained. The 1D 1H spectrum of the isolated NR3 (Figure 12B) was different from that of Lac3 and Cel3. Apart from the signals from the impurities (mainly Lac3) in isolated NR3, there were two anomeric protons at δH 5.42

(d J1-2 3.9 Hz) and δH 4.96 (d J1-2 3.7 Hz), both with an α configuration (Figure 12B). The presence of only two anomeric protons for a trisaccharide indicated that one of its

46 residues has a quaternary anomeric carbon and no corresponding anomeric proton. Because sucrose was the only sugar in the reaction system with this feature, NR3 was inferred to contain sucrose, which was supported by comparing its 13C assignments with those of sucrose (Table S3 in article III). The third residue exhibited resonances typical of a terminal α-(1→6)-linked α-glucopyranosyl unit, which was found attached to the fructofuranosyl residue of sucrose, as manifested by the downfield shift of fructofuranosyl C6 signal in the edited HSQC. The linkage was also confirmed by correlation peaks in the HMBC spectrum (Figure S5 in article III) between Fru-C2 (δC I II II 104.71) and Glc -H1 (δH 5.42), Fru-C6 (δC 70.08) and Glc -H1 (δH 4.96), and Glc -C1

(δC 99.58) and Fru-H6 (δH 4.06 and 3.76). NR3 was consequently identified as an α-D- Glcp-(1→2)-[α-D-Glcp-(1→6)]-β-D-Fruf (6Fru-α-Glcp-sucrose / 2Fru-α-Glcp- isomaltulose), also called isomelezitose.

NR3 was hydrolyzed to glucose, fructose, and sucrose by α-glucosidase (Figure 13). The release of glucose and sucrose is in line with the proposed structure, whereas fructose seems to result from the further hydrolysis of sucrose. NR3 was resistant to dextranase (data not shown), which requires two consecutive α-(1→6)-linked glucosyl units for its function (Maina et al., 2011). NR3 was not hydrolyzed by invertase, whereas raffinose (α-D-Galp-(1→6)-α-D-Glcp-(1→2)-β-D-Fruf) was hydrolyzed to fructose and melibiose (α-D-Galp-(1→6)-D-Glcp) (data not shown). This agrees with the structure of NR3 having a glucosyl unit, which is α-(1→6) linked to the fructosyl residue, whereas the galactosyl unit in raffinose is α-(1→6) linked to the glucosyl residue of sucrose.

In the product mixture of cellobiose acceptor reaction, NR3, if present, could not be directly detected by HPAEC-PAD due to its coelution with cellobiose. However, the analysis of the product fractions isolated by gel filtration (Figure 11C) showed the presence of an unknown product, probably a trisaccharide according to its elution behavior in gel filtration. This product was presumed to be NR3 based on its HPAEC retention time. Moreover, a product with a similar HPAEC retention time to NR3 was found in the product mixtures of all the tested acceptors except for maltose and isomaltose (Figure S1 in article III and others unpublished). However, higher dilution was used for the product mixtures of the two effective acceptors and thus the possibly formed NR3 could not be detected. These observations indicated the formation of NR3 as a side product from endogenous sugars in dextransucrase-catalyzed reactions.

47 5.3.6 Applicability of MSn analysis in the characterization of dextransucrase-synthesized GLOSs (III and IV)

13 glc 13 C6 sucrose (6 carbon atoms in the glucosyl residue are C-labeled, while those in the fructosyl residue are unlabeled) was used for the synthesis of GLOSs, the MSn analysis of which provided information on the monosaccharide composition of unknown GLOSs. In the beginning of the characterization of the minor products in the maltose acceptor reaction, they were first suspected to be fructose-containing GLOSs, as dextransucrases are known to use fructose and sucrose as acceptors (Moulis et al., 2006b; 13 glc Seo E et al., 2007). The reaction was thus carried out with C6 sucrose, as fructose- containing GLOSs (with one unlabeled residue and the remaining residues labeled) can be differentiated from the major maltose-based GLOSs (with two unlabeled residues). Therefore, if the minor series contain fructose, their m/z will be 6 plus that of the major maltose-based product of the same DP. However, when analyzing the two products of DP5 isolated from the maltose reaction mixture, only a single m/z signal was detected that corresponded to a pentaose with 2 unlabeled residues (data not shown). As the minor DP5 product was present in an appreciable amount in the isolated DP5 mixture (Figure 8D), the result indicated that the minor series probably also contained maltose rather than fructose. Unfortunately, the negative and positive MSn spectra of the labeled DP5 product mixture (data not shown) did not show any diagnostic fragment ions other than those known to be formed from the major LIM5, which was later explained by the fragmentation patterns of the minor products.

13 glc n The use of C6 sucrose was also useful in the MS characterization of the two novel structures (Lac3 and NR3) from lactose acceptor reaction after their structural elucidation by NMR spectroscopy analysis. Attempts to characterize unlabeled Lac3 were not successful due to the presence of isomeric impurity hypothesized to be a fructose-containing linear structure, the presence of which was later confirmed (article 13 glc III). To solve this problem, synthesis was carried out again with C6 sucrose. Because the m/z of resultant Lac3 and NR3 differed by 6, which enabled their separate fragmentation in MSn, the reaction mixture was directly analyzed without fractionation. The GLOSs composition and their sugar components were reflected in the MS spectrum of the labeled product mixture (Figure 14). Notably, the GLOSs containing two unlabeled hexose units (lactose-derived) were found in only DP3 and 4 and were not further elongated. Because of the addition of a labeled glucosyl branch in Lac3, the differentiation of the two nonreducing end residues was made possible, enabling a thorough MSn characterization of its fragmentation behavior (not within the scope of this thesis). In short, the combination of both positive and negative ionization techniques was beneficial for the analysis of rare oligosaccharides containing a 2,4-O-disubstituted

48 reducing end residue. The negative ion MS2 spectra of Lac3 and Cel3 showed an intense 2 signal corresponding to a loss of 60 Da+H2O. In positive ion MS spectra, the cross-ring ions resulting from a loss of <162 Da were not observed (others’ results in article III). These results provide reference MSn spectral data for such rare structures.

- 13 glc Figure 14. MS spectrum ([M+Cl] ) of the C6 -labeled oligosaccharide mixture from the 13 glc WcE392-rDSR lactose reaction. In the spectra, the number of * shows that of the C6 -labeled residues in the compound. Solid circles symbolize the labeled glucosyl residues transferred from sucrose, open circles represent unlabeled residues in the acceptor lactose, and triangles are unlabeled fructosyl residues. The inset shows the HPAEC-PAD chromatogram of the same mixture. (Adapted from Figure S9 in article III.)

The MSn fragmentation behavior of Lac3 was useful for the subsequent preliminary characterization of BIM3 by MSn. A mixture of panose and BIM3 with a ratio of ~7:3 was used for MSn analysis. It was analyzed in parallel with pure panose; thus, the difference in their fragment ion profiles indicated the behavior of BIM3 in the mixture. The fragmentation pattern of panose (Maina et al., 2013) is illustrated in Figure 15I. In the negative MS2 spectrum of the [M+Cl]- ions in the mixture (Figure 15II-A), the presence of BIM3 resulted in an increase in cross-ring fragment ions, such as m/z 425, 263, and 221, compared to pure panose (Figure 15I-A). The profile of the fragment ions originating from BIM3 resembled that from Lac3 and Cel3 (paper III), which contained a 2,4-disubstituted reducing residue.

In positive ionization, the MS2 spectrum of the [M+Li]+ ion of panose was generally similar to that of the mixture with BIM3 (Figure 15I-B and II-B, respectively). However, the fragment ions at m/z 391 and 451 in the spectrum of the mixture were lower in intensity compared to those in the spectrum of panose, indicating a lack of such fragment ions from BIM3. Similarly, cross-ring cleavage ions that formed by the loss of <162 Da (corresponding to ions at m/z 391 and 451) in the MS2 spectrum of the [M+Li]+ ion were also absent from Lac3 (article III). When the Y2 and C2 ions at m/z 349 were further fragmented at the MS3 stage, a marked increase in cross-ring fragment ion at m/z 229

49 was observed in the spectrum of the mixture (Figure 15II-C) compared to the corresponding spectrum of panose (Figure 15I-C). The cross-ring fragment ion at m/z 229 (loss of 120 Da) was previously reported to be diagnostic of a (1→2) linkage (article III). The above evidence indicates that BIM3 contains a (1→2)-linked hexopyranosyl residue on the reducing residue of maltose. The analysis of BIM3 based on negative and positive MSn spectral data of the reference compound Lac3 demonstrated the potential of MSn analysis in the structural characterization of rare oligosaccharides, even in impure samples.

50 Figure 15. MSn spectra of (I) pure panose and (II) a fraction containing panose and BIM3 at a ratio of ~7:3. (A) Negative ion MS2 of [M+Cl]- m/z 539, (B) positive ion MS2 of [M+Li]+ m/z 511, and (C) positive ion MS3 for fragment ions at m/z 349. The fragmentation pathways of panose are illustrated in section I. The downward arrows show the intensity changes of the diagnostic fragment ions due to the presence of BIM3 (Figure 3 in article IV).

