Polysaccharide production by lactic acid bacteria: from genes to industrial applications

summary


INTRODUCTION
Polysaccharides or glycans are widespread in nature, and display diverse chemical structures, physical properties and biological functions that are usually either structure or storage related (Ullrich 2009).These polymeric carbohydrate molecules consist of a large number of monosaccharides (generally, >10 units; oligosaccharides, ≤10 units) bound together by glycosydic linkages, and thus are characterized by high molecular weight.In contrast to DNA, RNA or protein synthesis, polysaccharide synthesis is not a template-driven process, but instead the structures are determined primarily by the complement of polysaccharide-modifying enzymes present in any organism.Polysaccharides can range in structure from linear to highly branched, and in composition they can be classified as homopolysaccharides (HoPS) when composed of only one type of monosaccharide or heteropolysaccharides (HePS) if two or more types of sugars are present.Variations in the sugar building blocks, anomeric configuration, glycosidic linkage, nonsugar decorations of the monosaccharides, conformation and molecular weight result in an enormous diversity of polysaccharides.This structural diversity contributes to the ensemble of biological functions and renders an array of physicochemical and rheological properties that can be exploited for varied commercial applications in the industrial, food and medical sectors (Ullrich 2009;Schmid, Sieber and Rehm 2015).Even though the global market is so far dominated by polysaccharides produced by plants and algae (e.g.pectin, cellulose, alginate), bacteria represent a relatively untapped source of an immense polysaccharide repertoire.
Bacteria can synthesize cytoplasmic storage polysaccharides (e.g.glycogen, bacterial starch; Wilkinson 1963) and cell surfaceassociated polysaccharides (peptidoglycan [PG]; lipopolysaccharides [LPS]; lipooligosaccharides [LOS]; teichoic acids [TA]; lipoteichoic acids [LTA] and other cell wall polysaccharides [CW-PS], such as pellicles, exopolysaccharides [EPS] and capsular polysaccharides [CPS]) (reviewed in Chapot-Chartier 2014; Tytgat and Lebeer 2014;Schmid, Sieber and Rehm 2015;Mistou, Sutcliffe and van Sorge 2016).EPS and CPS are exocellular polysaccharides that at most differ in their degree of attachment to the cell surface: CPS are tightly linked, often covalently, to the cell surface and form a capsule around the cell, whereas EPS are secreted into the extracellular matrix or loosely associated with the cell surface via electrostatic interactions often forming a slime layer (Tytgat and Lebeer 2014).Pellicles are interpolated with the cell wall (PG) layer, as opposed to capsules that are typically the outermost layer of the cell envelope (Chapot-Chartier et al. 2010;Mistou, Sutcliffe and van Sorge 2016).Whilst most of these polysaccharides are produced by both Gram-negative and Gram-positive bacteria, the latter cannot synthesize LPS and LOS, but instead produce TA and LTA (Fig. 1).
Storage polysaccharides accumulate as carbon and energy reserves to cope with the starvation conditions temporarily present in the environment.Cell surface-associated polysaccharides play critical roles in interactions between bacteria and their surroundings.Accordingly, they serve a multitude of biological functions, such as maintenance of cell shape and structural integrity, charge and cation homeostasis, protection against adverse conditions such as desiccation, toxic compounds (bile salts, hydrolyzing enzymes, e.g.lysozyme, gastric and pancreatic enzymes, metal ions, antibiotics, ethanol, etc.) and antibacterial stresses (varying pH, osmolarity and gas atmosphere), predation by protozoans, evasion of the immune system and phage attack (Donot et al. 2012;Patel and Prajapat 2013;Ryan et al. 2015;Caggianiello, Kleerebezem and Spano 2016;Mahony et al. 2016).CPS and EPS have been postulated to play important roles in bacteria-host interactions, namely facilitating colonization through their ability to adhere to surfaces (e.g.adhesion to eukaryotic cells and mucosa), and in microbial-mediated immunomodulation (Mazmanian and Kasper 2006;Ryan et al. 2015;Caggianiello, Kleerebezem and Spano 2016).Furthermore, EPS play key roles in bacterial biofilms as summarized by Flemming and Wingender (2010).Hitherto, a comprehensive understanding of the biological functions of the exocellular polysaccharides, and especially EPS, has not been obtained.
Despite the vast structural diversity, bacteria produce polysaccharides by using either a sequential or an en bloc mechanism via four different pathways: the Wzy-dependent pathway (en bloc), the ATP-binding ABC transporter pathway (sequential), the synthase-dependent pathway (sequential) and The ABC transporter-dependent (CPS in Gram-negative bacteria; potentially other CW-PS) represented for Gram-negative bacteria.The polysaccharide chain, anchored on a poly-2-keto-3-deoxyoctulosonic acid linker in the cytoplasmic face of the inner membrane, is assembled by the action of GTs; the finished polysaccharide is exported via an efflux pump complex composed of ABC transporters spanning the inner membrane and periplasmatic proteins of the OPX family spanning the outer membrane.(c) The synthase-dependent pathway (CPS, EPS) performs the polymerization and transport by a single synthase complex, which secretes the complete polymer strands across the membranes and cell wall.(d) Extracellular synthesis of HoPS by the use of a single GT (sucrase) protein.Blue circle, glucose; yellow diamond, galactose; pink square, rhamnose; green triangle, fructose.CM, cytoplasmic membrane, OM, outer membrane; PG, peptidoglycan; GT, glycosyltransferase; Und-P, undecaprenylphosphate; P, phosphate; k, poly-2-keto-3-deoxyoctulosonic acid linker.extracellular synthesis by use of a single glycosyltransferase (sucrase) enzyme (Fig. 2; Tytgat and Lebeer 2014;Schmid, Sieber and Rehm 2015).In one bacterial species, two or more pathways can coexist for the production of different macromolecules.A more detailed description of selected pathways will be provided in the following sections of this review.
Lactic acid bacteria (LAB) are a heterogeneous group of Grampositive bacteria known for their key roles in the manufacture of fermented foods and beverages, such as cheese, yoghurt and other fermented milks; fermented meat; sourdough and other breads; vegetables and alcoholic beverages, such as wine, cider and beer; as well as their ubiquitous presence in animal microbiomes, a phenomenon that is frequently associated with beneficial effects (reviewed in Mozzi, Raya and Vignolo 2010;Lahtinen et al. 2011).As indicated by their designation, the main fermentation product of LAB from carbohydrate is lactate (Kandler 1983).This feature has been extensively exploited to extend the shelf life of milk and other raw materials, since the low pH generated by the conversion of the carbon source into lactic acid inhibits the growth of spoilage and pathogenic bacteria.In addi-tion to acidification, LAB can also contribute to the functionality and organoleptic properties of the fermented products by producing antimicrobial compounds (e.g.bacteriocins), bioactive peptides, vitamins, low calorie sugars, flavor and aroma compounds, and polysaccharides (Gaspar et al. 2013).In particular, the in situ production of polysaccharides by LAB has strongly been associated with the diverse technological, functional and health-promoting properties displayed by these microorganisms (recently reviewed in Ryan et al. 2015;Torino, Font de Valdez and Mozzi 2015;Caggianiello, Kleerebezem and Spano 2016;Mende, Rohm and Jaros 2016).Among the cell surface-associated polysaccharides known to be produced by LAB (PG, TA, LTA, pellicles, EPS and CPS; Fig. 1), it is the exocellular polysaccharides EPS and CPS that are generally identified as contributing to the functional attributes in food.In LAB literature, the term EPS is often used to refer to both exocellular polysaccharides, but in this review we will use the terms EPS and CPS differentially for the sake of clarity.These polysaccharides are known to improve the rheological properties of LAB-fermented products by influencing viscosity, syneresis, firmness and sensory properties.The location of the polysaccharide (EPS or CPS), the primary structural features (monosaccharide type and configuration, glycosidic linkage, non-sugar decorations, charge), the conformation and molecular weight, the amount of polysaccharide and the interactions of the polysaccharide with other system components are all factors that can contribute to and/or influence the displayed technofunctional properties.Whether other cell surface-associated polysaccharides are also involved in technofunctional properties such as rheology augmentation remains to be investigated.
Regarding monomeric composition, the exocellular polysaccharides (EPS and CPS) produced by LAB can also be classified as HoPS and HePS.HoPS are generally synthesized in the extracellular matrix by an extracellular glycosyltransferase (GT) (Fig. 2) and are composed of D-glucose or D-fructose, and thereby denominated glucans (α-or β-) and fructans.Glucansucrases and fructansucrases catalyze the polymerization of the glucans and fructans usually using sucrose as the donor of the corresponding monosaccharide and transferring this residue to the reducing end of the growing HoPS (van Hijum et al. 2006).Different genera of LAB, including Weissella, Leuconostoc, Lactobacillus, Pediococcus and Streptococcus (namely oral streptococci), produce HoPS via this pathway.Diversity and synthesis of sucrase-dependent HoPS in LAB and their applications in the food industry have been the focus of several recent reviews (Ryan et al. 2015;Torino, Font de Valdez and Mozzi 2015) and will not be the main focus of this review.
A few examples of LAB that produce HoPS constituted solely of galactose moieties have also been described (van Kranenburg et al. 1999c;Mozzi et al. 2006).These polygalactans are composed of several repeating units containing four or five galactose residues, and are most likely the product of the Wzy-dependent pathway, which in LAB is the pathway of choice for the synthesis of heteropolymeric EPS and CPS (Ryan et al. 2015;Torino, Font de Valdez and Mozzi 2015).A detailed overview of their properties in LAB is provided in the following sections.
In this review, we will describe and discuss the genetics and biochemistry of exocellular polysaccharide biosynthesis in LAB, with particular emphasis on polysaccharides synthesized via the Wzy pathway.Furthermore, industrial applications associated with the ability to produce these polysaccharides and strategies to screen and improve LAB for superior rheological properties will be reviewed.

COMPARATIVE GENOMICS OF POLYSACCHARIDE SYNTHESIS
Various gene clusters for the synthesis of exocellular polysaccharides via the Wzy-dependent pathway and genes encoding sucrases for the extracellular synthesis of HoPS have been found in the genomes of LAB.While sucrase-encoding genes have been described in the LAB genera Weissella, Leuconostoc, Lactobacillus, Pediococcus and Streptococcus, studies on Wzy-dependent eps gene clusters focus almost exclusively on three genera: Lactococcus, Lactobacillus and Streptococcus.Other polysaccharide biosynthetic gene clusters, such as those for the synthesis of pellicles in Lactococcus lactis and CW-PS in lactobacilli, have been reported, but the biosynthetic routes for these polysaccharides remain largely elusive (Ruas-Madiedo, Salazar and de los Reyes-Gavil án 2009;Chapot-Chartier 2014;Tytgat and Lebeer 2014;Mistou, Sutcliffe and van Sorge 2016).
Genes encoding Wzy-dependent exocellular polysaccharide biosynthesis proteins in LAB are typically organized in a cluster with an operon structure and are generally chromosomal in Streptococcus thermophilus, but can reside on a plasmid or the chromosome in L. lactis and Lactobacillus sp.(Ruas-Madiedo, Salazar and de los Reyes-Gavil án 2009).The organization of the eps gene cluster in LAB is similar to that of Gram-positive pathogens (Fig. 3).Generally, eps gene clusters are highly diverse and their nucleotide sequences are among the most variable sequences in LAB genomes.Mobile genetic elements play a role in this diversity.Indeed, insertion sequence (IS) elements flanking or within the operon are consistently present in the architecture of eps gene clusters (De Vuyst et al. 2001;Hols et al. 2005;Cui et al. 2016).However, the modular gene organization in eps gene clusters is conserved (De Vuyst et al. 2001;Jolly and Stingele 2001;Bentley et al. 2006;Ruas-Madiedo, Salazar and de los Reyes-Gavil án 2009;Goh et al. 2011;Okura et al. 2013;Skov Sørensen et al. 2016).Genes in the eps operon can be categorized into groups based on the putative or established functions of their products.These include modulatory genes (phosphoregulatory module epsBCD), polysaccharide assembly machinery genes (initiation epsE, polymerization wzy, export/flippase wzx and attachment epsA), genes encoding the GT necessary for the assembly of the repeating units and genes encoding non-housekeeping functions required for the synthesis of activated sugar precursors and modification of the sugar residues (Fig. 3), according to the working model for Streptococcus pneumoniae (Nourikyan et al. 2015).
Generally, the eps gene clusters are 15-20 Kb in size and comprise less than 30 genes.In LAB, the eps genes usually have the same orientation and are transcribed as a single mRNA (Guidolin et al. 1994;van Kranenburg et al. 1997;Lamothe et al. 2002).Genes located at the 5 end of the cluster are involved in the modulation and assembly machinery of polysaccharide biosynthesis and display the highest level of overall conservation.A typical eps gene cluster consists of five highly conserved genes epsA, epsB, epsC, epsD and epsE, and a variable region, which includes the polymerase wzy, the flippase wzx, one or more glucosyltransferases and/or other polymer-modifying genes (Fig. 3).Lactococcus lactis harbors an additional gene, epsX, in the conserved 5 end.Moreover, in this organism, genes epsL and orfY represent a second conserved region at the 3 end (Fig. 3).To date, no putative functions have been assigned to the lactococcal genes epsX and epsL.Gene epsL can be disrupted or overproduced in L. lactis NIZO B40 without any effect on EPS production (van Kranenburg et al. 1999c), suggesting that it is either non-essential or there is an alternative gene in the genome which product has a similar function.The 12-kb eps cluster of L. lactis NIZO B40 is flanked by a gene encoding an IS element and orfY, which is oriented in the opposite direction of the genes in the eps cluster.The cluster consists of 14 genes, including the 5 conserved genes, the more variable wzy and wzx, and non-conserved genes encoding GTs.Lactococcus lactis B891 harbors an eps cluster displaying an identical architecture, but the gene organization in the eps cluster of L. lactis NIZO B35 differs considerably (van Kranenburg et al. 1999a).In S. thermophilus, despite the various architectures described, the eps cluster is generally flanked at the 5 end by deoD (purine-nucleoside phosphorylase presumably involved in the biosynthesis and catabolism of nucleotides) and at the 3 end by orf14.9and/or bglB (6-phospho-beta-glucosidase).The product of orf14.9 has been putatively associated with cell growth (De Vuyst et al. 2011), and whether the flanking gene at the 3 end is orf14.9or bglB is still under debate.In the pathogen S. pneumoniae, genes dexB and aliA take this role, respectively.In summary, despite differences in the genes flanking the eps clusters in streptococci (both pathogenic and LAB), the conservation of eps gene clusters appears to be restricted to the genes encompassed within the cluster.This view possibly holds true for lactobacilli, but a thorough comparative analysis of the eps gene clusters for this diverse group is still lacking.Indeed, in Lactobacillus delbrueckii ssp.bulgaricus Lfi5, the eps cluster is flanked by genes orfX and orfY (Lamothe et al. 2002).The variation in the eps gene cluster architecture in lactobacilli is seemingly higher than in L. lactis and S. thermophilus, which likely reflects the large diversity within the Lactobacillus genus.
The nomenclature currently used for designating the genes in eps clusters encoding the Wzy-dependent pathway differs among organisms (Fig. 3).In L. lactis, the first conserved lac-tococcal eps genes are designated epsR, epsA, epsB, epsC and epsD (van Kranenburg et al. 1997(van Kranenburg et al. , 1999c;;Forde and Fitzgerald 2003;Dabour and LaPointe 2005;Knoshaug, Ahlgren and Trempy 2007), whereas in S. thermophilus, the corresponding genes are called epsA, epsB, epsC, epsD and epsE, respectively (Jolly and Stingele 2001;Goh et al. 2011).In L. delbrueckii ssp.bulgaricus (thereafter denominated Lb. bulgaricus), the genes are named epsABCDE (Lamothe et al. 2002), but the order of the conserved genes is not the same as in S. thermophilus.Often the eps genes are merely designated alphabetically by order of occurrence in Table 1.Repeating units of EPS/CPS synthesized by the Wzy pathway.The number of elucidated repeating units, uniqueness of the repeating unit, the size range in number of sugar residues of the repeating unit and occurrence of different sugar types and other constituents in the structures are listed.Data derived from de Vuyst and de Vin (2007), Ruas-Madiedo, Salazar and de los Reyes-Gavil án (2009) and original works published since 2009 (see Tables 2 and 3). a given locus.As a consequence, eps genes with the same name tag encode proteins with different functions: epsB encodes the cytoplasmic domain of a tyrosine-protein kinase in L. lactis, a phosphotyrosine protein phosphatase in S. thermophilus and the membrane-bound domain of the tyrosine-protein kinase in Lb. bulgaricus (Fig. 3).Even for the same species, the genes can be named differently: in L. lactis NIZO strain B40, epsI and epsK code for the Wzy polymerase and the Wzx flippase (polysaccharide export), while in L. lactis SMQ-461, the genes with corresponding functions are named epsH and epsM, respectively (van Kranenburg et al. 1997;Dabour and LaPointe 2005).
A standardized nomenclature for Wzy-dependent polysaccharide synthesis genes in pathogenic bacteria has been proposed previously (Reeves et al. 1996).In S. pneumoniae, capsule biosynthesis genes utilize the prefix cps followed by the serotype number and gene designation, which follows an alphabetical order based on the gene location in the cps operon, a nomenclature which is closely related to that of S. thermophilus.Herein, we adopt the nomenclature commonly applied in S. thermophilus, as it is arguably the best-characterized LAB in terms of exocellular polysaccharide production and the most widely used LAB in industrial applications for its texturizing properties.For a generic LAB eps gene cluster, we propose to designate the five first conserved genes epsABCDE, the polymerase wzy and the flippase wzx (Fig. 3).
The genes typically encoding GT, the polymerase (wzy) and the flippase (wzx) are situated in a variable part of the eps gene cluster, and often have a low degree of similarity to already characterized genes, which makes the prediction of their putative functions difficult.A substantial effort to computationally group and categorize the eps gene products has been carried out for the human pathogen S. pneumoniae (Bentley et al. 2006;Aanensen et al. 2007).Comparison of polysaccharide synthesis operons from 90 pneumococcal serotypes revealed that central genes responsible for the synthesis and polymerization of the repeat unit are highly variable and often non-homologous among serotypes (Bentley et al. 2006).In that study, 40 ho-mology groups for polysaccharide polymerases (wzy) and 4 for priming GTs (wchA, wciI, wcjG, wcjH) were found.The predictions for initial sugars, and subsequent repeating unit polymerization linkage, correlate well with the polymerase homology groups: 32 polymerase homology groups associated with WchA, 5 with WciI, 4 with WcjG and 1 with WcjH.These associations are mostly exclusive, with only five polymerase homology groups associated with two initial transferases, which indicates a high specificity of the initial transferases (Bentley et al. 2006).In addition, 13 groups of flippases and a great diversity of GTs were found in the polysaccharide gene clusters of S. pneumoniae.The presence of multiple non-homologous or highly divergent forms of GTs, together with often different G+C content of the region in which these are encoded, indicates that the genes have been acquired from different sources.With the growing number of genome sequences of LAB, bioinformatic analyses similar to those conducted in S. pneumoniae can now be applied to categorize the functions encoded by the genes in eps clusters and predict biosynthetic mechanisms in LAB.
these genes are present elsewhere in the chromosome (Boels et al. 2004).The same is true for galU, galE and glmSMU genes (Fig. 4).Genes encoding proteins involved in the decoration of the sugar residues, such as O-acetyl transferases and pyruvoyl transferases, have also been found in the eps gene clusters of LAB (Stingele et al. 1999;Wu et al. 2014;Fig. 3).In S. thermophilus, genes potentially involved in the production of sugar nucleotide precursors (e.g.encoding phosphoglycerate mutase and phosphatase) have been found upstream of orf14.9.
The eps gene clusters encoding the Wzy-dependent pathway are present in the less characterized Leuconostoc, Oenococcus and Pediococcus, as determined by searches in publicly available genomes at NCBI or proprietary LAB genomes (Chr.Hansen Culture Collection).
Leuconostoc is known for its production of HoPS like dextran, alternan and levan, but putative eps clusters for the production of HePS can be found in some Leuconostoc strains (Fig. 3).For instance, Leuconostoc gelidum ssp.gasicomitatum KG16-1 (GenBank Accession LN890331) contains a cluster of genes annotated as epsAHBCDEFG followed by six putative dTDP-rhamnosyl transferase genes localized between loci LEGK 0689 and LEGK 0710.Similarly, in Lc. gasicomitatum LMG 18811 (GenBank Accession FN822744), a cluster of genes annotated as epsABCDEFGHIJKX followed by one putative dTDP-rhamnosyl transferase gene is lo-calized between loci LEGAS 0698 and LEGAS 0712.A putative eps cluster found by BLAST analysis in Lc. mesenteroides ssp.mesenteroides strain BD3749 (GenBank Accession CP014610) had a similar eps gene cluster structure as in the two Leuconostoc strains described above.
Oenococcus oeni, closely related to Leuconostoc, is able to synthesize both homo-and hetero-polysaccharides, via distinct metabolic pathways (Dimopoulou et al. 2016).All 50 studied genomes contained at least one of the pathways: (i) a Wzydependent synthetic pathway, allowing the production of HePS made of glucose, galactose and rhamnose, mainly in a capsular form; (ii) a glucan synthase pathway (Gtf), involved in βglucan synthesis in a free and a cell-associated form, giving a ropy phenotype to growth media; and (iii) HoPS synthesis from sucrose (α-glucan or β-fructan) by glycoside-hydrolases of the GH70 and GH68 families (Dimopoulou et al. 2014).For instance, O. oeni PSU-1 (GenBank Accession CP000411) contains a putative HePS cluster (Wzy-dependent pathway) between loci OEOE 1507 and OEOE 1496 (Fig. 3).
An eps gene cluster can also be identified in the genome sequence of Pediococcus pentosaceus SL4 (GenBank Accession CP006854).It is situated between locus tags T256 03080 and T256 03130, and the order of the conserved epsBCD genes is similar to that in L. lactis B40.
In summary, the major genera of LAB used in food processing (Lactococcus, Streptococcus, Lactobacillus, Leuconostoc, Oenococcus and Pediococcus) possess eps gene clusters.Even though the potential to synthesize exocellular polysaccharides is encoded within the genomes of many of these bacteria, production of polysaccharides and their functional properties need to be evaluated for successful industrial applications.
The genomes of many LAB species contain a second cluster for the production of polysaccharides, which are CW-PS anchored to PG that frequently contain rhamnose in the repeating unit forming the polysaccharide (Hols et al. 2005;Thevenard et al. 2014;Mistou, Sutcliffe and van Sorge 2016).These CW-PS are arguably involved in antibiotic stress response in S. thermophilus (Thevenard et al. 2014), while in L. lactis, where CW-PS are known as the pellicle, a role as receptors for phages and a protective effect against host phagocytosis in murine macrophages have been established (Chapot-Chartier et al. 2010;Mahony et al. 2013).CW-PS are encoded by rmlD-associated gene clusters, which vary between 14 and 28 kB in size containing between 12 and 27 genes with putative functions such as GT, polysaccharide biosynthesis proteins, rhamnose biosynthesis proteins (RmlABCD) and transport molecules (Mahony et al. 2013;Mistou, Sutcliffe and van Sorge 2016).
In the following sections, insights into the biosynthesis and regulation of polysaccharide production via the Wzy-dependent pathway will be presented.