51 5.4 Optimization of IMO synthesis by response surface modeling (IV)

The IMO product profile (mainly principal linear IMOs) was systematically studied as a function of sucrose and maltose concentrations and WcE392-rDSR dosage. The preliminary experiments showed that dextran synthesis was largely suppressed in the experimental region (Table S1 in article IV), within which a central composite design was carried out (Table 6). The yield of individual IMOs of DP3‒6 and the yield of total IMOs and the percent of consumed maltose were examined as responses after 24 h of reaction (Table 6). The branched minor products were not quantified; however, the content of BIM3 was affected in a similar way as was the content of panose (data not shown). According to a carbohydrate analysis of the product mixtures (Table S2 in article IV and Table 6), the substrates consumed and the products formed were roughly balanced. The IMO profile varied between the experimental points, and the four replicates at the center point of the design showed good reproducibility (Table 6 and Figure 16). A quadratic equation was used to model each response by excluding irrelevant terms to obtain the highest coefficients of determination (R2) and prediction (Q2) possible. The relatively high R2 and Q2 values (Table 7) indicated that the models fit well with the experimental data and could be used to predict the behavior of responses within the experimental region.

52 Table 6. The central composite experimental design and response data for modeling the effects of reaction conditions on IMO synthesis by WcE392-rDSR (Table 1 in article IV). Run Maltose Sucrose Enzyme Panose LIM4 LIM5 LIM6 Total Consumed (mol/L) (mol/L) (U/g (g/L) (g/L) (g/L) (g/L) IMOsa maltose (%) sucrose) (g/L) 1 0.15 0.15 1 10.39b 17.24 7.69 1.46 45.69 36.71 2 1 0.15 1 19.33 5.03 - - 25.89 20.55 3 0.15 1 1 3.89 22.33 39.67 31.35 162.84 86.25 4 1 1 1 120.10 105.46 31.20 3.90 287.16 38.15 5 0.15 0.15 10 13.45 22.90 15.63 4.50 70.01 37.03 6 1 0.15 10 47.87 23.38 2.58 - 82.98 28.50 7 0.15 1 10 3.86 13.18 25.98 31.19 204.43 94.19 8 1 1 10 120.55 138.99 80.15 19.00 408.68 50.19 9 0.15 0.575 5.5 5.96 19.30 33.83 27.96 154.46 84.20 10 1 0.575 5.5 91.66 98.77 37.78 5.92 265.03 40.37 11 0.575 0.15 5.5 37.91 31.57 6.58 0.64 89.98 23.02 12 0.575 1 5.5 45.70 98.17 97.11 46.89 365.89 64.74 13 0.575 0.575 1 48.53 79.17 40.06 8.82 199.51 39.77 14 0.575 0.575 10 43.40 62.11 45.95 14.28 191.76 59.17 15 0.575 0.575 5.5 47.62 79.08 56.26 17.00 229.36 49.84 16 0.575 0.575 5.5 49.39 83.00 55.05 15.82 240.15 46.69 17 0.575 0.575 5.5 47.50 81.13 58.39 18.23 236.31 49.96 18 0.575 0.575 5.5 49.00 84.61 58.82 17.83 243.86 49.26

a Total IMOs represents all of the IMO products, including DP>6 members of principal linear series and members of minor branched series. b Samples were analyzed twice, and the average values are presented. A hyphen (-) denotes that the value was below the lower limit of quantification.

53 Suc Fru Mal Pan LIM4 Run 1

Run 2

LIM5 LIM6 Run 3

Run 4

Run 5

Run 6

Run 7

Run 8

Run 9

Run 10

Run 11

Run 12

Run 13

Run 14

Run 15−18

0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 min Figure 16. HPAEC-PAD profiles of acceptor product mixtures in experimental design. The conditions of each run are shown in Table 6. Labeled peaks are: Fru, fructose; Suc, sucrose; Mal, maltose; and Pan and LIM4−6, the major IMO products in the DP range of 3–6. The chromatograms are under the same intensity scale, and the samples were diluted by the same amount. (Figure S2 in article IV.)

54 Table 7. Effects of factors expressed as their corresponding coefficients obtained by modeling for IMO yield (with respect to DP) and maltose consumption in WcE392-rDSR-catalyzed maltose acceptor reaction (Table 2 in article IV). Responses Regression equationsa Coefficients of determination (R2) and prediction (Q2) Panose (g/L) 44.78 + 36.20M + 16.52S + 2.69E + 23.69M∙S R2=0.95, Q2=0.89 LIM4 (g/L) 78.46 + 27.67M + 27.80S + 3.13E - 20.26M2 R2=0.95, Q2=0.82 - 14.43S2 + 27.58M∙S LIM5 (g/L) 55.09 + 2.22M + 18.58S + 3.97E - 10.09M2 R2=0.85, Q2=0.60 - 7.49E2 + 4.90M∙S + 4.22M∙E LIM6 (g/L) 18.79 - 5.19M + 9.64S + 1.80E - 4.32E2 R2=0.85, Q2=0.68 - 2.48M∙S + 0.88S∙E Total IMOs (g/L) 236.07 + 42.46M + 111.45S + 23.68E - 27.90M2 R2=0.97, Q2=0.87 - 45.86E2 + 41.93M∙S + 14.09M∙E Consumed 50.16 - 16.06M + 18.77S + 4.77E + 8.99M2 R2=0.97, Q2=0.90 maltose (%) - 9.42S2 - 8.43M∙S a M, initial maltose concentration (mol/L); S, initial sucrose concentration (mol/L); E, WcE392- rDSR concentration (U/g sucrose).

For each response, three two-dimensional contour plots with maltose and sucrose concentrations as the two axes (Figure S3 in article IV) were used to illustrate the models, and each plot corresponded to a different enzyme dosage. Because the enzyme dosage in relation to the sucrose concentration had a minor effect on the responses, contour plots at an enzyme dosage of 5.5 U/g sucrose were chosen to represent the effects of the two main variables in Figure 17. The yield of each IMO of DP3−6 and the total IMOs, as well as the percent of consumed maltose, increased by increasing the initial sucrose concentration, whereas the initial maltose concentration exhibited varying effects on the production of each IMO (Figure 17). The maximal production of panose, LIM4, and total IMOs was attained when maltose and sucrose were both at the highest concentrations tested. The production of LIM5 and LIM6, however, was favored at the medium and lowest maltose concentrations, respectively, while maintaining the highest sucrose concentration. The maltose utilization ratio was influenced positively by sucrose input and negatively by maltose input. Aiming for a good yield of total IMOs and a product profile towards higher DP, run 12 employing a medium enzyme load (5.5 U/g sucrose) and medium maltose (0.575 M) and the highest sucrose (1 M) concentrations was considered close to the optimum. The run led to sucrose depletion and a 65%

55 consumption of maltose. A yield of 366 g/L of total IMOs was attained, with a weighted average DP of >4.5 (taking into account the amounts of DP3−6, not higher DP).

Figure 17. Contour plots showing the influence of the initial sucrose and maltose concentrations in the WcE392-rDSR acceptor reaction on the yields (g/L) of (A) panose, (B) LIM4, (C) LIM5, (D) LIM6, (E) total IMOs, and (F) the consumed maltose (%). The reactions were conducted with an enzyme dosage of 5.5 U/g sucrose for 24 h (Figure 1 in article IV).

56 6. DISCUSSION

6.1 Properties of Weissella dextransucrases and dextrans (I and II)

Biochemical characterization showed that WcE392-rDSR was affected similarly to other reported Weissella dextransucrases by pH and temperature (Kang et al., 2009; Shukla and Goyal 2011; Amari et al., 2012; Bejar et al., 2013; Rao and Goyal 2013). It should be pointed out that in the study of the effect of pH on WcE392-rDSR activity, the Tris- HCl buffer used for pH 6.5‒7.0 contained Tris, which may have inhibited the enzyme activity (Côté and Robyt 1982). Thus, the decrease of activity in this pH range may have resulted from the inhibiting effect of Tris, in addition to that of pH per se. However, for the purpose of determining the pH optimum, the experiments were not repeated with another buffer of pH 6.5‒7.0, given that the enzyme activity at pH 6.0 already decreased drastically in Na-acetate buffer. The quick inactivation above 40 °C and the stabilizing effect of glycerol were observed for both WcCab3-DSR (others’ results in article II) and WcE392-rDSR. It should be noted that WcCab3-DSR was produced by sucrose induction and partially purified by PEG fractionation and thus was bound to dextran (Majumder et al., 2007), whereas WcE392-rDSR was produced in glucose-containing medium and devoid of bound dextran. The half-lives of WcCab3-DSR and WcE392- rDSR of similar protein concentrations at ~30 °C without any additive were 10.3 h and 26.3 h, respectively, despite the fact that the former was dextran bound and dextran is known to stabilize dextransucrase. In a recent study of the recombinant dextransucrase from W. confusa Cab3, the addition of dextran could increase the enzyme activity by at most 14% (Shukla et al., 2016). The stabilizing and activating effect of dextran on Weissella dextransucrases thus seems to be minor.