Biosynthetic pathways
Hitherto, only two pathways for the biosynthesis of the exocellular polysaccharides by LAB have been described in the literature: the Wzy-dependent pathway and the extracellular GT pathway for the synthesis of glucans and fructans (Ryan et al. 2015;Torino, Font de Valdez and Mozzi 2015).The latter is a relatively simple biochemical route that involves a specific GT (glucansucrase or fructansucrase), and an extracellular sugar donor, which is sucrose for the synthesis of glucans, but can also be other fructose-containing oligosaccharides (e.g.raffinose) for the synthesis of fructans (van Hijum et al. 2006;Galle and Arendt 2014).Extracellular biosynthesis of HoPS is essentially uncoupled from central metabolic processes and thereby not energetically demanding for the cell or prone to complex regulatory mechanisms.Indeed, cellular energy expenditure is restricted to the synthesis and export of the GT.In fact, the energy released by cleavage of the glycosidic bond in the substrate is used for the synthesis of new glycosidic bonds in the growing HoPS.The structural features of the HoPS (e.g.glycosidic bond, branching, size, molecular weight) are essentially determined by the type of GT encoded in the LAB genomes, and have been extensively reviewed by others (van Hijum et al. 2006;Patten and Laws 2015;Ryan et al. 2015;Torino, Font de Valdez and Mozzi 2015).Classification of the extracellular GTs has been briefly discussed in the previous section and will not be further presented in this review.
The biosynthesis of polysaccharides via the Wzy-dependent pathway is a complex intracellular process that was first studied for its involvement in the synthesis of the LPS O-antigen polysaccharide in Gram-negative bacteria (reviewed in Islam and Lam 2014), and later for production of CPS and EPS in both Gram-negative and Gram-positive bacteria (Whitfield 2006;Tytgat and Lebeer 2014;Schmid, Sieber and Rehm 2015).The full biosynthetic process can be divided into two distinct steps (Fig. 2): (i) generation of activated sugar precursors (sugar nucleotides) from central carbon metabolism in the cytoplasm (Fig. 4) and (ii) a committed cell membrane-associated assembly and polymerization of the polysaccharide (Fig. 5).
(i) Generation of activated sugar precursors: pathways for the generation of sugar nucleotides have been recurrently described in LAB (Welman and Maddox 2003;Neves et al. 2005;De Vuyst and De Vin 2007).The sugar nucleotides are synthesized in multistep pathways from glycolytic intermediates, generally glucose-6-phosphate (Glc-6-P) or fructose-6-phosphate (Fru-6-P) (Fig. 4), or in certain cases from intermediates of sugar specific catabolic pathways, such as the Leloir pathway intermediate α-glucose-1-phosphate (α-Glc-1-P) during the catabolism of galactose in L. lactis (Neves et al. 2006).The potential to produce different sugar nucleotides is intrinsically determined by the gene content of each LAB, which ultimately dictates the type of monomers present in the exocellular polysaccharides.Exocellular polysaccharides produced by LAB via the Wzy-dependent pathway consist of repeating units usually composed of two or more (usually 3-8) types of monosaccharides (Welman and Maddox 2003;Neves et al. 2005;De Vuyst and De Vin 2007; Ruas-Madiedo, Salazar and de los Reyes-Gavil án 2009).The most represented sugar moieties are galactose (Gal), glucose (Glc) and rhamnose (Rha), and to a lesser extent the Nacetylated sugar derivatives N-acetylglucosamine (GlcNAc) and N-acetylgalactosamine (GalNAc) (Table 1).Other monosaccharides and acetylated derivatives that have occasionally been found in the repeating units of LAB HePS are fucose (Fuc, 2 structures), ribose (Rib, 1 structure), glucuronic acid (GlcA, 1 structure), mannose (Man, 6 structures) and N-acetylmannosamine (ManNAc, 2 structures) (Robijn et al. 1996;Low et al. 1998;Faber et al. 2002;Li et al. 2014;Patten et al. 2014;Zhou et al. 2016).The pathways to generate the respective sugar nucleotides from the milk carbohydrate lactose and from glucose are shown in Fig. 4. The genes encoding these pathways can either be found in the polysaccharide biosynthetic clusters (see previous section) or at other locations in the genome.
Small quantities of uronic acids, fructose, arabinose and xylose have also been reported in the monosaccharide composition of polysaccharides determined by complete hydrolysis of purified HePS, but the presence of these monosaccharides has not been confirmed in the elucidated structures of the repeating units.Thus, these sugars are either contaminants from other CW-PS, cell lysis or from the growth medium as also indicated previously by others (De Vuyst and De Vin 2007; Ruas-Madiedo, Salazar and de los Reyes-Gavil án 2009).
(ii) Cell membrane-associated assembly and polymerization: the committed steps of the Wzy-dependent biosynthetic pathway have been mainly investigated in pathogenic streptococci, in particular for the synthesis of capsules in  Nourikyan et al. (2015).The genetic locus shows the genes involved in the synthesis and export of exocellular polysaccharidess in LAB.As an example, assembly and polymerization of the polysaccharide produced by L. lactis ssp.cremoris NIZO B40 is shown.In the eps gene clusters (generic and B40), the genes coding for the polysaccharide assembly machinery, glycosyltransferases, the EpsBCD phosphoregulatory system and synthesis of NDP-sugars are shown in green, orange, yellow and pink, respectively.The same color scheme is used for representing the proteins in the Wzy-dependent pathway.The reactions catalyzed by each GT gene product are indicated in the upper-left corner gray box.These reactions occur in the cytoplasm CM, cytoplasmic membrane.
S. pneumoniae (Bentley et al. 2006;Henriques et al. 2011;Yother 2011;Schaffner et al. 2014;Nourikyan et al. 2015;Grangeasse 2016), but are poorly characterized in LAB, with only a few functional studies described so far (Minic et al. 2007;Nierop Groot and Kleerebezem 2007;Cefalo, Broadbent and Welker 2011a;Dertli et al. 2013;Suzuki, Kobayashi and Kimoto-Nira 2013).Considering the importance and value of LAB exocellular polysaccharides, this is fairly surprising.Indeed, except for the extracellularly synthesized glucans and fructans, CPS and EPS are presumably the products of this pathway.Furthermore, CW-PS, such as the pellicle in L. lactis (Chapot-Chartier et al. 2010), the rhamnose-glucose polysaccharides in S. thermophilus (Hols et al. 2005) and the recently described lactobacilli CW-PS (Vinogradov et al. 2013(Vinogradov et al. , 2015(Vinogradov et al. , 2016) ) might also be synthesized via the Wzy-dependent pathway.However, the absence of several typical pathway components in some CW-PS biosynthetic clusters and the presence in others of genes with homology to ABC transporters (vide supra) raises the question to which pathway is used: the Wzy-dependent or the ABC transporter-dependent pathway (reviewed in Mistou, Sutcliffe and van Sorge 2016).Thus, for the synthesis of CW-PS, more work needs to be performed in order to firmly ascertain the biosynthetic routes.
Homologs of the key functions characterizing the Wzydependent pathway are ubiquitously present in the modular gene clusters for the synthesis of exocellular polysaccharides in LAB (Fig. 3).Some of the more recent and interesting findings concerning the functions of these genes arise from studies of capsule biosynthesis in the human pathogen S. pneumoniae (Bentley et al. 2006;Henriques et al. 2011;Yother 2011;Eberhardt et al. 2012;Schaffner et al. 2014;Nourikyan et al. 2015;Grangeasse 2016).Considering the relatively close genetic proximity between the pathogenic streptococci and LAB, the LAB Wzy protein functions will herein be surmised as similar to those reported for the pneumococcal proteins.The current model for polysaccharide synthesis by the Wzy-dependent pathway starts with the cytoplasmic en bloc synthesis of single-repeat units by sequential addition of the activated sugar precursors (Fig. 5).The first committed step consists of the activation of undecaprenylphosphate (Und-P), the lipid carrier, by transfer of the first residue from an activated sugar precursor via the priming GT.The second step comprises the assembly of the repeating unit by sequential addition of sugar nucleotides in reactions catalyzed by soluble and/or membrane-bound GTs.The repeating units are subsequently transported or flipped across the cytoplasmic membrane via a flippase (Wzx, CpsJ).Polymerization is catalyzed by the Wzy polymerase (CpsH), which adds single repeating units via generation of new glycosidic bonds to the reducing terminus of a growing polymer building up the exocellular polysaccharide structure.In Gram-negative bacteria, polysaccharide co-polymerase (PCP) proteins are postulated to be responsible for chain length control of the polymer (Islam and Lam 2014).These proteins are, however, not present in Gram-positive bacteria.The modulation proteins CpsC (Wzd or EpsC) and CpsD (Wze or EpsD), which constitute an active bacterial tyrosine (BY)-kinase, presumably act as surrogates (Yother 2011;Grangeasse, Nessler and Mijakovic 2012;Nourikyan et al. 2015;Grangeasse 2016).CpsC is a membrane-bound polypeptide that harbors two transmembrane spanning helices and a cytoplasmic C-terminal domain required for kinase activation.CpsD, a cytoplasmic protein harboring the kinase activity, possesses the Walker A and B ATP/GTP-binding motifs and a C-terminal tyrosine cluster motif.CpsC triggers CpsD kinase activity leading to autophosphorylation of the tyrosine cluster.Moreover, CpsC likely acts as a scaffold, keeping together the assembly machinery, i.e. the priming GT (e.g.CpsE), the Wzx flippase, the Wzy polymerase and CpsA.Although the mechanism by which the BY-kinases control exocellular polysaccharides synthesis remains elusive, the current model proposed that cycling between phosphorylated and non-phosphorylated forms of BY-kinases is required for proper synthesis and export of the polymer.Dephosphorylation is catalyzed by CpsB (Wzh or EpsB), which is a metal-dependent phosphotyrosine-protein phosphatase of the PHP family.After completing polymerization, linkage of the polysaccharide to PG might or might not occur, but this step remains largely elusive.Recently, attachment of the CPS to the cell wall in S. pneumoniae was shown to depend on the activity of members of the LytR-CpsA-Psr (LCP) proteins (Eberhardt et al. 2012), including Cps2A (EpsA, Wzg).LCP proteins also mediate the attachment to PG of CPS in Staphylococcus aureus (Chan et al. 2014) as well as the attachment of TAs in Bacillus subtilis (Kawai et al. 2011).Additional modifications to the polysaccharide, including addition of other noncarbohydrate constituents such as acetyl (O-acetylation), glycerol, pyruvoyl and phosphate groups, can also occur and the respective functions are presumably encoded in the dedicated Wzy-dependent pathway gene cluster.Indeed, in S. pneumoniae, O-acetylation of serotype 9V has been tentatively attributed to an O-acetyl transferase encoded by the wcjE gene (Bentley et al. 2006), while the epsH of S. thermophilus is a putative O-acetyltransferase (Stingele et al. 1999).In this section, the pneumococcal gene/protein nomenclature was used to describe the Wzy-dependent pathway proteins, but when possible we will use the adopted nomenclature for LAB (epsABCDE, wzy, wzx as in Fig. 3).
In summary, the dedicated Wzy pathway gene clusters comprise genes involved in the synthesis and export of exocellular polysaccharides that can be categorized into four groups (or modules) according to their functions (Figs 3-5): polysaccharide assembly machinery (the priming GT, Wzx flippase, Wzy polymerase, EpsA), the phosphoregulatory system that controls the polysaccharide assembly machinery (EpsB, EpsC, EpsD), the GT and sugar nucleotide biosynthetic pathways.Genes encoding enzymes involved in monosaccharide decoration can also be present in the cluster.
The priming GTs (also known as initiating GTs) are membrane-bound polyprenyl-P sugar-1-P transferases involved in the transfer of a phosphorylated monosaccharide from a sugar nucleotide to Und-P, thereby catalyzing the formation of an energy-rich phosphate bond.The membrane-bound Und-P-P-sugar is the substrate for the sequential action of GTs that build up the repeating unit in the cytoplasmic side of the membrane (Fig. 5).Priming GTs can either catalyze the addition of a hexose-1-phosphate (Glc-1-P or Gal-1-P) or of a hexosamine-1phosphate (GlcNAc-1-P or GalNAc-1-P) showing high degree of homology to the Salmonella enterica serovar typhimurium WbaP or to the enterobacterial WecA, respectively.These proteins belong to the bacterial sugar transferase family (Bac transf, PF02397).In S. pneumoniae, the WbaP homolog CpsE (WchA) comprises 455 amino acids and catalyzes the addition of Glc-1-P from UDPglucose to Und-P (Kolkman, van der Zeijst and Nuijten 1997;Cartee et al. 2005).Bentley et al. (2006) observed a perfect correlation between the presence/absence of cpsE and the presence/absence of glucose in the repeating unit of elucidated structures.These authors proposed that CpsE performs the same function in all 65 serotypes where it is present.Streptococcus thermophilus NCFB 2393 (aka LMG18311) EpsE, a homolog of the pneumococcal CpsE, was functionally characterized and shown to transfer Glc-1-P to Und-P (Almiron-Roig et al. 2000) (Fig. 6).Based on NCBI BlastP searches, proteins homologous to EpsE are present in other S. thermophilus strains, such as Sfi39 and EU20, for which the polysaccharide structures have been elucidated (Germond et al. 2001;Marshall et al. 2001).As for S. pneumoniae, a positive correlation between the presence of EpsE and the presence of glucose in the repeating unit is observed for the three S. thermophilus strains (Fig. 6).Whether this phenomenon is more general requires further characterization.
A different priming GT, the EpsE homolog in L. lactis NIZO B40 (EpsD), also catalyzes the transfer of Glc-1-P from UDP-Glc to UndP (van Kranenburg et al. 1999b).This phosphoglycosyltransferase is much smaller than the streptococcal EpsE, comprising only 228 amino acids.A homologous protein is present in NIZO B891, an L. lactis strain that produces a polysaccharide containing Glc and Gal (van Kranenburg et al. 1999c).Furthermore, homologs sharing over 90% identity and 100% coverage with EpsE are present in several lactococcal genomes deposited in NCBI (Fig. 6).This smaller glucosyl-1-P transferase shares high sequence homology with the C-terminus of the larger EpsE, which is characterized by the presence of the bac transf domain (PF02397) (Fig. 6).
The S. thermophilus Sfi6 EpsE protein was shown to transfer Gal-1-P from UDP-Gal to Und-P (Stingele et al. 1999).A homolog is found in strain MR-1C, for which prediction of galactosyl-1phosphotransferase activity can also be inferred from the elucidated structure of the repeating unit (Low et al. 1998).Furthermore, BlastP searches of S. thermophilus genomes deposited in NCBI reveal additional homologs, as for example in strain LY03, which contains galactose in the repeating unit of its polysaccharide (Fig. 6).Lactococcus lactis NIZO B35 (van Kranenburg et al. 1999c), a polygalactan-producing strain harbors a homolog of Sfi6 EpsE.The S. pneumoniae functional cognate is most likely the WcjG galactosyl-1-P transferase described by Bentley et al. (2006).It should be noted, however, that S. thermophilus Sfi6 EpsE and L. lactis B40 share considerable sequence homology (about 50% identity) despite the different sugar specificity, which in our opinion limits firm functional assignment of putative hexose-1-P transferases based solely on sequence homology.
Streptococcus pneumoniae possesses two additional phosphoglycosyltransferases, an alternative galactosyl-1-P transferase WcjH and a hexosamine-1-P transferase WciI, postulated to transfer GalNAc-1-P to the Und-P, as in the repeating unit of CPS of strain TIGR4 (Bentley et al. 2006).A potential homolog of the latter is present in the genome sequence of S. thermophilus strain TH982 (Treu et al. 2014); however, no information regarding exocellular polysaccharides synthesis is available for this strain.As for WcjH, no homologs were found in LAB by BlastP of NCBI deposited sequences.
A systematic investigation on the presence of the different priming GTs in LAB has not been performed so far.Most of the information available on functional characterization has been obtained from a few selected S. thermophilus or L. lactis strains in the 1990s (reviewed in Jolly and Stingele 2001;Broadbent et al. 2003).In lactobacilli and other LAB, the information is even scarcer.A recent study postulates that EpsE of Lb. johnsonii FI9785 adds Gal-1-P to Und-P, but firm confirmation is still required (Dertli et al. 2013), while Lebeer et al. (2009) showed that the EpsE protein of Lb. rhamnosus GG is a galactosyl-1-P transferase.Inactivation of epsE genes in LAB renders strains that no longer secrete polysaccharides to the extracellular matrix (van Kranenburg et al. 1997;Germond et al. 2001;Dabour and LaPointe 2005;Minic et al. 2007;Dertli et al. 2013;Rosini et al. 2015).These genetic studies provide strong evidence that the priming GT is essential for exocellular polysaccharide production.Considering the fast build-up of LAB genomics data, it is timely to gain further insights into the characteristics of genes encoding priming GTs.The different types of priming GTs identified in LAB are presented in Fig. 6.
GTs catalyze the formation of a glycosidic bond between a sugar moiety from an activated sugar precursor (donor) and a specific substrate molecule, which in the case of exocellular polysaccharide synthesis via the Wzy-dependent pathway, is the membrane-bound Und-P-P-sugar.GTs are a very diverse group of enzymes that can recognize a range of donor molecules and acceptors, and are often described as promiscuous.Currently, 101 GT families are described online in the Carbohydrate-Active enZymes (CAZy) database (Lombard et al. 2014), but promiscuity toward different substrates shown by some GTs makes it difficult to predict activity based merely on sequence analysis.The diversity of GT-encoding genes in LAB is enormous (see the previous section) and to the best of our knowledge functional characterization was only performed for the GTs encoded in the eps gene clusters of S. thermophilus Sfi6 and L. lactis NIZO B40 (Stingele et al. 1999;van Kranenburg et al. 1999b).The Sfi6 GT gene region comprises the epsEFGHI genes, with epsE encoding the galactosyl-1-P transferase (priming GT).EpsF encodes a galactosyltransferase that adds the branching α-1,6-galactose, EpsG is α-1,3-N-acetylgalactosaminyltransferase that carries out the second reaction in the build-up of the repeating unit, and by exclusion EpsI was postulated to be a β-1,3-glucosyltransferase. Thus, the order of biosynthesis of the S. thermophilus Sfi6 polysaccharide repeating unit is Gal, GalNAc, Glc and the side chain Gal.A similar functional analysis was performed for the GTs encompassed in the eps gene cluster of L. lactis NIZO B40: EpsD (priming GT), EpsE, EpsF and EpsG link Glc-1-P to Und-P (glucosyl-1P transferase), glucose to lipid-linked glucose (glucosyltransferase) and galactose to lipid-linked Glc-Glc (galactosyltransferase), respectively (van Kranenburg et al. 1999b).As for the priming GTs, more studies are required to grasp the diversity of LAB GTs.
Except for the priming GTs (EpsE), functional characterization of the Wzy-dependent pathway assembly machinery proteins, EpsA, Wzx flippase and Wzy polymerase has not been pursued.Recent studies in pathogenic Gram-positive bacteria showed that EpsA is a phosphotransferase belonging to the LCP family of proteins that catalyzes the attachment of CPS to N-acetylmuramic acid residues of PG by the formation of a phosphodiester bond (Eberhardt et al. 2012;Chan et al. 2014).The exact role of EpsA in LAB remains to be elucidated, as well as its function in LAB strains that presumably produce mainly EPS.
The presence of Wzy flippase encoding genes in the genomes of LAB is based primarily on sequence homology to genes encoding the few Wzy proteins that have been functionally characterized in other organisms (Islam and Lam 2014).The Wzx flippases are membrane-bound proteins often characterized by the presence of 12 transmembrane segments (TMS), which are classified as part of the polysaccharide transporter family.Wzx flippases recognize the UndP-P-repeating unit and flip it across the cytoplasmic membrane.Whether flippases display strict substrate specificity for a unique repeating unit is still a matter of debate, but the Und-P-P-linked sugar seems to play a key role serving as an initial point of discrimination by the flippase (Islam and Lam 2014).
The Wzy polymerases are membrane-bound proteins putatively harboring 10 to 14 TMS.The role of the Wzy polymerase is to add a single repeating unit via generation of a new glycosidic bond to the reducing terminus of a polysaccharide composed of multiple copies of the repeating unit.According to Islam and Lam (2014), Wzy polymerases should be viewed as GT enzymes, but their substrate specificity remains largely uncharacterized.The EpsBCD phosphoregulatory system involved in the control of synthesis and export of polysaccharide in S. pneumoniae will be discussed in more detail below.