The kinetics for a recombinant dextransucrase from W. confusa LBAE C39-2 and a native one from W. cibaria JAG8 have been reported to follow Michaelis-Menten kinetics, with KM and Vmax values of 8.6 mM and 20 µmol/(mg∙min), respectively, for the former, and 13 mM and 27.5 µmol/(mg∙min), respectively, for the latter (Amari et al., 2012; Rao and Goyal 2013). The kinetics parameters of the C39-2 dextransucrase were determined by a 3,5-dinitrosalicylic acid (DNS) assay (Amari et al., 2012), which has been suggested to overestimate the dextransucrase activity due to the significant over-oxidation of dextran and other carbohydrates in the sample, thus giving invalid enzyme assay and kinetics results (Vettori et al., 2011; McIntyre et al., 2013). According to Vettori et al. (2011), the DNS method generates a 5.8-fold higher activity than the 14C-sucrose radioisotope method. However, it is not known how accurate is the Nelson- Somogyi assay, the other commonly used reducing-value assay for dextransucrase

57 activity and kinetics measurements. It was thus compared with the 14C-sucrose radioisotope assay in this study. The Nelson-Somogyi method gave an approximately 14 2.5-fold higher activity and Vmax than did the C-sucrose radioisotope method, while the

KM values were similar regardless of the method used. Because small amounts of glucose and leucrose formed as side products, the higher values produced by the Nelson- Somogyi method seemed to be due to the oxidation of several residues down the dextran chain other than the reducing residue. As determined by Nelson-Somogyi method, the kinetic values for WcE392-rDSR (KM of 13.0 mM and Vmax of 19.9 µmol/(mg∙min)) were similar to those of W. cibaria JAG8 dextransucrase (KM of 13 mM and Vmax of 27.5 14 µmol/(mg∙min)) (Rao and Goyal 2013). The WcE392-rDSR KM obtained by the C- sucrose radioisotope method (14.7 mM) was comparable to that of a glucansucrase from Leuconostoc mesenteroides NRRL B-1118 (12 mM) using the same method (Côté and 14 Skory 2012). In this thesis, the Vmax (8.2 µmol/(mg∙min)) determined by the C-sucrose radioisotope assay was reported for the first time for a glucansucrase.

The dextran produced in vitro by WcE392-rDSR exhibited high similarity to that produced in vivo by native and heterologous hosts as well as that by the other enzyme WcCab3-DSR, as characterized by 1H NMR spectroscopy and HPSEC. The Weissella dextrans showed an α-(1→6) to α-(1→3) linkage ratio of 97:3, being slightly less branched than the dextran from Lc. mesenteroides NRRL B-512F (with a ratio of 96:4), which was analyzed comparatively (Maina et al., 2008; Bounaix et al., 2009). The molar masses reported for Weissella dextrans ranged between 106 and 107 g/mol (Amari et al., 2012; Maina et al., 2014). Similarly high molar masses (107‒108 g/mol) were obtained in this study (articles I and II) by HPSEC. Significantly high molar masses (108 g/mol) were reported for dextrans synthesized in vitro with Lc. mesenteroides B-512F dextransucrase mutants applying the asymmetrical flow field flow fractionation technique (Irague et al., 2012). The high molar mass and low branching make Weissella dextrans attractive hydrocolloids for food applications, such as in sourdough baking (Lacaze et al., 2007).

6.2 Specificity of Weissella dextransucrases in acceptor reactions

6.2.1 Comparison with commercial Lc. mesenteroides dextransucrase (III and IV) Before this thesis work, the dextransucrase acceptor reactions for GLOS synthesis were only investigated for DSR-S from Lc. mesenteroides B-512F. WcE392-rDSR was similar to DSR-S in its selectivity towards the acceptors tested (in decreasing order):

58 maltose, isomaltose, nigerose, lactose, cellobiose, and melibiose (Robyt and Eklund 1983). The unusual acceptors tested in this thesis work, laminaribiose and forkose, showed poor effectiveness for WcE392-rDSR. Moreover, the product patterns of WcE392-rDSR were also similar to those of DSR-S based on the product profiles from maltose, isomaltose, maltotriose, nigerose, lactose, and cellobiose (Robyt and Eklund 1983; Fu and Robyt 1990).

In the presence of lactose or cellobiose, both WcE392-rDSR and DSR-S form mainly a trisaccharide with a glucosyl residue α-(1→2) linked to the acceptor’s reducing end (Robyt 1995). In their maltose acceptor reaction, a principal series of linear IMOs are synthesized from further elongation on panose. Moreover, a number of previously unknown minor products were also observed in the reactions of both enzymes. In this thesis, the minor trisaccharide from WcE392-rDSR was identified as centose with a glucosyl residue α-(1→2) linked to the maltose-reducing end. However, a trace amount of forkose (α-D-Glcp-(1→4)-[α-D-Glcp-(1→6)]-D-Glcp) has been reported to be formed by DSR-S from Lc. mesenteroides B-512F (Fu and Robyt 1990). To determine whether the minor trisaccharide derived from maltose varied with the enzyme source, the commercial Lc. mesenteroides enzyme (results not shown) was compared to WcE392-rDSR and WcCab3-DSR in maltose reaction. Their product profiles were similar in HPAEC-PAD analysis, showing that the ability to synthesize centose (instead of forkose) and branched IMOs from maltose is shared by these dextransucrases. The finding was consistent with the use of maltose acceptor product patterns to classify glucansucrases of different specificities (Côté and Leathers 2005). For dextransucrases producing typical dextrans (with predominantly α-(1→6) linkages and a low degree of α-(1→3) branching), their GLOS products from maltose showed a unique pattern, although this was based on only principal IMO series detected in TLC analysis (Côté and Leathers 2005). This thesis demonstrates that the dextransucrases highly specific for α-(1→6) linkage formation also produce the same minor IMO series.

The shared specificity between the two Weissella dextransucrases is consistent with the close resemblance in their sequences (93% amino acid sequence identity; GenBank: AKE50934 and AHU88292). The amino acid substitutions are unlikely to cause drastic differences in their catalytic properties (others’ results in article I). According to the latest constructed phylogenetic tree of GH70 enzymes (Vuillemin et al., 2016), WcE392-rDSR and W. confusa LBAE C39-2 dextransucrase (with 100% amino acid identity to WcCab3-DSR) cluster together and form a branch with a subgroup of two W. cibaria dextransucrases and a Lactobacillus fermentum dextransucrase. Although DSR- S from Lc. mesenteroides NRRL B-512F is separated from Weissella dextransucrases, and its sequence (GenBank: AAD10952) bears only 52% amino acid similarity to the

59 two Weissella dextransucrases studied, they seem to belong to a phylogenetic group of typical dextransucrases producing predominantly α-(1→6) linkages. This corresponds to the similar selectivity and specificity observed for the three dextransucrases.

6.2.2 Specificity to form an α-(1→2) branch (III and IV) A typical dextransucrase DSR-S was shown to form an α-(1→2)-branched trisaccharide from β-(1→4)-linked lactose or cellobiose but form an α-(1→6) linkage on the nonreducing end of α-linked acceptors (Robyt 1995). However, this thesis reports for the first time that α-linked maltose can be α-(1→2) branched as well and form centose via typical dextransucrases. The higher DP homologues were proposed to be IMOs carrying a single-unit α-(1→2) branch at the reducing end. This correlates with an earlier observation that the α-(1→2) branch added to a maltose derivative, whose C-6 of the nonreducing end was blocked, could not grow further (Bhattacharjee and Mayer 1993). Such branched products have not been previously described from the acceptor reaction of an enzyme of GH family 70. Dextransucrase from Lc. citreum NRRL B-1299 and its truncated form GBD-CD2 (a branching sucrase) are able to form α-(1→2) branches on α-(1→6)-linked residues in GLOSs (Dols et al., 1998; Brison et al., 2013). Nevertheless, in the α-(1→2)-branched IMOs formed by typical dextransucrases, the branch is located on the reducing-end residue.

Based on the observation that centose did not serve as an acceptor, which ruled out the possibility of main chain elongation after branch formation, the higher DP homologues were proposed to be formed through direct branch formation on an elongated linear IMO. In this thesis, the dextransucrases have been shown to form an α-(1→2) branch not only on lactose and cellobiose but also on maltose, panose and higher linear IMOs. Thus, these acceptor sugars interact with the enzymes so that the C-2 of the reducing residue can be oriented toward the active site and be linked to the C-1 of the glucosyl-enzyme intermediate. The resultant branched GLOSs, however, were expectedly no longer accommodated in the enzyme acceptor site to form a regular α-(1→6) linkage at the nonreducing end. This complies with the fact that the branched trisaccharides derived from maltose (this study) and cellobiose (Ruiz-Matute et al., 2011) do not elongate further. GLOSs of only DP3 and DP4 were observed to contain lactose under the 13 glc synthesis with C6 -labeled sucrose (Figure 14). Although other lactose-derived GLOSs were too low in amount to be characterized in this and previous studies (Díez- Municio et al., 2012b), a portion of lactose is expected to be elongated linearly via α- 13 glc (1→6) linkages, explaining the presence of the C6 -labeled lactose-containing DP4 product. In the isolation of Lac3 or Cel3 by gel filtration, they were found coeluting closely with a small amount of another product (not shown), which may be the linear

60 trisaccharide with an α-(1→6)-linked glucosyl residue. It is interesting to note that although typical dextransucrases form α-(1→3) branches in dextran, no α-(1→3) linkage has been found in GLOSs. The recognition of α-(1→2)-branch linkage specificity sheds light on a new interaction mode between acceptors and the enzyme active site. Enzyme structure-guided mutations may alter dextransucrase specificity to produce a higher portion of α-(1→2)-branched GLOSs for novel applications.

6.2.3 Novel sucrose-containing trisaccharide (III) In the WcE392-rDSR acceptor reaction in the presence of lactose, the production of sucrose-containing isomelezitose (NR3) by a dextransucrase was described for the first time. It was independent from the exogenous acceptor and was promoted by higher concentrations of sucrose and dextransucrase. Previously, an unknown trisaccharide product was noticed in the lactose reaction mixture of Lc. mesenteroides B-512F dextransucrase, which may have been isomelezitose (Díez-Municio et al., 2012b). This product may have also been formed in the presence of cellobiose and other acceptors tested. However, the reaction rate for isomelezitose formation is much lower compared to that for effective acceptors to form IMO.

Isomelezitose was previously obtained from the fructose acceptor reaction of an alternansucrase, where two possible pathways were proposed for its synthesis (Côté et al., 2008). One involves glucosyl transfer to the 2-OH of the reducing residue of isomaltulose (α-D-Glcp-(1→6)-D-Fruf, an acceptor product of fructose), resulting in the formation of a sucrose-type (β-D-Fruf-(2→1)-α-D-Glcp) linkage. The other pathway involves the transfer of a glucosyl residue to the 6-OH of the fructofuranosyl residue of another sucrose molecule. In this thesis, isomelezitose was hypothesized to be formed by the first pathway i.e., as an acceptor product of isomaltulose. Some supportive evidence has been reported. First, a sucrose-type linkage is formed on lactulose (β-D- Galp-(1→4)-D-Fruf) and maltulose (α-D-Glcp-(1→4)-D-Fruf) by Lc. mesenteroides dextransucrase (Demuth et al., 2002; Díez-Municio et al., 2012a), indicating that a similar sucrose-type linkage could be formed on isomaltulose. Second, a major imbalance existed between the consumption of isomaltulose and the yield of isomaltulose-derived IMOs according to the data published on the isomaltulose acceptor reaction of Lc. mesenteroides dextransucrase (Barea-Alvarez et al., 2014), indicating that other compounds may have also been produced from isomaltulose. Third, if sucrose acts as an acceptor, a glucosyl residue is added to the sucrose-glucosyl moiety (Moulis et al., 2006b; Dobruchowska et al., 2013) and not on the fructosyl moiety to form 13 glc isomelezitose. To test the hypothesis, an acceptor reaction with isomaltulose and C6 - labeled sucrose can be carried out. According to the hypothesis, the two glucosyl

61 residues in the isomelezitose thus synthesized can be differentiated in MSn analysis. By reference to the fragmentation pattern of isomelezitose reported in article III, one should be able to prove its origin from isomaltulose.

Other sucrose analogs, β-D-Galp-(1→4)-β-D-Fruf-(2→1)-α-D-Glcp (lactulosucrose) and α-D-Glcp-(1→4)-β-D-Fruf-(2→1)-α-D-Glcp, derived from lactulose and maltulose, respectively, have been considered donor substrates for dextransucrase after the consumption of sucrose (Demuth et al., 2002; Díez-Municio et al., 2012a). Additional study is needed to test the possible capacity of isomelezitose to act as a donor substrate. Because isomelezitose derives from endogenous sugars in acceptor reactions of both alternansucrase (Côté et al., 2008) and dextransucrase, it may widely occur as a side product among glucansucrases.

6.3 Potential of dextransucrases in prebiotic GLOS synthesis

6.3.1 Synthesis of prebiotic IMOs from maltose (IV) The dextransucrase maltose acceptor reaction represents a promising synthetic pathway for prebiotic IMOs because of the cost-effectiveness and the abundance of the substrates used (sucrose and maltose), as well as its ability to control the size distribution of IMOs (Su and Robyt 1993; Lee et al., 2008). Compared to classical IMO production via a fungal α-transglucosidase on high-maltose syrup, which preferentially produces low-DP IMOs (Goffin et al., 2011; Hu et al., 2013), the dextransucrase-catalyzed reaction can be tailored for the synthesis of higher-DP IMOs. A slight increase in DP (from 2.7 to 3.3) endowed IMOs with non-digestibility for their function as prebiotics (Iwaya et al., 2012). Moreover, in the presence of the effective acceptor maltose, dextran synthesis can be largely inhibited to maximize sucrose utilization for IMO synthesis (Su and Robyt 1993; Rabelo et al., 2009).

In this thesis, the effects of reaction factors (concentrations of sucrose, maltose, and dextransucrase) on the IMO product profile were studied for the first time by optimization design. A high sucrose concentration was favorable for IMO yield. Enzyme dosage showed a small effect on the yield and profile of IMOs over a prolonged reaction time for maximal sucrose conversion. Maltose input, however, had varying effects on the production of IMO components. It promoted the production of lower DP while inhibiting that of DP6. This indicates that maltose has a higher affinity to dextransucrase than do the higher-DP IMOs. Thus, only when maltose concentration was low, did the intermediate IMO products have a better chance to act as acceptors. The total IMO yield

62 peaked when the sucrose and maltose inputs were at their highest. This was also observed in a previous study with Lc. mesenteroides NRRL B-512F dextransucrase, in which the IMO size distribution and maltose consumption were unfortunately not reported (Rabelo et al., 2009). However, this thesis showed that for a good yield of higher-DP IMOs, medium maltose (~0.5 M) and high sucrose (1 M) concentrations were optimal and also resulted in a better maltose utilization.

In vitro fermentation by human fecal bacteria showed that linear maltose-based IMOs with DP5−7 had a relatively high selectivity toward beneficial bacteria compared to those with a lower DP (Sanz et al., 2006). The higher IMOs also persist longer in the colon and can reach the most distal regions, where most chronic intestinal disorders originate (Sanz et al., 2006). In addition, dextransucrases can simultaneously synthesize minor α-(1→2)-branched IMOs, which may have enhanced resistance to gastrointestinal digestion and selectivity toward probiotic bacteria (Sanz et al., 2005). Thus, the dextransucrase-catalyzed maltose acceptor reaction represents a promising synthetic pathway for prebiotic IMOs. Dextransucrase from Lc. mesenteroides B-512F was successfully immobilized for the repetitive production of IMOs (Tanriseven and Doğan 2002). Several purification methods have been developed to remove unwanted monosaccharides (e.g., glucose and fructose) and disaccharides (e.g., maltose and sucrose) from IMO products, including adsorption separation and selective fermentation by yeasts or bacteria (Crittenden and Playne 2002; Yoon et al., 2003; Pan and Lee 2005; Goffin et al., 2011). Beneficial IMOs can also be produced in situ by bacterial cells during food fermentation (Katina et al., 2009; Cho et al., 2014).

6.3.2 Synthesis of other GLOSs (III) Leuconostoc dextransucrases have been used to synthesize GLOSs from a variety of acceptors, such as lactose, cellobiose, isomaltulose, lactulose, and gentiobiose. These products have been demonstrated in vitro as potential prebiotics (Ruiz-Matute et al., 2011; Coelho et al., 2014; Díez-Municio et al., 2014; Barea-Alvarez et al., 2014; García- Cayuela et al., 2014; Kothari and Goyal 2015). Although these acceptors are less effective than maltose, a reasonable yield of acceptor products can be achieved by increasing the substrate concentrations and the ratio of acceptor to sucrose, which diverts more sucrose for use in the acceptor reaction at the cost of dextran and leucrose production (Díez-Municio et al., 2012b). In the Lc. mesenteroides B-512F dextransucrase-catalyzed synthesis of 2Glc-α-Glcp-lactose (Lac3) using concentrated sucrose and lactose both at 0.88 M, the trisaccharide yield was 130 g/L corresponding to a consumption of approximately 36% of the initial lactose (Díez-Municio et al., 2012b). In this thesis, the lactose product yield was limited by the low lactose input

63 concentration of 0.15 M. The highest Lac3 yield was achieved (with a 27% lactose consumption) when the sucrose and dextransucrase concentrations were at their highest, which resulted in the concomitant formation of a large amount of dextran. Under these conditions for lactose acceptor reaction, a novel side product, isomelezitose (NR3), was identified and is a potential nutraceutical (Görl et al., 2012).

Dextransucrase-synthesized 2Glc-α-Glcp-lactose has found use as a precursor for the synthesis of kojibiose, a bioactive disaccharide with low availability due to difficulties in its isolation or synthesis. After the initial lactose acceptor reaction, the sugars present in the mixture (fructose, glucose, galactose, and sucrose) were removed by yeast Saccharomyces cerevisiae treatment. The remaining lactose and 2Glc-α-Glcp-lactose were then hydrolyzed by Kluyveromyces lactis β-galactosidase without removing the yeast cells. Kojibiose was obtained with good yield: 38% in weight with respect to the initial amount of lactose (Díez-Municio et al., 2014). Dextransucrase acceptor reactions may provide affordable synthesis processes for a variety of rare di- and oligosaccharides, thus expanding their applications.

64 7. CONCLUSION

In this thesis, two W. confusa dextransucrases (mostly WcE392-rDSR and partly WcCab3-DSR) were studied for their catalytic properties and dextran- and GLOS- synthesizing functionalities. Focus was given to the structural characterization of their GLOS products because the information on dextransucrase-synthesized GLOSs was previously only available for a commercial Lc. mesenteroides dextransucrase.

The effects of pH and temperature on the WcE392-rDSR activity were determined. The enzyme lost activity quickly above 40 °C and was stabilized by glycerol. It showed

Michaelis-Menten kinetics with a KM of 14.7 mM and a Vmax of 8.2 µmol/(mg∙min) as determined by the sucrose radioisotope assay. A similar KM value and a 2.5-fold higher

Vmax were obtained by the reducing-value Nelson-Somogyi assay. Dextrans from the Weissella dextransucrases consisted of 97% α-(1→6) linkages and 3% α-(1→3) linkages, which resembled that from the commercial Lc. mesenteroides B-512F dextransucrase. The dextrans obtained showed high molar masses in the range of 107‒108 g/mol in HPSEC analysis.