Regulation of polysaccharide synthesis
Bacteria possess various regulatory mechanisms for controlling their metabolic activities, which enable them to interact with the surrounding environment and secure their niche among other forms of life.Understanding the mechanisms regulating exocellular polysaccharide production in pathogenic bacteria has long attracted the attention of scientists in the quest for identifying novel antibiotic targets.Likewise, comprehensive understanding of the regulatory mechanisms underlying polysaccharide biosynthesis in LAB allows industrial microbiologists to maximally exploit them in a myriad of commercial applications.
Exocellular polysaccharide production via the Wzydependent pathway is an energy-intensive process involving various regulatory enzymes, GTs as well as transport and polymerizing enzymes, in addition to housekeeping enzymes that are involved in general cellular processes and not exclusive to polysaccharide biosynthesis (vide supra).It is, therefore, expected that bacteria employ complex regulatory mechanisms at different cellular levels for tightly controlling the production of those polymers.It is also expected that competition for intracellular resources essential for growth, e.g.lipid carrier, sugar nucleotides and ATP, would result in maximum exocellular polysaccharide production occurring at the expense of the maximum specific growth rate or biomass yield of the organism.However, no clear patterns or systematic correlation with growth kinetics can be drawn based on the dynamics of polysaccharide production observed in different bacteria.For example, exocellular polysaccharide production in some strains of S. thermophilus and Lb.bulgaricus appears to reach maximum levels under conditions favoring optimal growth, whereas production of polysaccharide by mesophilic LAB was reported to have its highest levels under suboptimal growth conditions (Cerning 1990(Cerning , 1995;;De Vuyst et al. 2001;Broadbent et al. 2003;Welman and Maddox 2003;Li et al. 2016).Taken together, the regulatory mechanisms of the energy-demanding process of polysaccharide production may vary considerably and are likely to be dependent on the physiological significance of the polymer in a given organism.
Medium composition and growth conditions were shown to strongly influence the levels of exocellular polysaccharide production in LAB as well as its sugar composition and molecular mass, alluding to an obvious role of environmental factors in directly or indirectly modulating polysaccharide biosynthesis.Exocellular polysaccharide production levels were found to vary with the nature and/or concentration of the carbon source in the medium in S. thermophilus (Petit et al. 1991;De Vuyst et al. 1998;Degeest andDe Vuyst 1999, 2000;Li et al. 2016), Lb. bulgaricus (Grobben et al. 1995(Grobben et al. , 1996(Grobben et al. , 1998)), Lb. helveticus (Torino et al. 2001;Torino, Mozzi and Font de Valdez 2005), Lb. casei (Cerning et al. 1994;Mozzi et al. 2001), Lb. rhamnosus (Gamar, Blondeau and Simonet 1997) and L. lactis (Looijesteijn et al. 1999(Looijesteijn et al. , 2000;;Welman and Maddox 2003).In Lb. casei, the sugar distribution in HePS also varied depending on the carbon source (Cerning et al. 1992(Cerning et al. , 1994)), an observation which has been described in S. thermophilus when the lactose feeding rate in fed-batch cultures was reduced (Petit et al. 1991).The proportion of high molecular weight to low molecular weight polysaccharides in Lb. bulgaricus NCFB 2772 was shown to be dependent on the carbohydrate source in the growth medium (Grobben et al. 1997).In S. thermophilus LY03, this proportion varied with the carbon/nitrogen ratio in the medium (Degeest and De Vuyst 1999), which was shown to have a strong impact on exocellular polysaccharide production in other LAB strains (Cerning 1990;De Vuyst et al. 1998;De Vuyst and Degeest 1999;Harutoshi 2013).Other factors reported to modulate polysaccharide production include pH, temperature, oxygen tension, amino acids, phosphate and minerals (Mozzi et al. 1995;Grobben et al. 1998;De Vuyst and Degeest 1999;Looijesteijn et al. 2000;De Vuyst et al. 2001;Mozzi, Savoy de Giori and Font de Valdez 2003;Aslim et al. 2005;Harutoshi 2013) as well as vitamins (Grobben et al. 1998).In S. thermophilus, the overall effect of optimizing medium composition and culture conditions was a 4.2-fold increase in exocellular polysaccharide titers and a 9-fold increase in the molecular mass of the polymer (Li et al. 2016).
Despite the presence of numerous studies demonstrating the effect of environmental factors in modulating exocellular polysaccharide production in LAB, little is known about the underlying regulatory mechanisms.It is to be noted that, even within the same species, strains differ significantly in their response to changes in the above-mentioned factors.This emphasizes again the lack of generalized patterns that can help elucidating the regulatory mechanisms for exocellular polysaccharide production in LAB.The first four genes at the 5 end of the eps gene cluster in different LAB are highly conserved (Fig. 3) and their products have been traditionally proposed to play a modulatory role in exocellular polysaccharide production, based primarily on sequence similarity with corresponding proteins in other bacteria (van Kranenburg et al. 1999c;De Vuyst et al. 2001;Jolly and Stingele 2001;Broadbent et al. 2003;Péant et al. 2005;Wu et al. 2014).In most LAB, the product of the first gene in the cluster, EpsA, possesses an LCP domain, which is often present in multiple proteins in Gram-positive bacteria (H übscher et al. 2008).Transcriptional regulation of the eps operon in LAB has long been attributed to EpsA (Jolly and Stingele 2001;Broadbent et al. 2003).In B. subtilis, the homologous protein, LytR, was suggested to play a role in the transcriptional regulation of the ly-tRABC operon encoding cell wall-modifying enzymes (Lazarevic et al. 1992), a reason behind propagating the transcriptional regulatory function to EpsA in LAB by virtue of sequence homology.The EpsA homolog in L. lactis, EpsR, was also proposed to have a regulatory function in exocellular polysaccharide production based on sequence similarity with the regulatory proteins Xre, PrtR and RdgA, which all contain a DNA-binding domain (van Kranenburg et al. 1997).Moreover, there is another gene in L. lactis downstream of the eps gene cluster, orfY, the product of which is homologous to B. subtilis LytR (van Kranenburg et al. 1997).However, there exists limited biochemical or genetic evidence supporting the regulatory role of EpsA homologs in LAB or elucidating the nature of the proposed regulation.Only a recent study in Lb. johnsonii FI9785 showed that EpsA plays a crucial role in exocellular polysaccharide biosynthesis since deletion of the epsA homolog was shown to completely abolish exocellular polysaccharide production (Dertli et al. 2016).Unlike B. subtilis LytR, which acts as a transcriptional attenuator (Lazarevic et al. 1992), data from Lb. johnsonii FI9785 suggested that EpsA is a positive regulator of the eps operon (Dertli et al. 2016).The role of EpsA as a positive regulator of eps operon has been previously postulated in other Gram-positive bacteria, e.g. S. agalactiae (Cieslewicz et al. 2001) and S. pneumoniae (Morona et al. 2000;Broadbent et al. 2003).In these organisms, the epsA homolog (cpsA) was not found essential for polysaccharide repeating unit biosynthesis, though its deletion resulted in reduced capsule formation compared to the wild type.In S. agalactiae, cpsA deletion was associated with significant reductions in the transcription of eps genes (Cieslewicz et al. 2001).In addition, the purified CpsA from S. agalactiae (Hanson et al. 2012) and S. iniae (Hanson, Lowe and Neely 2011) was shown to bind specifically to DNA containing the putative promoter region of the cps operon of the respective strain in vitro using electrophoretic mobility shift assays.The specific DNA binding was attributed to the small intracellular domain of the protein, whereas the large extracellular domain, including the LCP domain, was not required (Hanson, Lowe and Neely 2011;Hanson et al. 2012).Besides the proposed regulatory function, data in S. agalactiae suggested the implication of CpsA in modulating cell wall, in cell division and in mediating CPS attachment to the cell surface (Hanson et al. 2012;Rowe et al. 2015;Toniolo et al. 2015).Indeed, the primary role recently proposed for members of the LCP protein family, including EpsA, in Gram-positive bacteria points towards an enzymatic function in the attachment of CW-PS to PG, which is supported by crystallographic (Kawai et al. 2011;Eberhardt et al. 2012) and genetic evidence (Dengler et al. 2012;Chan et al. 2013Chan et al. , 2014;;Wang et al. 2015;Baumgart et al. 2016).Furthermore, it was assumed that many of the reported changes in polysaccharide levels in LCP mutant strains of various species are indirect consequences of stress response induced by cell wall defects rather than a disrupted regulatory function of LCP proteins (Kawai et al. 2011;Dengler et al. 2012).As the LCP family comprises diverse subgroups of proteins, functional differences cannot be excluded (H übscher et al. 2008), which calls for a more cautious interpretation of protein homology searches and emphasizes the need for detailed functional studies in LAB.New insights on transcriptional-level regulation of eps gene expression may also be gained from the recent work in S. pneumoniae, which shows that sequence polymorphisms in the cps promoter play a role in fine-tuning the level of encapsulation (Wen et al. 2016).
As exocellular polysaccharide biosynthesis requires a constant supply of intracellular metabolites essential for cellular growth and maintenance, the modulation of polysaccharide levels by environmental factors may also be ascribed to transcriptional regulators outside the eps gene cluster which orchestrate growth and carbohydrate metabolism on a global level, such as CcpA or other sugar-specific regulators in LAB (Ravcheev et al. 2013).This was postulated in several LAB strains, where changes in exocellular polysaccharide levels could be correlated with the availability of sugar nucleotides, indicating the potential impact of any regulatory mechanism controlling the metabolic fluxes to those precursors on polysaccharide production.This might be evident in S. thermophilus, where a correlation between exocellular polysaccharide production and enzymes involved in sugar nucleotide biosynthesis (Fig. 4) could be clearly observed (Escalante et al. 1998;Degeest and de Vuyst 2000;Levander, Svensson and Radstrom 2002;Svensson et al. 2005).In S. thermophilus LY03, the activity levels of α-phosphoglucomutase (PgmA) as well as the Leloir pathway enzymes UDP-galactose 4-epimerase (GalE) and glucose-1-phosphate uridylyltransferase (GalU) were highly correlated with the amount of polysaccharide produced, whereas no correlation with the activity of the key glycolytic enzyme fructosebisphosphatase (Fbp) was observed (Degeest and de Vuyst 2000).A galactose-fermenting mutant of the same strain with increased activities of the Leloir pathway enzymes also showed a higher polysaccharide yield as compared to the parent strain (Levander, Svensson and Radstrom 2002).Overexpression of galU in combination with pgmA in S. thermophilus LY03 or galU alone in its galactose-fermenting mutant resulted in the production of significantly higher levels of polysaccharide, which correlated with higher levels of GalU and PgmA activities (Levander, Svensson and Radstrom 2002).The combined effect of pgmA and galU overexpression together with the enhanced Leloir enzyme activities in the galactose-fermenting strain was found to further boost exocellular polysaccharide yields (Svensson et al. 2005).In that case, UDP-glucose levels did not differ from those in the parent strain suggesting that this sugar nucleotide is channeled in favor of exocellular polysaccharide biosynthesis in the engineered strain rather than accumulated (Svensson et al. 2005).The combined data illustrates an important role of Leloir pathway enzymes in S. thermophilus, which generally cannot ferment galactose, in the generation of sugar nucleotides for exocellular polysaccharide biosynthesis and raises the possibility of potential control mechanisms outside the eps gene cluster.In Lb. bulgaricus and Lb.casei, correlations could also be established between the levels of various enzymes involved in sugar nucleotide biosynthesis and the amount of polysaccharide produced (Grobben et al. 1996;Mozzi, Savoy de Giori and Font de Valdez 2003).
On the other hand, studies in L. lactis did not always conform to the same pattern as above.In L. lactis ssp.cremoris NIZO B40, changes in the levels of exocellular polysaccharide production with the sugar source were indeed attributed to differences in the capacity to produce sugar nucleotides, whilst the transcriptional levels of eps genes remained constant (Looijesteijn et al. 1999).However, the activities of different enzymes required for sugar nucleotide biosynthesis did not change with the levels of sugar nucleotides.In fructose-grown cells, Fbp activity was the limiting factor for exocellular polysaccharide production, as the enzyme is needed for the biosynthesis of sugar nucleotides when fructose is used as the sugar source (Looijesteijn et al. 1999).In a later study, overexpression of galU, and the subsequent increase in UDP-glucose and UDP-glucose levels, did not result in any apparent increase in polysaccharide production (Boels et al. 2001).It was suggested that the control of exocellular polysaccharides biosynthesis in L. lactis by the enzymes involved in sugar nucleotide generation might be dependent on the sugar composition of the polymer (Boels et al. 2001).
In addition to the regulation of the overall amounts of exocellular polysaccharide produced, there exist apparent mechanisms for regulating its polymerization and chain length.In Gram-negative pathogens, production of preferred chain lengths of LPS O-antigen, mostly synthesized through the Wzydependent pathway, was deemed crucial for their survival in different environments (Kintz and Goldberg 2011;Osawa et al. 2013;Chang et al. 2015) and is believed to be controlled by the PCP Wzz (Morona et al. 2009;Islam and Lam 2014).In Gram-positive bacteria, this regulation appears to be achieved through the concerted actions of EpsB, EpsC and EpsD homologs, which constitute a phosphorylation-dependent regulatory system (Yother 2011;Grangeasse 2016).In that system, EpsC, a transmembrane activation protein, is required for the autophosphorylation of the tyrosine cluster of the cytoplasmic protein EpsD (vide supra), i.e. both proteins constitute a functional autophosphorylating BY kinase, the activity of which is modulated via interaction with the phosphotyrosine phosphatase EpsB (Morona et al. 2000(Morona et al. , 2002;;Bender, Cartee and Yother 2003;Soulat et al. 2006;Nourikyan et al. 2015;Toniolo et al. 2015).Although most of the knowledge on these proteins comes from the work on Gram-positive pathogens, similar roles were proposed for EpsB-D in LAB based on sequence homology and a small number of functional studies.In S. thermophilus, the requirement of EpsC for the phosphorylation of EpsD was verified and both proteins were found essential for exocellular polysaccharide synthesis, whereas EpsB was not (Minic et al. 2007).Protein-protein interactions between EpsC and EpsD as well as EpsB and EpsD homologs were demonstrated in S. thermophilus and L. lactis using the yeast two-hybrid system (Cefalo, Broadbent and Welker 2011a,b).The phosphotyrosine phosphatase function of purified EpsB has also been verified in S. thermophilus (Cefalo, Broadbent and Welker 2013).
Since BY kinases are known to be involved in different regulatory processes (Bechet et al. 2009;Kobir et al. 2011;Grangeasse, Nessler and Mijakovic 2012), the alternated phosphorylation and dephosphorylation of EpsD represents a potential regulatory system for exocellular polysaccharide assembly.In addition, in S. pneumonaie, CpsD is proposed to play a dual function in capsule assembly and cell division, where its phosphorylation acts as a signaling system coordinating CPS synthesis with chromosome segregation (Nourikyan et al. 2015).
Mutations in the components of the phosphoregulatory system are known to affect exocellular polysaccharide levels and/or molecular weight (Yother 2011).However, the detailed mechanisms and the modulatory environmental stimuli are still not fully understood and do not appear to follow the same pattern across different exocellular polysaccharide-producing bacteria.For instance, the phosphorylated form of EpsD homolog in S. pneumoniae was found to negatively regulate CPS production in one strain (Morona et al. 2000) while stimulating its production in another (Bender, Cartee and Yother 2003).In S. agalactiae, inactivation of EpsC or EpsD homologs resulted in a marked decrease in capsule formation (Cieslewicz et al. 2001), whereas epsC deletion in Lb. johnsonii apparently enhanced polysaccharide production (Dertli et al. 2013;Horn et al. 2013).Deletion of epsC or epsD homologs in S. pneumoniae resulted in the production of only short-chain polymers (Bender, Cartee and Yother 2003), alluding to their involvement in chain-length determination.Similarly, CpsC extracellular domain in S. agalactiae appeared necessary for the production of high molecular weight polysaccharides (Toniolo et al. 2015).In S. thermophilus, both EpsC and EpsD were required for the phosphoglycosyltransferase (priming GT) function of EpsE through the phosphorylation of the Tyr200 residue of the latter (Minic et al. 2007).A positive correlation between epsC expression levels and exocellular polysaccharide molecular weight in S. thermophilus could also be established (Li et al. 2016).
New perspectives into the control of exocellular polysaccharide production through the EpsBCD phosphoregulatory system can be gained from a recent study in B. subtilis.It was suggested that the control of EpsD autophosphorylation is mediated by the polysaccharides produced by the organism itself, since polysaccharide inhibits the autophophorylation through its interaction with the extracellular domain of EpsC homolog (Elsholz, Wacker and Losick 2014).The unphosphorylated EpsD was suggested as the active form of the protein that can activate different GTs through phosphorylation, making the exocellular polysaccharide a signaling molecule controlling its own production through a positive feedback loop (Elsholz, Wacker and Losick 2014).
In summary, despite the considerable advances in the understanding of the functions involved in polysaccharide biosynthesis via the Wzy-dependent pathway, a complete picture of the underlying mechanisms in LAB is still missing, which raises the need for more functional studies on the proteins involved in exocellular polysaccharide biosynthesis in this important group of bacteria.