The selectivity of WcE392-rDSR towards several acceptors and the acceptor product patterns were similar to those of Lc. mesenteroides B-512F dextransucrase. The most efficient acceptor maltose was tested with Weissella dextransucrases and the Lc. mesenteroides dextransucrase, all exhibiting a similar product profile in HPAEC-PAD analysis. The GLOS products were isolated according to DP for structural characterization. Interestingly, in addition to a series of principal products, there seemed to be another series of minor products. The principal products were identified as linear maltose-based IMOs, which corresponded to the predominant α-(1→6) linkages in dextran products from typical dextransucrases. The minor products were confirmed to form another homologous series, where a single glucosyl residue was α-(1→2) linked to the reducing residue of linear IMOs. Such branched IMOs have not been reported for a glucansucrase. The branch was proposed to be added to each member of the linear IMOs. Unlike maltose, WcE392-rDSR produced dominantly branched trisaccharides with analogous lactose and cellobiose and with a glucosyl residue α-(1→2) linked to the acceptor’s reducing end. Moreover, the side product isomelezitose was proven for the first time to be formed by a dextransucrase. This nonreducing trisaccharide was apparently synthesized from the fructose acceptor product isomaltulose. An MSn 13 glc analysis of the C6 -labeled GLOS products was proven useful in, e.g., distinguishing fructose-containing side products from products derived from exogenous acceptors.

65 To apply WcE392-rDSR for the synthesis of higher-DP linear IMOs for better prebiotic properties, the effects of sucrose, maltose, and dextransucrase concentrations on the IMO product profile were studied here for the first time. According to response surface modelling, the yields of individual IMOs of DP3−6 were all maximized with the highest sucrose input (1 M) but varying maltose inputs (0.15−1 M). The optimal maltose concentration for DP3−4, DP5 and DP6 was at its highest, medium and lowest levels, respectively. The dextransucrase dosage (1–10 U/g sucrose) had a smaller effect on these responses.

Overall, this study examined the catalytic properties of Weissella dextransucrases and the structures of their dextran and GLOS products. Several GLOSs from Weissella dextransucrases were identified, which enhanced the understanding of the catalytic mechanism of these enzymes and more generally of typical dextransucrases producing predominantly α-(1→6) linkages, such as the commercial one from Lc. mesenteroides B-512F. The study also demonstrated the usability of Weissella dextransucrases in synthesizing dextran and oligosaccharides for food use.

66 REFERENCES

Ahmed RZ, Siddiqui K, Arman M, Ahmed N. 2012. Characterization of high molecular weight dextran produced by Weissella cibaria CMGDEX3. Carbohydr. Polym. 90:441- 6.

Al-Sheraji S, Ismail A, Manap MY, Mustafa S, Yusof RM, Hassan FA. 2013. Prebiotics as functional foods: A review. J. Funct. Foods 5:1542-53.

Amari M, Arango LFG, Gabriel V, Robert H, Morel S, Moulis C, Gabriel B, Remaud- Siméon M, Fontagné-Faucher C. 2012. Characterization of a novel dextransucrase from Weissella confusa isolated from sourdough. Appl. Microbiol. Biotechnol. 97:5413-22.

Anwar MA, Kralj S, Pique AV, Leemhuis H, van der Maarel MJEC, Dijkhuizen L. 2010. Inulin and levan synthesis by probiotic Lactobacillus gasseri strains: characterization of three novel fructansucrase enzymes and their fructan products. Microbiology (London, U. K.) 156:1264-74.

Argüello Morales MA, Remaud-Simeon M, Willemot R, Vignon MR, Monsan P. 2001. Novel oligosaccharides synthesized from sucrose donor and cellobiose acceptor by alternansucrase. Carbohydr. Res. 331:403-11.

Ayeni FA, Sánchez B, Adeniyi BA, Clara G, Margolles A, Ruas-Madiedo P. 2011. Evaluation of the functional potential of Weissella and Lactobacillus isolates obtained from Nigerian traditional fermented foods and cow's intestine. Int. J. Food Microbiol. 147:97-104.

Bai Y, van der Kaaij RM, Leemhuis H, Pijning T, van Leeuwen SS, Jin Z, Dijkhuizen L. 2015. Biochemical characterization of the Lactobacillus reuteri glycoside hydrolase family 70 GTFB type of 4,6-α-glucanotransferase enzymes that synthesize soluble dietary starch fibers. Appl. Environ. Microbiol. 81:7223-32.

Barea-Alvarez M, Benito MT, Olano A, Jimeno ML, Moreno FJ. 2014. Synthesis and characterization of isomaltulose-derived oligosaccharides produced by transglucosylation reaction of Leuconostoc mesenteroides dextransucrase. J. Agric. Food Chem. 62:9137-44.

Basedow AM, Ebert KH. 1979. Production, characterization, and solution properties of dextran fractions of narrow molecular weight distributions. J. Polym. Sci., Polym. Symp. 66:101-15.

Bejar W, Gabriel V, Amari M, Morel S, Mezghani M, Maguin E, Fontagné-Faucher C, Bejar S, Chouayekh H. 2013. Characterization of glucansucrase and dextran from Weissella sp. TN610 with potential as safe food additives. Int. J. Biol. Macromol. 52:125-32.

Bhattacharjee MK, Mayer RM. 1993. Formation of alpha-(1→6), alpha-(1→3), and alpha-(1→2) glycosidic linkages by dextransucrase from Streptococcus sanguis in acceptor-dependent reactions. Carbohydr. Res. 242:191-201.

67 Boels I, van Kranenburg R, Hugenholtz J, Kleerebezem M, de Vos W. 2001. Sugar catabolism and its impact on the biosynthesis and engineering of exopolysaccharide production in lactic acid bacteria. Int. Dairy J. 11:723-32.

Bounaix MS, Gabriel V, Morel S, Robert H, Rabier P, Remaud-Siméon M, Gabriel B, Fontagné-Faucher C. 2009. Biodiversity of exopolysaccharides produced from sucrose by sourdough lactic acid bacteria. J. Agric. Food Chem. 57:10889-97.

Bounaix M, Robert H, Gabriel V, Morel S, Remaud-Siméon M, Gabriel B, Fontagné- Faucher C. 2010. Characterization of dextran-producing Weissella strains isolated from sourdoughs and evidence of constitutive dextransucrase expression. FEMS Microbiol. Lett. 311:18-26.

Brison Y, Laguerre S, Lefoulon F, Morel S, Monties N, Potocki-Veronese G, Monsan P, Remaud-Simeon M. 2013. Branching pattern of gluco-oligosaccharides and 1.5 kDa dextran grafted by the α-1,2 branching sucrase GBD-CD2. Carbohydr. Polym. 94:567- 76.

Brison Y, Pijning T, Malbert Y, Fabre É, Mourey L, Morel S, Potocki-Véronèse G, Monsan P, Tranier S, Remaud-Siméon M, Dijkstra BW. 2012. Functional and structural characterization of α-(1→2) branching sucrase derived from DSR-E glucansucrase. J. Biol. Chem. 287:7915-24.

Bubb WA. 2003. NMR spectroscopy in the study of carbohydrates: Characterizing the structural complexity. Concepts Magn. Reson., Part A 19:1-19.

Cho SK, Eom H, Moon JS, Lim S, Kim YK, Lee KW, Han NS. 2014. An improved process of isomaltooligosaccharide production in kimchi involving the addition of a Leuconostoc starter and sugars. Int. J. Food Microbiol. 170:61-4.

Chung C, Day D. 2002. Glucooligosaccharides from Leuconostoc mesenteroides B-742 (ATCC 13146): a potential prebiotic. J. Ind. Microbiol. Biotechnol. 29:196-9.

Coelho R, Araújo A, Fontes C, da Silva A, da Costa J, Rodrigues S. 2014. Powder lemon juice containing oligosaccharides obtained by dextransucrase acceptor reaction synthesis and dehydrated in sprouted bed. J. Food Sci. Technol. (New Delhi, India) 52:5961-7.

Collins MD, Samelis J, Metaxopoulos J, Wallbanks S. 1993. Taxonomic studies on some leuconostoc-like organisms from fermented sausages: description of a new genus Weissella for the Leuconostoc paramesenteroides group of species. J. Appl. Microbiol. 75:595-603.

Corradini C, Cavazza A, Bignardi C. 2012. High-performance anion-exchange chromatography coupled with pulsed electrochemical detection as a powerful tool to evaluate carbohydrates of food interest: Principles and applications. Int. J. Carbohydr. Chem. 10.1155/2012/487564.

Côté GL, Cormier RS, Vermillion KE. 2014. Glucansucrase acceptor reactions with D- mannose. Carbohydr. Res. 387:1-3.

68 Côté GL, Dunlap CA, Vermillion KE. 2009. Glucosylation of raffinose via alternansucrase acceptor reactions. Carbohydr. Res. 344:1951-9.

Côté GL, Leathers TD. 2005. A method for surveying and classifying Leuconostoc spp. glucansucrases according to strain-dependent acceptor product patterns. J. Ind. Microbiol. Biotechnol. 32:53-60.

Côté GL, Robyt JF. 1982. Isolation and partial characterization of an extracellular glucansucrase from Leuconostoc mesenteroides NRRL B-1355 that synthesizes an alternating (1→ 6), (1→ 3)-α-D-glucan. Carbohydr. Res. 101:57-74.Côté GL, Sheng S, Dunlap CA. 2008. Alternansucrase acceptor products. Biocatal. Biotransform. 26:161- 8.

Côté GL, Skory CD. 2015. Water-insoluble glucans from sucrose via glucansucrases. Factors influencing structures and yields. In: H. N. Cheng, R. A. Gross, P. B. Smith, editors. Green polymer chemistry: biobased materials and biocatalysis III. ACS Symposium Series, Washington, DC. p 101-12.

Côté GL, Skory CD. 2012. Cloning, expression, and characterization of an insoluble glucan-producing glucansucrase from Leuconostoc mesenteroides NRRL B-1118. Appl. Microbiol. Biotechnol. 93:2387-94.