Amount and molecular weight of polysaccharides produced by LAB
As described in the previous section, both environmental and intrinsic factors affect the yield and molecular weight of polysaccharides produced by LAB.Amount and molecular weight are properties that have often been correlated with the technofunctional properties of the polysaccharides in diverse applications (Mende, Rohm and Jaros 2016).These macromolecular properties have been extensively reviewed by others, but a concise overview is presented henceforth.
Polysaccharides synthesized via the Wzy-dependent pathway are usually produced in amounts from 20 to 600 mg L −1 (reviewed in Vaningelgem et al. 2004; Ruas-Madiedo, Salazar and de los Reyes-Gavil án 2009; Leroy and De Vuyst 2016), with only a few strains reported to produce higher amounts, e.g. 1 g L −1 in S. thermophilus ASCC 1275 (Wu et al. 2014), or over 2.7 g L −1 in Lb. rhamnosus RW-9595M (Bergmaier, Champagne and Lacroix 2003).The amount of sucrase-dependent HoPS produced by LAB is usually higher than 1 g L −1 (Ruas-Madiedo, Salazar and de los Reyes-Gavil án 2009; Torino, Font de Valdez and Mozzi 2015), with a few strains reaching values of about 10 g L −1 as in the case of Lb. reuteri Lb121, which produces both α-glucan and β-fructan (Van Geel-Schutten et al. 1999).The higher titers observed for the extracellularly synthesized HoPS reflect the lack of interdependency with central metabolic processes.On the other hand, polysaccharide synthesis via the Wzy-dependent pathway is strongly intertwined with carbon and energy metabolism.Care should be taken when using for comparative purposes the values reported in the literature for polysaccharide titers.Indeed, differences in isolation procedures can impact the exocellular polysaccharide concentration measured, thus leading to either underestimation or overestimation of the 'true' amounts (Ruas-Madiedo and de los Reyes-Gavilan 2005, Mende, Rohm and Jaros 2016).
LAB produce both polymers of high and low molecular weight.Polysaccharides synthesized via the Wzy-dependent pathway can range in molecular weight from 8 to over 5000 kDa (Mozzi et al. 2006).A simultaneous occurrence of polysaccharides of different sizes is frequent, i.e.LAB can produce simultaneously the same exocellular polysaccharide with different molecular weights (Mozzi et al. 2006; Ruas-Madiedo, Salazar and de los Reyes-Gavil án 2009; Mende et al. 2013).It has been speculated that the low molecular weight polymer correlates with CPS, whereas the high molecular weight is found free in Table 2. Monosaccharide composition and non-monosaccharide constituents present in structures of repeating units forming EPS/CPS synthesized by the Wzy pathway in LAB that have been reported after the reviews by de Vuyst and de Vin (2007) the extracellular matrix (Mende, Rohm and Jaros 2016).Despite the variation in molecular weight, most commonly LAB produce one type of exocellular polysaccharides.Some studies report the ability to produce different types of polysaccharides simultaneously, for instance different types of HoPS in Lb. reuteri 121 (Van Geel-Schutten et al. 1999), or a mixture of HoPS and HePS in e.g.Lb. johnsonii (Dertli et al. 2013) or O. oeni (Dimopoulou et al. 2016).
The molecular weight of glucan and fructan HoPS is usually between 10 5 and 10 6 Da, but polymers with molecular weight up to more than 10 7 Da have been reported (Ruas-Madiedo, Salazar and de los Reyes-Gavil án 2009; Torino, Font de Valdez and Mozzi 2015).

Polysaccharide structures
The structural diversity of exocellular polysaccharides produced by LAB is enormous.These polymers are comprised of a single type of sugar (HoPS) or two or more monosaccharides (HePS), can either be unbranched or branched, and neutral or charged.This range of structural features is based on the variation of sugar monomers, a wide range of glycosidic linkages, the presence of branches and decoration with non-carbohydrate constituents.The monosaccharides present in LAB polysaccharides can vary in nature, anomeric configuration (α-or β-anomer), stereochemistry (L-or D-chiral forms) and cyclic conformation (pyranose or furanose).This assortment of monosaccharides and the ensemble of different ways they can be linked together generate an impressive diversity.As an example, just two glucose residues can be joined in 30 different ways (Laine 1994).
The structural features of polysaccharides produced by LAB have been thoroughly reviewed by de Vuyst and de Vin (2007) and Ruas-Madiedo, Salazar and de los Reyes-Gavil án (2009).More recently, Torino, Font de Valdez and Mozzi (2015) revisited the topic, but focused mainly on extracellularly synthesized HoPS.An updated overview of repeating unit structures synthesized via the Wzy-dependent pathway is presented in Tables 1-3.Out of 81 structures elucidated so far, 55 are unique (Table 1).Regarding structure availability, the LAB genera are not equally represented: Lactobacillus 47 structures, Streptococcus 28 structures, Lactococcus 6 structures and none for other LAB, even though the Wzy-dependent pathway gene clusters have been found in Leuconostoc and Oenococcus (Fig. 3).The greatest variety is found in lactobacilli (Table 1), a trait that is likely due to the large diversity within the genus.Exocellular polysaccharides synthesized by LAB are mostly branched; so far only seven non-branched structures have been reported, with all except one being produced by different species of Lactobacillus (Table 3; De Vuyst and De Vin 2007; Ruas-Madiedo, Salazar and de los Reyes-Gavil án 2009).
Streptococcus thermophilus 8S also produces a linear polysaccharide.Neutral polysaccharides are the most common, but decoration with phosphate, as in a few L. lactis strains and one Lb.rhamnosus, renders the polymer anionic (Table 1).Other noncarbohydrate decorations like acetylation and pyruvoylation are postulated to be of importance for the biological role and the functionality of the polymers (Tables 1 and 2; Torino, Font de Valdez and Mozzi 2015; Wang et al. 2015).Besides phosphate, acetyl and pyruvoyl groups, glycerol-phosphate groups are also found as decorations on certain polymers produced by lactobacilli (Tables 1 and 2).
The sugars galactose and glucose, and to a lesser extent rhamnose, are the most frequently occurring in repeating units of exocellular polysaccharides.Galactose and glucose occur both in the pyranose and furanose forms, while rhamnose is generally in the pyranose form.The latter is quite frequent in L. lactis and S. thermophilus, but less in lactobacilli.Indeed, to date it has only been found in the species Lb. rhamnosus and Lb.bulgaricus.In contrast, mannose, which is not present in any repeating unit of polysaccharides produced by L. lactis and S. thermophilus, is relatively common in lactobacilli (Tables 2 and 3).GalNAc and Glc-NAc are present in polysaccharides produced by a few strains in the three genera.Other sugars that have occasionally been found in repeating units are as follows: ManNAc, fucose, ribose and a non-ionic sugar.
Table 3. Structures of the repeating units of exocellular polysaccharides synthesized via the Wzy-dependent pathway published since 2009, and thereby not presented in previous reviews by de Vuyst and de Vin (2007) and Ruas-Madiedo, Salazar and de los Reyes-Gavil án (2009).The structure of all repeating units has been elucidated using nuclear magnetic resonance spectroscopy.Determination of the monomeric composition of exocellular polysaccharide produced by LAB has been attempted in several studies (Ruas-Madiedo et al. 2010;Qin et al. 2011;Suzuki, Kobayashi and Kimoto-Nira 2013;Mende et al. 2014;Wang et al. 2014a,b), but these data were not considered for the analysis presented in Table 2 as no structure for the repeating unit of the polysaccharides was reported.