Crittenden RG, Playne M. 2002. Purification of food-grade oligosaccharides using immobilised cells of Zymomonas mobilis. Appl. Microbiol. Biotechnol. 58:297-302.

Demuth K, Jördening H, Buchholz K. 2002. Oligosaccharide synthesis by dextransucrase: new unconventional acceptors. Carbohydr. Res. 337:1811-20.

Di Cagno R, De Angelis M, Limitone A, Minervini F, Carnevali P, Corsetti A, Gaenzle M, Ciati R, Gobbetti M. 2006. Glucan and fructan production by sourdough Weissella cibaria and Lactobacillus plantarum. J. Agric. Food Chem. 54:9873-81.

Di Cagno R, Coda R, De Angelis M, Gobbetti M. 2013. Exploitation of vegetables and fruits through lactic acid fermentation. Food Microbiol. 33:1-10.

Díez-Municio M, Montilla A, Moreno FJ, Herrero M. 2014. A sustainable biotechnological process for the efficient synthesis of kojibiose. Green Chem. 16:2219- 26.

Díez-Municio M, Herrero M, Jimeno ML, Olano A, Moreno FJ. 2012a. Efficient synthesis and characterization of lactulosucrose by Leuconostoc mesenteroides B-512F dextransucrase. J. Agric. Food Chem. 60:10564-71.

Díez-Municio M, Montilla A, Jimeno ML, Corzo N, Olano A, Moreno FJ. 2012b. Synthesis and characterization of a potential prebiotic trisaccharide from cheese whey permeate and sucrose by Leuconostoc mesenteroides dextransucrase. J. Agric. Food Chem. 60:1945-53.

Dobruchowska JM, Meng X, Leemhuis H, Gerwig GJ, Dijkhuizen L, Kamerling JP. 2013. Gluco-oligomers initially formed by the reuteransucrase enzyme of Lactobacillus

69 reuteri 121 incubated with sucrose and malto-oligosaccharides. Glycobiology 23:1084- 96.

Dols M, Simeon MR, Willemot RM, Vignon MR, Monsan PF. 1998. Structural characterization of the maltose acceptor-products synthesized by Leuconostoc mesenteroides NRRL B-1299 dextransucrase. Carbohydr. Res. 305:549-59.

Domon B, Costello CE. 1988. A systematic nomenclature for carbohydrate fragmentations in FAB-MS/MS spectra of glycoconjugates. Glycoconj. J. 5:397-409.

Dubois M, Gilles KA, Hamilton JK, Rebers PA, Smith F. 1956. Colorimetric method for determination of sugars and related substances. Anal. Chem. 28:350-6.

European Food Safety Authority (EFSA). 2011. Scientific opinion on the substantiation of health claims related to fructose and reduction of post-prandial glycaemic responses (ID 558) pursuant to Article 13 (1) of Regulation (EC) no 1924/2006. EFSA J. 9:2223- 38.

Fang Y, Wu J, Xu Z. 2015. Dextransucrase-catalyzed elongation of polysaccharide brushes with immobilized mono-/di-saccharides as acceptors. Chem. Commun. (Cambridge, U. K.) 51:129-32.

Fu D, Robyt JF. 1990. Acceptor reactions of maltodextrins with Leuconostoc mesenteroides B-512FM dextransucrase. Arch. Biochem. Biophys. 283:379-87.

Fusco V, Quero GM, Cho G, Kabisch J, Meske D, Neve H, Bockelmann W, Franz CMAP. 2015. The genus Weissella: taxonomy, ecology and biotechnological potential. Front. Microbiol. 6:10.3389/fmicb.2015.00155.

Galle S, Schwab C, Arendt E, Gänzle M. 2010. Exopolysaccharide-forming Weissella strains as starter cultures for sorghum and wheat sourdoughs. J. Agric. Food Chem. 58:5834-41.

Galle S, Schwab C, Dal Bello F, Coffey A, Ganzle MG, Arendt EK. 2012. Influence of in-situ synthesized exopolysaccharides on the quality of gluten-free sorghum sourdough bread. Int. J. Food Microbiol. 155:105-12.

García-Cayuela T, Díez-Municio M, Herrero M, Martínez-Cuesta MC, Peláez C, Requena T, Moreno FJ. 2014. Selective fermentation of potential prebiotic lactose- derived oligosaccharides by probiotic bacteria. Int. Dairy J. 38:11-5.

Germaine GR, Schachtele CF, Chludzinski AM. 1974. Rapid filter paper assay for the dextransucrase activity from Streptococcus mutans. J. Dent. Res. 53:1355-60.

Goffin D, Delzenne N, Blecker C, Hanon E, Deroanne C, Paquot M. 2011. Will isomalto-oligosaccharides, a well-established functional food in Asia, break through the European and American market? The status of knowledge on these prebiotics. Crit. Rev. Food Sci. Nutr. 51:394-409.

70 Görl J, Timm M, Seibel J. 2012. Mechanism-Oriented Redesign of an Isomaltulose Synthase to an Isomelezitose Synthase by Site-Directed Mutagenesis. ChemBioChem 13:149-56.

Hansen PI, Larsen FH, Motawia SM, Blennow A, Spraul M, Dvortsak P, Engelsen SB. 2008. Structure and hydration of the amylopectin trisaccharide building blocks– Synthesis, NMR, and molecular dynamics. Biopolymers 89:1179-93.

Harvey DJ. 2012. Analysis of carbohydrates and glycoconjugates by matrix-assisted laser desorption/ionization mass spectrometry: An update for 2007-2008. Mass Spectrom. Rev. 31:183-311.

Holt S, Teresi JM, Côté GL. 2008. Influence of alternansucrase-derived oligosaccharides and other carbohydrates on α-galactosidase and α-glucosidase activity in Bifidobacterium adolescentis. Lett. Appl. Microbiol. 46:73-9.

Hu Y, Ketabi A, Buchko A, Gänzle MG. 2013. Metabolism of isomalto- oligosaccharides by Lactobacillus reuteri and bifidobacteria. Lett. Appl. Microbiol. 57:108-14.

Irague R, Tarquis L, André I, Moulis C, Morel S, Monsan P, Potocki-Véronèse G, Remaud-Siméon M. 2013. Combinatorial engineering of dextransucrase specificity. PLoS ONE 8:e77837.

Irague R, Rolland-Sabaté A, Tarquis L, Doublier JL, Moulis C, Monsan P, Remaud- Siméon M, Potocki-Véronèse G, Buléon A. 2012. Structure and property engineering of α-D-glucans synthesized by dextransucrase mutants. Biomacromolecules 13:187-95.

Ito K, Ito S, Shimamura T, Weyand S, Kawarasaki Y, Misaka T, Abe K, Kobayashi T, Cameron AD, Iwata S. 2011. Crystal structure of glucansucrase from the dental caries pathogen Streptococcus mutans. J. Mol. Biol. 408:177-86.

Iwaya H, Jae-Sung L, Yamagishi S, Shinoki A, Lang W, Thawornkuno C, Hee-Kwon K, Kumagai Y, Suzuki S, Kitamura S, Hara H, Okuyama M, Mori H, Kimura A, Ishizuka S. 2012. The delay in the development of experimental colitis from isomaltosyloligosaccharides in rats is dependent on the degree of polymerization. PLoS ONE 7:e50658.

Jeanes A, Haynes WC, Wilham C, Rankin JC, Melvin E, Austin MJ, Cluskey J, Fisher B, Tsuchiya H, Rist C. 1954. Characterization and classification of dextrans from ninety- six strains of bacteria. J. Am. Chem. Soc. 76:5041-52.

Jung SM, Mayer RM. 1981. Dextransucrase: Donor substrate reactions. Arch. Biochem. Biophys. 208:288-95.

Juvonen R, Honkapää K, Maina NH, Shi Q, Viljanen K, Maaheimo H, Virkki L, Tenkanen M, Lantto R. 2015. The impact of fermentation with exopolysaccharide producing lactic acid bacteria on rheological, chemical and sensory properties of pureed carrots (Daucus carota L.). Int. J. Food Microbiol. 207:109-18.

71 Kang H, Oh J, Kim D. 2009. Molecular characterization and expression analysis of the glucansucrase DSRWC from Weissella cibaria synthesizing a α(1→6) glucan. FEMS Microbiol. Lett. 292:33-41.

Katina K, Maina NH, Juvonen R, Flander L, Johansson L, Virkki L, Tenkanen M, Laitila A. 2009. In situ production and analysis of Weissella confusa dextran in wheat sourdough. Food Microbiol. 26:734-43.

Ketabi A, Dieleman LA, Gänzle MG. 2011. Influence of isomalto-oligosaccharides on intestinal microbiota in rats. J. Appl. Microbiol. 110:1297-306.

Kim M, Day DF, Kim D. 2010. Potential physiological functions of acceptor products of dextransucrase with cellobiose as an inhibitor of mutansucrase and fungal cell synthase. J. Agric. Food Chem. 58:11493-500.

Kim G, Kang H, Seo E, Jung S, Park J, Kim D, Kim D, Ahn S, Sunwoo C, Kim D. 2012. Glucosylation of the flavonoid, astragalin by Leuconostoc mesenteroides B-512FMCM dextransucrase acceptor reactions and characterization of the products. Enzyme Microb. Technol. 50:50-6.

Kothari D, Goyal A. 2013. Structural characterization of enzymatically synthesized dextran and oligosaccharides from Leuconostoc mesenteroides NRRL B-1426 dextransucrase. Biochemistry (Moscow) 78:1164-70.