APPLICATIONS IN FOOD
Exocellular polysaccharide-producing LAB cultures are on high demand for applications in food, mainly in fermented dairy products, such as yoghurt, cheese and dairy-based desserts, and also in fermented bakery products, due to their positive contribution to textural properties.The in situ production of exocellular polysaccharides arguably has the potential to decrease the amount of added ingredients and stabilizers, such as dairy proteins, starch, pectin and hydrocolloids, often used to improve the textural properties of fermented dairy and cereal products.Ultimately, replacement of these additives leads to a clean label and a reduced production cost.In addition, exocellular polysaccharides in fermented dairy and cereal products can be considered as a functional ingredient due to the postulated health benefits (Ryan et al. 2015;Caggianiello, Kleerebezem and Spano 2016).Many studies are available on the effects that polysaccharide-producing strains or cultures have on the textural properties and stability of fermented dairy products, and several hypotheses and conclusions have been formulated (Table 4).Yet, there is not a clear understanding of the role of exocellular polysaccharides in the microstructure of fermented dairy foods.The use of different exocellular polysaccharideproducing strains, and the different experimental conditions used, such as fermentation temperature, composition of the milk base, including amount and type of milk proteins, leads to a discrepancy in the obtained conclusions (Table 4).The different polysaccharide characteristics, such as attachment to the cell (CPS vs EPS), molecular composition, molecular weight, conformation and size, rigidity, degree of branching and charge, will influence the textural properties (Table 4).This is mainly due to the different interactions occurring in the system, which influence the texture properties of the matrix.

Exopolysaccharides and capsular polysaccharides
Mesophilic and thermophilic LAB, including the common species used in yoghurt, Streptococcus thermophilus and Lb.bulgaricus, are able to produce EPS and CPS.In fermented milk, EPS and CPS can contribute to enhanced viscosity, creaminess, shear resistance, water-holding capacity, and to low syneresis, with some of the EPS having higher contribution to ropiness than most of CPS (Hassan et al. 1996a(Hassan et al. ,b, 2003;;Folkenberg et al. 2005Folkenberg et al. , 2006a,b;,b;Yang et al. 2010;Kristo, Miao and Corredig 2011;Costa et al. 2012a;Buldo et al. 2016).Due to the limited knowledge on the structure of exocellular polysaccharides produced by different LAB, the effect that they have on textural properties is often generalized.Among the studied strains, which positively contribute to texture, the following association is proposed: generally polysaccharide-producing strains of L. lactis ssp.cremoris are known to contribute to high viscoelastic properties, including viscosity and gel stiffness, and hence moduli, ropiness and sliminess (Ruas-Madiedo, Hugenholtz and Zoon 2002;Girard and Schaffer-Lequart 2007a,b;Ayala-Hern ández et al. 2009;Kristo, Miao and Corredig 2011;Costa et al. 2012a).Some EPS-producing strains from Lb. rhamnosus, Lb. casei and Lb.helveticus result in high viscosity products (Doleyres, Schaub and Lacroix 2005;Yang et al. 2010).Some of the EPS-producing Lb. bulgaricus strains seem to negatively contribute to the building of the structure, and thus to lower moduli, probably due to a depletion effect with casein (Girard andSchaffer-Lequart 2007a, Hess, Roberts andZiegler 1997;Purohit et al. 2009).Generally, EPS-producing S. thermophilus strains appear to positively contribute to the viscosity of the product, but negatively to the gel stiffness, and thus to the moduli (Hassan et al. 2003;Amatayakul et al. 2006;Purohit et al. 2009;Gentès, St-Gelais and Turgeon 2011;Mende et al. 2012).The cause of higher viscosity has been attributed to different factors, including higher EPS molecular weight, linearity and stiffness of the backbone, and complexity of the side chains (Faber et al. 1998;Tuinier et al. 2001;De Vuyst et al. 2003;Gentès, St-Gelais andTurgeon 2011, 2013).Polysaccharides which occupy larger volume in solution tend to produce high viscosity of the system.In addition, the complex interactions with other elements in the system, such as proteins and/or cells, strongly contributed to the viscosity of the system (Folkenberg et al. 2005;Girard and Schaffer-Lequart 2007a.b;Corredig, Sharafbafi and Kristo 2011;Gentès, St-Gelais and Turgeon 2011;Buldo et al. 2016).The number of side chains (branching) and the radius of gyration (R G ) of polysaccharides have been positively associated to gel firmness (Tuinier et al. 2001;Ruas-Madiedo, Hugenholtz and Zoon 2002;Petry et al. 2003).Reduced syneresis is a common property conferred by different EPS-producing strains.A positive correlation with EPS neutrally charged and with a stiff backbone was observed, whereas anionic EPS seem to negatively contribute to syneresis (Gent ès, St-Gelais andTurgeon 2011, 2013).
CPS has better water-binding properties than EPS (Low et al. 1998;Broadbent et al. 2001).In mozzarella cheese, CPS produced by S. thermophilus MR-1C can improve water retention and meltability (Low et al. 1998;Broadbent et al. 2001).In half-fat cheddar cheese, EPS-and/or CPS-producing strains lead to higher water retention, cheese yield and viscosity of the whey compared to EPS-negative control strains (Dabour and LaPointe 2005;Dabour et al. 2006).CPS are less prone to increase the viscosity of the whey compared to ropy EPS, hence resulting in a more favorable application in cheesemaking (Broadbent et al. 2001).Generally, application of EPS-producing strains to cheesemaking has been less extensively studied compared to fermented milk.
The charge of EPS, together with the EPS production time during fermentation, determines the distribution of the EPS in the microstructure, and hence their interference with gelation and with the protein matrix.The majority of the exocellular polysaccharides produced by LAB is uncharged, but a few charged polymers have also been reported (Table 1).Negatively charged EPS has been associated with higher viscoelastic properties, stiffness and viscosity (Gent ès, St-Gelais and Turgeon 2011;Kristo, Miao and Corredig 2011).During acidification, milk proteins go from negatively charged at the initial pH of the milk, approximately 6.5, to positively charged at the end of the fermentation, around pH 4.6-4.3.At their isoelectric point, the proteins are neutrally charged.If the EPS can interact with the protein network, e.g.via electrostatic interactions, therefore interfering with protein coagulation, a continuous branched protein network will be formed, which will lead to higher viscoelastic properties (Girard and Schaffer-Lequart 2007a;Ayala-Hern ández et al. 2009;Corredig, Sharafbafi and Kristo 2011;Gentès, St-Gelais and Turgeon 2013;Hahn et al. 2014;Buldo et al. 2016).In case of surface inert properties between EPS and proteins (e.g.casein), such as neutral or positively low charged EPS, depletion effect and/or phase separation can occur, leading to formation of both protein and EPS aggregates in the serum phase (Tuinier et al. 1999;Tuinier, Dhont and De Kruif 2000;de Kruif and Tuinier 2001;Hassan, Frank and Qvist 2002;Corredig, Sharafbafi and Kristo 2011;Gentès, St-Gelais and Turgeon 2011;Costa et al. 2012a;Buldo et al. 2016).This phenomenon has been linked either to a decrease in viscoelastic properties, due to the interference of the EPS with the protein network formation or to an increase in viscoelastic properties, due to the formation of a dense casein network (Hassan et al. 2003;Amatayakul et al. 2006;Gentès, St-Gelais and Turgeon 2011;Buldo et al. 2016).The presence of EPS in the serum phase also leads to higher water-binding capacity, which results in lower syneresis and high viscosity (Costa et al. 2012a;Gentès, St-Gelais and Turgeon 2013).However, one has to keep in mind that the explanations are mainly based on hypotheses, since the mechanism of interaction between exocellular polysaccharides and other elements in the system remains largely unknown, mainly due to limitations in identification and/or visualization of the polysaccharides.
The most significant contribution to textural properties is related to the properties of the exocellular polysaccharides, rather than their amount.Several studies report a lack of positive correlation between the amount of exocellular polysaccharides and textural properties (Marshall and Rawson 1999;Petry et al. 2003;Doleyres, Schaub and Lacroix 2005;Gentès, St-Gelais and Turgeon 2011;Hahn et al. 2014;Buldo et al. 2016).This could be due to the higher degree of interference of the polysaccharide with the  protein network (Hahn et al. 2014).However, whether the concentration of polysaccharide affects texture remains controversial (Mende, Rohm and Jaros 2016).
Isolation and characterization of exocellular polysaccharides from fermented milk is still a challenging task.Several protocols are available in the literature for exocellular polysaccharide isolation.Mende, Rohm and Jaros (2016) illustrate the principal steps with the related problems for exocellular polysaccharide isolation.The main concerns for the isolation and purification steps are related to the purity and/or loss of exocellular polysaccharides during extraction and also to degradation of the structure during the extraction process.The difference in purity leads to a broad and inaccurate estimation of the exocellular polysaccharide titer, as well as an inaccurate characterization of the structure.The exocellular polysaccharide titer varies from 20 to 600 mg L -1 , based on strain type, milk composition and fermentation conditions (Cerning 1995;Amatayakul et al. 2006;Mende, Rohm and Jaros 2016).Polysaccharide production seems to be closely related to the amount and type of proteins present in the media (Amatayakul et al. 2006;Buldo et al. 2016), which in some strains was simply attributed to the direct effect of the protein source on bacterial growth (Zisu and Shah 2003).
The effect of isolated EPS from LAB, added to milk prior to fermentation, on the textural properties of several fermented milk products has also been studied.Generally, no beneficial effects on textural properties or on cost-in-use have been reported (Vlahopoulou, Bell and Wilbey 2001;Doleyres, Schaub and Lacroix 2005; Laneuville and Turgeon 2014).However, Mende et al. (2013) have shown that the stiffness of the gel increased with the concentration of S. thermophilus ST-143 EPS added to milk prior to chemical acidification.Based on this observation, the authors proposed that the molecular weight of the polysaccharide rather than the charge contributes to the stiffness.The positive contribution of exocellular polysaccharides to textural properties of fermented dairy products seems to be driven by high molecular weight and complexity of the side chain.The presence of both non-ropy EPS and CPS improves the textural properties without contributing to ropiness.
In addition to dairy products, exocellular polysaccharides produced by LAB are known to improve the textural properties and the staling of bread, as well as the technological properties of the dough, such as water absorption, workability and rising of the dough (Tieking and G änzle 2005).In bakery applications, the most studied polysaccharide-producing LAB are Lb.reuteri, Lb. pontis, Lb. panis, Lb. acidophilus, Lb. sanfranciscencis and Lb. frumenti (Tieking et al. 2003).Dextran, reuteran and levan improve the textural properties of bread, with in situ production being more effective than added EPS (Tieking and G änzle 2005).
The rational design of a LAB culture that possesses superior rheological properties will thus require (a priori) knowledge regarding exocellular polysaccharide production, structural and macromolecular features of such polysaccharides, and structure-function correlations in different environmental conditions.

DEVELOPING STRAINS FOR FOOD AND OTHER APPLICATIONS
Use of exocellular polysaccharide-producing strains is relevant for fermented milk and cheese applications (see the 'Applications in food' section), other food applications (e.g.bakery) as well as applications outside the food sector (e.g.pharma and medical applications).Despite the presence of genes encoding exocellular polysaccharide synthesis in many LAB, only few strains give the desired textural properties in relevant food matrices, e.g.fermented milk.Screening tools for exocellular polysaccharide production or improved texture properties of isolates are therefore of interest to various industries including starter culture producers.Alternatively, methods to improve strains with regard to their ability to either produce more exocellular polysaccharides or modified exocellular polysaccharides with better viscosifying features, or methods that allow development of strains that give enhanced texture to fermented milk, are of great interest to the industry.