Kothari D, Goyal A. 2015. Gentio-oligosaccharides from Leuconostoc mesenteroides NRRL B-1426 dextransucrase as prebiotics and as a supplement for functional foods with anti-cancer properties. Food Funct. 6:604-11.

Lacaze G, Wick M, Cappelle S. 2007. Emerging fermentation technologies: Development of novel sourdoughs. Food Microbiol. 24:155-60.

Lee CY, Yun CY, Yun JW, Oh TK, Kim CJ. 1997. Production of glucooligosaccharides by an acceptor reaction using two types of glucansucrase from Streptococcus sobrinus. Biotechnol. Lett. 19:1227-30.

Lee KW, Park JY, Jeong HR, Heo HJ, Han NS, Kim JH. 2012. Probiotic properties of Weissella strains isolated from human faeces. Anaerobe 18:96-102.

Lee MS, Cho SK, Eom HJ, Kim SY, Kim TJ, Han NS. 2008. Optimized substrate concentrations for production of long-chain isomaltooligosaccharides using dextransucrase of Leuconostoc mesenteroides B-512F. J. Microbiol. Biotechnol. 18:1141-5.

Leemhuis H, Pijning T, Dobruchowska JM, van Leeuwen SS, Kralj S, Dijkstra BW, Dijkhuizen L. 2013. Glucansucrases: Three-dimensional structures, reactions, mechanism, α-glucan analysis and their implications in biotechnology and food applications. J. Biotechnol. 163:250-72.

72 Liang C, Zhang Y, Jia Y, Wenzhao W, Li Y, Lu S, Jin JM, Tang SY. 2016. Engineering a carbohydrate-processing transglycosidase into glycosyltransferase for natural product glycodiversification. Sci. Rep. 6:10.1038/srep21051.

Lombard V, Golaconda Ramulu H, Drula E, Coutinho PM, Henrissat B. 2014. The carbohydrate-active enzymes database (CAZy) in 2013. Nucleic Acids Res. 42:D490-5.

Maina NH, Juvonen M, Domingues RM, Virkki L, Jokela J, Tenkanen M. 2013. Structural analysis of linear mixed-linkage glucooligosaccharides by tandem mass spectrometry. Food Chem. 136:1496-507.

Maina NH. 2012. Structure and macromolecular properties of Weissella confusa and Leuconostoc citreum dextrans with a potential application in sourdough. [dissertation]. Finland: Univeristy of Helsinki. ISBN:978-952-10-7984-9.

Maina NH, Pitkänen L, Heikkinen S, Tuomainen P, Virkki L, Tenkanen M. 2014. Challenges in analysis of high-molar mass dextrans: Comparison of HPSEC, AsFlFFF and DOSY NMR spectroscopy. Carbohydr. Polym. 99:199-207.

Maina NH, Virkki L, Pynnönen H, Maaheimo H, Tenkanen M. 2011. Structural analysis of enzyme-resistant isomaltooligosaccharides reveals the elongation of α-(1→3)-linked branches in Weissella confusa dextran. Biomacromolecules 12:409-18.

Maina NH, Tenkanen M, Maaheimo H, Juvonen R, Virkki L. 2008. NMR spectroscopic analysis of exopolysaccharides produced by Leuconostoc citreum and Weissella confusa. Carbohydr. Res. 343:1446-55.

Majumder A, Purama RK, Goyal A. 2007. An overview of purification methods of glycoside hydrolase family 70 dextransucrase. Indian J. Microbiol. 47:197-206.

Malang SK, Maina NH, Schwab C, Tenkanen M, Lacroix C. 2015. Characterization of exopolysaccharide and ropy capsular polysaccharide formation by Weissella. Food Microbiol. 46:418-27.

Malik A, Radji M, Kralj S, Dijkhuizen L. 2009. Screening of lactic acid bacteria from Indonesia reveals glucansucrase and fructansucrase genes in two different Weissella confusa strains from soya. FEMS Microbiol. Lett. 300:131-8.

McIntyre AP, Mukerjea R, Robyt JF. 2013. Reducing values: dinitrosalicylate gives over-oxidation and invalid results whereas copper bicinchoninate gives no over- oxidation and valid results. Carbohydr. Res. 380:118-23.

Meng X, Dobruchowska JM, Pijning T, Gerwig GJ, Dijkhuizen L. 2015a. Synthesis of new hyper-branched α-glucans from sucrose by Lactobacillus reuteri 180 glucansucrase mutants. J. Agric. Food Chem. 64:433-42.

Meng X, Dobruchowska JM, Pijning T, Gerwig GJ, Kamerling JP, Dijkhuizen L. 2015b. Truncation of domain V of the multidomain glucansucrase GTF180 of Lactobacillus reuteri 180 heavily impairs its polysaccharide-synthesizing ability. Appl. Microbiol. Biotechnol. 99:5885-94.

73 Meng X, Pijning T, Dobruchowska JM, Gerwig GJ, Dijkhuizen L. 2015c. Characterization of the functional roles of amino acid residues in acceptor-binding subsite +1 in the active site of the glucansucrase GTF180 from Lactobacillus reuteri 180. J. Biol. Chem. 290:30131-41.

Miao M, Jia X, Jiang B, Wu S, Cui SW, Li X. 2016. Elucidating molecular structure and prebiotics properties of bioengineered α-D-glucan from Leuconostoc citreum SK24.002. Food Hydrocoll. 54:227-33.

Mischnick P. 2012. Mass spectrometric characterization of oligo- and polysaccharides and their derivatives. In: M. Hakkarainen, editor. Mass Spectrometry of Polymers–New Techniques. Berlin Heidelberg: Springer. p 105-74.

Monsan P, Remaud-Simeon M, Andre I. 2010. Transglucosidases as efficient tools for oligosaccharide and glucoconjugate synthesis. Curr. Opin. Microbiol. 13:293-300.

Moulis C, Joucla G, Harrison D, Fabre E, Potocki-Veronese G, Monsan P, Remaud- Simeon M. 2006b. Understanding the polymerization mechanism of glycoside- hydrolase family 70 glucansucrases. J. Biol. Chem. 281:31254-67.

Moulis C, Arcache A, Escalier P, Rinaudo M, Monsan P, Remaud-Simeon M, Potocki- Veronese G. 2006a. High-level production and purification of a fully active recombinant dextransucrase from Leuconostoc mesenteroides NRRL B-512F. FEMS Microbiol. Lett. 261:203-10.

Naessens M, Cerdobbel A, Soetaert W, Vandamme EJ. 2005. Leuconostoc dextransucrase and dextran: Production, properties and applications. J. Chem. Technol. Biotechnol. 80:845-60.

Nelson N. 1944. A photometric adaptation of the Somogyi method for the determination of glucose. J. Biol. Chem. 153:375-80.

Nihira T, Mori T, Asakura M, Okahata Y. 2011. Kinetic studies of dextransucrase enzyme reactions on a substrate- or enzyme-immobilized 27 MHz quartz crystal microbalance. Langmuir 27:2107-11.

Olvera C, Centeno-Leija S, López-Munguía A. 2007. Structural and functional features of fructansucrases present in Leuconostoc mesenteroides ATCC 8293. Antonie Van Leeuwenhoek 92:11-20.

Pan YC, Lee WC. 2005. Production of high-purity isomalto-oligosaccharides syrup by the enzymatic conversion of transglucosidase and fermentation of yeast cells. Biotechnol. Bioeng. 89:797-804.

Passerini D, Vuillemin M, Ufarté L, Morel S, Loux V, Fontagné-Faucher C, Monsan P, Remaud-Siméon M, Moulis C. 2015. Inventory of the GH70 enzymes encoded by Leuconostoc citreum NRRL B-1299 – identification of three novel α-transglucosylases. FEBS J. 282:2115-30.

74 Pijning T, Vujicic-Zagar A, Kralj S, Dijkhuizen L, Dijkstra B. 2012. Structure of the α- 1,6/α-1,4-specific glucansucrase GTFA from Lactobacillus reuteri 121. Acta Crystallogr., Sect. F: Struct. Biol. Cryst. Commun. 68:1448-54.

Pitkänen L. 2011. The effect of structure on the dilute solution properties of branched polysaccharides studied with SEC and AsFlFFF. [dissertation]. Finland: University of Helsinki. ISBN:978-952-10-7351-9.

Plou FJ, Martín MT, de Segura AG, Alcalde M, Ballesteros A. 2002. Glucosyltransferases acting on starch or sucrose for the synthesis of oligosaccharides. Can. J. Chem. 80:743-52.

Purama RK, Goyal A. 2008b. Identification, effective purification and functional characterization of dextransucrase from Leuconostoc mesenteroides NRRL B-640. Bioresour. Technol. 99:3635-42.

Purama RK, Goyal A. 2008a. Application of response surface methodology for maximizing dextransucrase production from Leuconostoc mesenteroides NRRL B-640 in a bioreactor. Appl. Biochem. Biotechnol. 151:182-92.

Rabelo MC, Honorato T, Goncalves L, Pinto G, Rodrigues S. 2009. Optimization of enzymatic synthesis of isomalto-oligosaccharides production. J. Food Biochem. 33:342- 54.

Rao T, Goyal A. 2013. Purification, optimization of assay, and stability studies of dextransucrase isolated from Weissella cibaria JAG8. Prep. Biochem. Biotechnol. 43:329-41.

Remaud-Slmeon M, Lopez-Munguia A, Pelenc V, Paul F, Monsan P. 1994. Production and use of glucosyltransferases from Leuconostoc mesenteroides NRRL B-1299 for the synthesis of oligosaccharides containing α-(1→2) linkages. Appl. Biochem. Biotechnol. 44:101-17.