Screening for exocellular polysaccharide-producing strains
Screening methods for exocellular polysaccharides or phenotypes associated with these polysaccharides are valuable tools for academia and industry.These methods allow selection of polysaccharide-producing strains with appropriate rheological properties to be used in starter cultures for applications in food or other industrial sectors.Generally, screening methods should be effective in differentiating bacterial isolates with the desired phenotype, e.g.enable screening for isolates with high exocellular polysaccharide production or alternatively screening for isolates that generate improved texture in the medium used.The screening method should ideally allow rapid measurement or assessment of the desired feature, permitting high numbers of isolates to be screened, and finally the method should be inexpensive.When designing the method, it is also important to ensure that the screening is done in a relevant matrix, optimally close to the application; if in situ texture development in acidified milk is desired, the procedure should be suitable to screen for texture in acidified milk.Figure 7 shows the different screening methods described hereafter.
Several screening strategies have been devised to identify exocellular polysaccharide-producing bacteria.Vedamuthu and Neville (1986) used 1.0 mL pipettes to test the ropiness of acidified reconstituted non-fat drymilk, assessing mucoidness by the stringiness of the free flowing milk gel.Van den Berg et al. (1993) used this approach to screen close to 600 LAB and found 30 strains displaying a ropy phenotype using EPS selection medium.Using a modified version of the medium, Ludbrook, Russell and Greig (1997) isolated 11 LAB from salami, ham and olives that showed levels of exocellular polysaccharides ranging from 114 to 530 mg L −1 .LAB from traditional Nigerian fermented foods were screened for ropiness by the above-mentioned method allowing identification of texturizing Lc. mesenteroides, Lb. plantarum, Lb. acidophilus and Lb.brevis (Sanni et al. 2002).The method of measuring ropiness with a pipet is also described by Mozzi et al. (2001), defining ropy as strings longer than 6 mm.The method was used to screen 201 LAB from Argentina (Mozzi et al. 2006).Only six thermophilic and six mesophilic strains were identified to generate ropy milk.This feature did not correlate with exocellular polysaccharide production measured as polymer dry mass, but out of six high molecular weight (>1000 kDa) polysaccharide-producing strains five were ropy (Mozzi et al. 2006).Moreover, the method was used to screen 94 putative LAB isolates from Indian Dahi and sour raw milk, identifying 47 mesophilic isolates with ropy phenotype (Behare, Singh and Singh 2009).A variant to the ropy-screen method employs an inoculation wire loop to assess string generation upon touching an LAB colony.When the string is longer than 5 mm, the strain is considered ropy (Dierksen, Sandine and Trempy 1997).In a study of putative LAB isolates from traditional Iranian goat and ewe's milk, a total of 102 isolates were screened for a ropy phenotype, but none of the isolates showed ropiness (Hajimohammadi Farimani et al. 2016).Although the ropy screen is easy to perform, it is difficult to standardize, since the output is qualitative rather than quantitative.Furthermore, although the ropy phenotype observed with a particular isolate demonstrates that the strain can produce exocellular polysaccharide, it is not clear whether exocellular polysaccharide is produced in low or high amounts (Vijayendra et al. 2008).
A screening approach employing microhaematocrit capillaries to measure efflux, a parameter that correlates with viscosity measurements, has also been developed (Ricciardi, Parente and Clementi 1997).LAB grown on various carbon sources were screened for viscosity using this method and several isolates of Lb. bulgaricus, Lc. dextranicum and Lactococcus sp. were identified as polysaccharide-producing strains.Forty-one LAB isolates from Italian sourdough fermentations were screened for exocellular polysaccharide production by both picking colonies from solid media and detecting ropiness using microhaematocrit capillaries to screen for ropy phenotype (Zotta et al. 2008).Ropy phenotypes were detected for Lb.plantarum isolates grown on glucose, maltose or sucrose; Lb. paraplantarum on maltose or sucrose; and Lc.mesenteroides and Weissella cibaria grown solely on sucrose as carbon source.The use of microhaematocrit capillaries as a screening method is appealing, since it is easy to perform and gives an indication of whether the produced polysaccharide causes changes in the food matrix.
Visual inspection of slimy or mucoid colonies on solidified media is probably the most frequently used screening method for EPS production.This approach is particularly good at identifying high-level EPS producers, since the slime around the colony is hydrated EPS.This phenotype is often associated with extracellular HoPS (De Vuyst and De Vin 2007).Tallgren et al. (1999) screened 600 soil isolates by looking for slime development around colonies grown on agar plates with sucrose and found 170 slimy isolates.A similar strategy was used by Smitinont et al. (1999), screening 104 isolates from traditional Thai fermented foods and finding 7 slimy colonies.There are several other reports using slimy colony morphology as first screening step (Malik et al. 2009;Bennama et al. 2012;Ishola and Adebayo-Tayo 2012;Paulo et al. 2012).The disadvantage of this method is that it typically detects only strains with high polysaccharide production.Strains that produce polysaccharides via the Wzy-dependent pathway are rarely giving rise to slimy colonies; this was demonstrated for the polysaccharide-producing S. thermophilus LY03 (De Vuyst et al. 1998).
Staining methods are used for detecting exocellular polysaccharide-producing cells, e.g. by using ruthenium red, neutral red, calcofluor white, congo red or Indian ink.Ruthenium red was used to screen for S. thermophilus mutants with loss of ropy phenotype, which appear as red colonies on milk agar media, while ropy wild-type strains appear white (Stingele, Neeser and Mollet 1996).This red/white selection was also applied to L. lactis strains.Use of ruthenium red as a general polysaccharide indicator is difficult.We have tested approximately 90 S. thermophilus and 30 L. lactis isolates with ruthenium red and we observed high strain to strain variation and no clear correlation between 'white' ruthenium red isolates and their ability to texturize fermented milk (unpublished results).Neutral red dye was used to select ethidium bromide mutants of ropy S. thermophilus that produced higher amounts of exocellular polysaccharide than the wild-type strain (Escalante et al. 2002).Calcofluor white has been used to select polysaccharide negative mutants of Rhizobium meliloti (Leigh, Signer and Walker 1985) and to detect polysaccharides in biofilms by Flavobacterium columnare (Cai, De La Fuente and Arias 2013) and S. pneumoniae (Domenech, Garcia and Moscoso 2012).Congo red has been used in agar media for biofilm staining of B. subtilis (Bedrunka and Graumann 2016) and Staphylococcus aureus (Darwish and Asfour 2013).Although staining assays are relatively easy to perform, the visual interpretation can be difficult due to small differences in positive and negative signals.It has been suggested that lectins, glycoproteins that bind to sugar monomers of polysaccharides, can be used to differentiate between CPS and EPS in S. thermophilus (Robitaille et al. 2006).Lectins typically have specificity for certain carbohydrate epitopes and may therefore not be used as generic CPS screening tools.An alternative to lectins are polyclonal antibodies, used in Quellung reaction and agglutination tests, which have been the gold standard for serotyping of human pathogens, such as S. pneumoniae (Geno et al. 2015).In particular, agglutination test can be envisioned as a screening tool for identification of polysaccharide-producing LAB.These tests based on the agglutination reaction that occurs between the polysaccharide-producing strain and the type-specific antibody in the serum are reported to be highly sensitive.The limitation however is the availability of the specific polyclonal antibody.To the best of our knowledge, this method has not been applied to screen for polysaccharide-producing LAB.
All the screening approaches mentioned above rely on exocellular polysaccharides causing a physical change in the environment that can be observed by visual inspection.A more direct approach involves quantification of the exocellular polysaccharide produced.In a screening study of 182 Lactobacillus isolates cultivated in liquid broth, the mass of EPS in the supernatant was determined gravimetrically (van Geel-Schutten et al. 1998).Several samples with EPS dry weight above 100 mg L −1 were found when the strains were grown on sucrose.Growth on raffinose or lactose only rarely resulted in EPS production.A similar approach to determine polymer dry weight was reported by Smitinont et al. (1999).When LAB are grown in milk or other high protein content media, protein precipitation is required for reliable use of the polymer dry weight method (De Vuyst et al. 1998).This procedure was used to determine the monomeric composition of EPS from four different S. thermophilus (De Vuyst et al. 1998).
We have recently developed a method for high-throughput texture screening of fermented milk samples.Bacterial isolates are grown in 96-well microtiter plates with milk as cultivation medium.After acidification of the milk, samples are aspirated and dispensed by a Hamilton MicroLab Star liquid handling unit.The liquid handler has a pressure sensor located in the headspace of each pipetting channel.Pressure data from each sensor are collected and the resulting pressure curves used to differentiate between strains with low, medium and high texture phenotypes (Cantor et al. 2014).We find this screening approach useful as it allows selection of strains with texturizing phenotype in the desired matrix, i.e. fermented milk.
Molecular screening approaches have also been used in the search for new exocellular polysaccharide-producing isolates or as a means to differentiate between polysaccharide-producing isolates at the gene level.The strategy typically comprises the design of degenerate primers based on LAB gene sequences followed by PCR reaction, cloning and sequencing of PCR products.This approach has been mainly applied to identify glucan and fructan-producing strains (Krajl et al. 2003;Tieking and Gänzle 2005;Malik et al. 2009).With the drop in prices for genome sequencing, it is now easier to study eps gene sequences and architecture of the clusters encoding eps genes, and hence differentiate strains at the genetic level.However, the identification of eps gene clusters is unlikely to allow drawing conclusions regarding amount and type of polysaccharides or even rheological properties when applied in milk fermentations.