Roberfroid M, Gibson GR, Hoyles L, McCartney AL, Rastall R, Rowland I, Wolvers D, Watzl B, Szajewska H, Stahl B, Guarner F, Respondek F, Whelan K, Coxam V, Davicco M, Léotoing L. 2010. Prebiotic effects: metabolic and health benefits. Br. J. Nutr. 104:S1-S63.

Robyt JF. 1995. Mechanisms in the glucansucrase synthesis of polysaccharides and oligosaccharides from sucrose. Adv. Carbohydr. Chem. Biochem. 51:133-68.

Robyt JF, Eklund SH. 1983. Relative, quantitative effects of acceptors in the reaction of Leuconostoc mesenteroides B-512F dextransucrase. Carbohydr. Res. 121:279-86.

Ruas-Madiedo P, de los Reyes-Gavilán CG. 2005. Methods for the screening, isolation, and characterization of exopolysaccharides produced by lactic acid bacteria. J. Dairy Sci. 88:843-56.

Ruas-Madiedo P, Salazar N, de los Reyes-Gavilán, Clara, G. 2009. Biosynthesis and chemical composition of exopolysaccharides produced by lactic acid bacteria. In: M.

75 Ullrich, editor. Bacterial Polysaccharides, Current Innovations and Future Trends. Norfolk, U.K.: Caister Academic Press. p 279-310.

Ruiz-Matute AI, Brokl M, Sanz ML, Soria AC, Côté GL, Collins ME, Rastall RA. 2011. Effect of dextransucrase cellobiose acceptor products on the growth of human gut bacteria. J. Agric. Food Chem. 59:3693-700.

Ryan PM, Ross RP, Fitzgerald GF, Caplice NM, Stanton C. 2015. Sugar-coated: exopolysaccharide producing lactic acid bacteria for food and human health applications. Food Funct. 6:679-93.

Sanz ML, Gibson GR, Rastall RA. 2005. Influence of disaccharide structure on prebiotic selectivity in vitro. J. Agric. Food Chem. 53:5192-9.

Sanz ML, Côté GL, Gibson GR, Rastall RA. 2006. Influence of glycosidic linkages and molecular weight on the fermentation of maltose-based oligosaccharides by human gut bacteria. J. Agric. Food Chem. 54:9779-84.

Sawai T, Tohyama T, Natsume T. 1978. Hydrolysis of fourteen native dextrans by Arthrobacter isomaltodextranase and correlation with dextran structure. Carbohydr. Res. 66:195-205.

Schwab C, Mastrangelo M, Corsetti A, Gänzle M. 2008. Formation of oligosaccharides and polysaccharides by Lactobacillus reuteri LTH5448 and Weissella cibaria 10M in sorghum sourdoughs. Cereal Chem. 85:679-84.

Seibel J, Jördening H, Buchholz K. 2010. Extending synthetic routes for oligosaccharides by enzyme, substrate and reaction engineering. In: Christoph Wittmann, Rainer Krull, editors. Biosystems Engineering I. Springer. p 163-93.

Seo DM, Kim S, Eom H, Han N. 2007. Synbiotic synthesis of oligosaccharides during milk fermentation by addition of Leuconostoc starter and sugars. J. Microbiol. Biotechnol. 17:1758-64.

Seo E, Nam S, Kang H, Cho J, Lee H, Ryu H, Kim D. 2007. Synthesis of thermo- and acid-stable novel oligosaccharides by using dextransucrase with high concentration of sucrose. Enzyme Microb. Technol. 40:1117-23.

Seto H, Kawakita H, Ohto K, Harada H, Inoue K. 2008. Novel carbonyl-group- containing dextran synthesis by pyranose-2-oxidase and dextransucrase. Carbohydr. Res. 343:2417-21.

Shukla S, Goyal A. 2011. Optimization of fermentation medium for enhanced glucansucrase and glucan production from Weissella confusa. Braz. arch. biol. technol. 54:1117-24.

Shukla S, Goyal A. 2011. 16S rRNA-based identification of a glucan-hyperproducing Weissella confusa. Enzyme Res. 10.4061/2011/250842.

76 Shukla S, Verma A, Kajala I, Nyyssolä A, Baruah R, Katina K, Juvonen R, Tenkanen M, Goyal A. 2016. Structure modelling and functional analysis of recombinant dextransucrase from Weissella confusa Cab3 expressed in Lactococcus lactis. Prep. Biochem. Biotechnol. 10.1080/10826068.2016.1141299.

Somogyi M. 1945. A new reagent for the determination of sugars. J. Biol. Chem. 160:61- 8.

Stam MR, Danchin EG, Rancurel C, Coutinho PM, Henrissat B. 2006. Dividing the large glycoside hydrolase family 13 into subfamilies: towards improved functional annotations of α-amylase-related proteins. Protein Eng., Des. Sel. 19:555-62.

Su D, Robyt JF. 1993. Control of the synthesis of dextran and acceptor-products by Leuconostoc mesenteroides B-512FM dextransucrase. Carbohydr. Res. 248:339-48.

Sumner JB, Howell SF. 1935. A method for determination of saccharase activity. J. Biol. Chem. 108:51-4.

Tanriseven A, Doğan Ş. 2002. Production of isomalto-oligosaccharides using dextransucrase immobilized in alginate fibres. Process Biochem. (Oxford, U. K.) 37:1111-5.

Timm M, Görl J, Kraus M, Kralj S, Hellmuth H, Beine R, Buchholz K, Dijkhuizen L, Seibel J. 2013. An unconventional glycosyl transfer reaction: glucansucrase GTFA functions as an allosyltransferase enzyme. Chembiochem 14:2423-6.

Usui T, Ogata M, Murata T, Ichikawa K, Sakano Y, Nakamura Y. 2009. Sequential analysis of α-glucooligosaccharides with α-(1→4) and α-(1→6) linkages by negative ion Q-TOF MS/MS Spectrometry. J. Carbohydr. Chem. 28:421-30. van Leeuwen SS, Leeflang BR, Gerwig GJ, Kamerling JP. 2008a. Development of a H- 1 NMR structural-reporter-group concept for the primary structural characterisation of α-D-glucans. Carbohydr. Res. 343:1114-9. van Leeuwen SS, Kralj S, van Geel-Schutten IH, Gerwig GJ, Dijkhuizen L, Kamerling JP. 2008b. Structural analysis of the α-D-glucan (EPS180) produced by the Lactobacillus reuteri strain 180 glucansucrase GTF180 enzyme. Carbohydr. Res. 343:1237-50.

Vettori MHPB, Franchetti SMM, Contiero J. 2012. Structural characterization of a new dextran with a low degree of branching produced by Leuconostoc mesenteroides FT045B dextransucrase. Carbohydr. Polym. 88:1440-4.

Vettori MHPB, Mukerjea R, Robyt JF. 2011. Comparative study of the efficacies of nine assay methods for the dextransucrase synthesis of dextran. Carbohydr. Res. 346:1077- 82.

Vuillemin M, Claverie M, Brison Y, Severac E, Bondy P, Morel S, Monsan P, Moulis C, Remaud-Simeon M. 2016. Characterization of the first α-(1→3) branching sucrases of the GH70 family. J. Biol. Chem. 291:7687-702.

77 Vujicic-Zagar A, pijnig T, Kralj S, Lopez CA, Eeuwema W, Dijkhuizen L, Dijkstra BW. 2010. Crystal structure of a 117 kDa glucansucrase fragment provides insight into evolution and product specificity of GH70 enzymes. Proc. Natl. Acad. Sci. U. S. A. 107:21406-11.

Waldherr FW, Vogel RF. 2009. Commercial exploitation of homo-exopolysaccharides in non-dairy food systems. In: M. Ullrich, editor. Bacterial Polysaccharides, Current Innovations and Future Trends. Norfolk, U.K.: Caister Academic Press. p 313-30.

Welman AD. 2009. Exploitation of Exopolysaccharides from Lactic Acid Bacteria: Nutritional and Functional Benefits. In: M. Ullrich, editor. Bacterial Polysaccharides, Current Innovations and Future Trends. Norfolk, U.K.: Caister Academic Press. p 331- 40.

Wuhrer M, de Boer AR, Deelder AM. 2009. Structural glycomics using hydrophilic interaction chromatography (HILIC) with mass spectrometry. Mass Spectrom. Rev. 28:192-206.

Yamauchi F, Ohwada Y. 1969. Synthesis of oligosaccharides by growing culture of Leuconostoc mesenteroides. Agric. Biol. Chem. 33:1295-300.

Yen C, Tseng Y, Kuo Y, Lee M, Chen H. 2011. Long-term supplementation of isomalto- oligosaccharides improved colonic microflora profile, bowel function, and blood cholesterol levels in constipated elderly people—a placebo-controlled, diet-controlled trial. Nutrition (N. Y., NY, U. S.) 27:445-50.

Yoon S, Mukerjea R, Robyt JF. 2003. Specificity of yeast (Saccharomyces cerevisiae) in removing carbohydrates by fermentation. Carbohydr. Res. 338:1127-32.

Zannini E, Mauch A, Galle S, Gänzle M, Coffey A, Arendt E, Taylor J, Waters D. 2013. Barley malt wort fermentation by exopolysaccharide-forming Weissella cibaria MG1 for the production of a novel beverage. J. Appl. Microbiol. 115:1379-87.

Zannini E, Waters DM, Coffey A, Arendt EK. 2015. Production, properties, and industrial food application of lactic acid bacteria-derived exopolysaccharides. Appl. Microbiol. Biotechnol. 100:1121-35.

78