Strain improvement for polysaccharide production
The demand for exocellular polysaccharide-producing LAB in various applications is the catalyst for the development of new strains with increased production or improved technofunctional properties.Strain improvement can be pursued by two major approaches: (i) genetic and/or metabolic engineering and (ii) classical strain improvement methods (Fig. 8).The choice of either approach is dictated mainly by the final application.The tight requirements of regulatory agencies and the negative perception by consumers of genetically modified foods exclude the use of recombinant DNA technologies to improve the performance of bacteria used in food manufacture.Thus, for any food application the approach of choice is the use of natural strain improvement methodologies.Here we will summarize the research efforts to improve polysaccharide production and/or the rheological properties in LAB.
(i) Genetic/metabolic engineering approaches.Engineering of complex pathways poses great challenges, and efforts to increase exocellular polysaccharide production in LAB were met with limited success (Welman and Maddox 2003;Gaspar et al. 2013).This complexity is further exacerbated by the paucity of knowledge on key functions of the Wzy-dependent pathway, such as the Wzx flippase, Wzy polymerase and priming GTs.The potential of LAB to generate sugar nucleotides required for exocellular polysaccharide synthesis has been exploited for the heterologous expression of pneumococcal CPS and hyaluronic acid in L. lactis (reviewed in Gaspar et al. 2013).The resulting recombi-nant strains are potential hosts for the safe production and delivery of pneumococcal capsule vaccine antigens or hyaluronic acid, a polysaccharide with many applications in medicine, cosmetics and specialty foods.
In L. lactis NIZO B40, an increase in polysaccharide production was achieved by cloning the entire eps gene cluster present in plasmid pNZ4000 into a high copy number vector (Boels et al. 2003).A 9-fold elevated copy number led to a 3-fold increase in the expression level of eps genes, resulting in circa 4-fold increase in polysaccharide production.This result further substantiated the view that increased levels of activated sugar precursors do not influence NIZO B40 exocellular polysaccharide production levels, but rather that the production is limited by the activity levels of the enzymes in the biosynthetic pathway.The relative carbon flux towards polysaccharide production increased 3-fold, and was accompanied by a substantial reduction of growth rate and a lower final optical density.Together these results indicate that increased exocellular polysaccharide production imposes a significant metabolic burden, possibly due to competition for sugar nucleotides and energy, which are utilized in both polysaccharide production and growth.
The complete eps gene cluster from S. thermophilus SFi6 was transferred to L. lactis MG1363, a non-EPS-producing model LAB.The polysaccharide obtained in L. lactis had a similar high molecular weight, but a different structure, namely a change in backbone composition from GalNAc to Gal and a side chain modification, i.e. loss of the branching Gal (Stingele et al. 1999).The substitution of GalNAc by Gal most likely results from the absence of an UDP-N-acetylglucosamine C4-epimerase activity in L. lactis MG1363, which thereby is incapable of generating UDP-GalNAc for incorporation in the repeating unit of the recombinant polysaccharide.Based on the findings, the authors speculated that the S. thermophilus Sfi6 Wzy polymerase could recognize and polymerize a repeating unit that differs in the backbone, as well as in the side chain from its native substrate.
Galactose is assimilated via the Leloir pathway in S. thermophilus (Fig 4).The enzyme that links the Leloir pathway with glycolysis is α-phosphoglucomutase (α-PGM), which catalyzes interconversion of Glc-6-P and Glc-1-P.Streptococcus thermophilus mutants lacking α-PGM activity as well as mutants with overexpressed pgmA produced the same amount of exocellular polysaccharide as the parent strain when grown on lactose, indicating that the Leloir pathway is important for supplying exocellular polysaccharide precursors (Levander and Radstrom 2001).Homologous overexpression of the galU gene in S. thermophilus LY03 led to a 10-fold increase in GalU activity, but did not have any effect on the exocellular polysaccharide yield when lactose was the carbon source.However, when galU was overexpressed in combination with pgmA the exocellular polysaccharide yield increased from 0.17 to 0.31 g mol −1 of carbon from lactose.A galactose-fermenting S. thermophilus LY03 mutant, TMB 6010, with increased activity of the Leloir pathway enzymes, and higher levels of Glc-1-P and Glc-6-P, was also found to have a higher exocellular polysaccharide yield (0.24 g mol −1 of carbon) than the parent strain.The exocellular polysaccharide yield was further improved to 0.27 g mol −1 of carbon by overexpressing galU in this strain, but the highest exocellular polysaccharide yield, 0.36 g mol −1 of carbon, was obtained when pgmA was knocked out in the galactose-proficient strain (Levander, Svensson and Radstrom 2002).Overexpression of both pgmA and galU in TMB6010 led to a 3.3-fold increased polysaccharide production (Svensson et al. 2005).Strikingly, the viscosity in fermented milk did not increase significantly in response to the increased exocellular polysaccharide amounts.
Exocellular polysaccharide production could also be increased in S. thermophilus by introducing a galactokinase (galK) gene into a galactose-deficient strain (Robitaille et al. 2009).The recombinant strain produced 1.3 times more polysaccharide, but when used in combination with Lb. bulgaricus for yoghurt manufacture, no significant increase in polysaccharide production was detected.
(ii) Classical strain improvement.eps gene clusters are often plasmid encoded in L. lactis (Vedamuthu and Neville 1986;Neve, Geis and Teuber 1988;van Kranenburg et al. 1997;van Kranenburg, Kleerebezem and de Vos 2000;Forde and Fitzgerald 2003) and in some lactobacilli, e.g.Lb. casei (Kojic et al. 1992) and Lb.paraplantarum (Zivkovic et al. 2015).When eps gene clusters are located on plasmids, the mucoid phenotype can be transferred by conjugation from one strain to another strain as demonstrated in L. lactis for EPS plasmids pSRQ2202 (Vedamuthu and Neville 1986) and pNZ4000 (van Kranenburg and de Vos 1998), or by protoplast transformation as shown for plasmid pVS5 (von Wright and Tynkkynen 1987).From a regulatory point of view, such conjugation events are natural, and hence the plasmid recipients are not considered genetically modified organisms.
The galactose-negative phenotype, identified for the majority of S. thermophilus strains, is due to the lack of GalK activity.The isolation of spontaneous galactose-proficient mutants from the polysaccharide-producing S. thermophilus CHCC6008 by selective pressure in galactose broth resulted in strain CHCC11379.This mutant showed slightly increased exocellular polysaccharide production, and improved viscosity and shear stress of a fermented milk (Janzen and Christiansen 2011b).
Based on the hypothesis that polysaccharide overproducing mutants confer bacteriophage resistance by increasing a physical barrier against infecting phages, exocellular polysaccharide production was further increased by isolating phage-resistant mutants of strain CHCC11342 (Janzen and Christiansen 2011a).CHCC11342 is a galactose-proficient, exocellular polysaccharide-producing S. thermophilus strain, isolated as mutant from CHCC6008.The resulting phage hardened mutant, CHCC11977, when used to acidify milk, led to viscosity, shear stress and gel firmness increased approximately by 20% as compared to CHCC6008.This stands in contrast to the results of the study by Deveau, Van Calsteren and Moineau (2002), which showed no impact of exocellular polysaccharide production on either phage sensitivity or resistance.The combination of CHCC11977 with a polysaccharide-producing Lb. bulgaricus strain resulted in increased shear stress and mouth feel in yoghurt application trials.In a similar way, mutants isolated after phage challenge of Lb. bulgaricus strain CHCC10019 increased shear stress of fermented milk by 25% (Janzen and Christiansen 2011a).
Resistance to antibiotics was recently discovered as an attractive way to increase rheological parameters of starter cultures.Six out of 21 Lb.bulgaricus CHCC13995 ampicillin-resistant mutants revealed doubling of efflux time of fermented milk measured by a volumetric pipette, indicating increased viscosity.A positive effect on ropiness and gel stiffness of the fermented product was also observed when the mutant was used in combination with the non-polysaccharide-producing S. thermophilus strain CHCC4895 (Johansen, Soerensen and Kibenich 2013).Resistance towards the antibiotic D-cycloserine was proven successful to increase shear stress and gel stiffness in milk acidified with the resistant Lb. bulgaricus and S. thermophilus strains.A D-cycloserine-resistant mutant of Lb. bulgaricus CHCC13995, CHCC12945, triggered a shear stress increase of 10% and a gel stiffness increase of 50% in fermented milk as compared to the wild-type strain.For S. thermophilus, the D-cycloserine-resistant mutant CHCC13236 induced a 15% increase in shear stress and 19% in gel stiffness of the fermented milk product, as compared to the wild-type CHCC13994 (Kibenich, Soerensen and Johansen 2012).
Isolation of LAB strains with improved rheological properties by selection with antibiotics, phages or essential nutrients allows direct application of these superior strains in food applications.The examples presented here demonstrate the feasibility to obtain LAB isolates with improved ability to produce polysaccharides.The work done so far provides the foundations for further exploitation of LAB-produced polysaccharides in novel applications.

CONCLUDING REMARKS
LAB have a long tradition of safe usage in the manufacture of fermented beverages and foods.One of the beneficial effects of LAB is enhancement of rheological properties through in situ production of polysaccharides during fermentation of the raw food materials.Due to their ability to form hydrocolloids in aqueous solutions, polysaccharides can serve as emulsifiers, thickeners and gelling agents.Other attributes in food are stabilization, suspension of particulates, control of crystallization, inhibition of syneresis, encapsulation and film formation.Despite this array of versatile applications, the use of LAB polysaccharides as food ingredients/additives has so far been restricted to a single polymer, the HoPS dextran produced by Lc. mesenteroides.
Therefore, LAB are essentially an untapped reservoir of structurally diverse polysaccharides to be used as food additives.Industrial use of microbial polysaccharides is, however, not restricted to the food industry, and the number of (bio)technological applications in pharmaceutical and cosmetics industries, medicine, water treatment, oil industry (enhanced oil recovery), agriculture and construction materials continuously increases (Ullrich 2009;Freitas, Alves and Reis 2011;Srinivasan 2013;Freitas, Alves and Reis 2014;Moscovici 2015;Schmid, Sieber and Rehm 2015).Exploring the natural diversity and versatile technofunctional properties of polysaccharides produced by LAB holds great promise in highvalue market niches, such as pharmaceuticals, cosmetics and medical applications (adhesives, sutures, vaccine adjuvants).Furthermore, polysaccharides are biocompatible, biodegradable and non-toxic natural biopolymers.Thus, the demand for biosustainable products in modern, ecofriendly societies render bacterial polysaccharides as promising replacements for petroleum-derived polymers.
A major limitation to the use of bacterial, and in particular LAB, polysaccharides for industrial applications is the cost associated with their production.On the technological side, highvalue substrates (carbon sources) are used for bacterial biomass production and the downstream process costs associated with purification and isolation are also on the high end.Adding to this, polysaccharides are generally produced by bacteria in low titers and yields, most likely because their synthesis is metabolically costly to the cell.
Thus, harnessing polysaccharide diversity and functionality for commercial purposes requires a multidisciplinary approach to boost polysaccharide production in LAB in a rational way.To this end, metabolic engineering can be applied to develop suitable hosts for sustainable and cost-effective production of polysaccharides, ultimately leading to product development and commercialization.Effective metabolic engineering relies on an in-depth understanding of metabolic pathways, protein functions and underlying regulatory mechanisms.To date, a detailed functional characterization of the major polysaccharide biosynthetic route in LAB, the Wzy-dependent pathway, has not been undertaken.In this review, the pneumococcal model for biosynthesis of CPS was extrapolated for the synthesis of exocellular polysaccharides by LAB.The assumptions were that S. pneumoniae and LAB are taxonomically closely related, and there is considerably high-sequence homology for the assembly machinery and modulatory proteins.In addition, regulation of polysaccharide biosynthesis in LAB remains mostly elusive.To overcome these obstacles, a systems approach combining genomics, functional genomics and high-throughput experimentation with powerful computational tools for data integration and interpretation is required.The knowledge generated combined with an ever-growing genetic engineering toolbox is expected to provide the foundations for the development of optimized cell factories.
In the context of systems metabolic engineering, the focus has shifted from improving production of native compounds to harnessing the microbial machinery for the production of novel molecules.Therefore, systems metabolic engineering offers a unique opportunity to generate designer polysaccharides with improved functionalities.Combinatorial assembly of pathways has been made easy with the advance of synthetic biology.In principle, tailor made polysaccharides could be generated by designing synthetic eps gene clusters combining genes or modules with the desired functions from different sources and expressing the synthetic cluster in an optimized host.The plethora of GTs observed in the loci for polysaccharide production provides an opportunity to continually generate new strains producing unique polysaccharides by gene shuffling.Despite the relative simplicity in pathway assembly, main hurdles arise from poor functional characterization of the molecular factors involved.Indeed, the substrate specificity of Wzy polymerases and Wzx flippases has never been studied in LAB and to the best of our knowledge remains a mystery even in well-characterized enterobacteria.As for GTs, only a very limited number has been biochemically characterized, and thus substrate specificity and GT activity remain largely unknown in LAB.
LAB are clearly a great source of polysaccharides with industrial applications.The already existing natural diversity can even be further expanded by implementation of synthetic biology approaches.However, increasing polysaccharide production via metabolic engineering and/or generating tailored polysaccharides requires a better understanding of the molecular mechanisms underlying polysaccharide synthesis.This can be achieved by combining structural polysaccharide data, biochemical characterization and systems biology approaches (omics experimentation and bioinformatics/modeling).Finally, we surmise that the availability of tailored polysaccharides is an asset to improve our understanding of structure-function-application relationships.

Figure 1 .
Figure 1.Cell surface-associated polysaccharides in bacteria.Capsular polysaccharides (CPS) and exopolysaccharides (EPS) are common to Gram-negative (a) and Grampositive bacteria (b).Lipopolysaccharides (LPS) and lipooligosaccharides (LOS) are found only in Gram-negative bacteria, whereas Gram-positive bacteria synthesize cell wall polysaccharides (CW-PS) such as wall teichoic acids (TA), lipotheichoic acids (LTA) and pellicles.A peptidoglycan (PG) layer is present in the cell walls of both Gram-positive and Gram-negative bacteria.Closed circles, monosaccharides.

Figure 2 .
Figure 2. Pathways for the synthesis of bacterial polysaccharides.(a) The Wzy-dependent pathway (synthesis of LPS O-antigen polysaccharide in Gram-negative bacteria, CPS and EPS in both Gram-negative and Gram-positive bacteria) represented for Gram-positive bacteria.Main players are the Wzy polymerase and Wzx flippase; details for this pathway are shown on Fig. 5, but a repeating unit is synthesized intracellularly, flipped and polymerized in the outer face of the cytoplasmic membrane.(b) The ABC transporter-dependent (CPS in Gram-negative bacteria; potentially other CW-PS) represented for Gram-negative bacteria.The polysaccharide chain, anchored on a poly-2-keto-3-deoxyoctulosonic acid linker in the cytoplasmic face of the inner membrane, is assembled by the action of GTs; the finished polysaccharide is exported via an efflux pump complex composed of ABC transporters spanning the inner membrane and periplasmatic proteins of the OPX family spanning the outer membrane.(c) The synthase-dependent pathway (CPS, EPS) performs the polymerization and transport by a single synthase complex, which secretes the complete polymer strands across the membranes and cell wall.(d) Extracellular synthesis of HoPS by the use of a single GT (sucrase) protein.Blue circle, glucose; yellow diamond, galactose; pink square, rhamnose; green triangle, fructose.CM, cytoplasmic membrane, OM, outer membrane; PG, peptidoglycan; GT, glycosyltransferase; Und-P, undecaprenylphosphate; P, phosphate; k, poly-2-keto-3-deoxyoctulosonic acid linker.

Figure 3 .
Figure 3. Schematic genetic organization of the eps gene clusters in different LAB based on several representative strains and S. pneumoniae strain D39.Gene functional grouping is marked with different colors.Relative localizations of eps genes with homologous functions are indicated with connection bars.All the genes are transcribed in one direction except for a few genes oriented in the opposite transcriptional sense, which are indicated with arrows.Genes with unknown functions or functions not related to the EPS biosynthesis are in white.GT, glycosyltransferase; IS, transposase; NDP-sugar, nucleotide diphospho-sugar.The 'GT' functional module does mainly include GTs, but it can also include other enzymes that modify the oligosaccharide repeat unit structure.For instance, Oenococcus oeni PSU-1 ORF marked with a star is a putative galactoside O-acetyltransferase.

Figure 5 .
Figure5.Proposed model for biosynthesis of polysaccharides via the Wzy-dependent pathway.This model is inspired in the findings from S. pneumoniae and adapted from the model proposed byNourikyan et al. (2015).The genetic locus shows the genes involved in the synthesis and export of exocellular polysaccharidess in LAB.As an example, assembly and polymerization of the polysaccharide produced by L. lactis ssp.cremoris NIZO B40 is shown.In the eps gene clusters (generic and B40), the genes coding for the polysaccharide assembly machinery, glycosyltransferases, the EpsBCD phosphoregulatory system and synthesis of NDP-sugars are shown in green, orange, yellow and pink, respectively.The same color scheme is used for representing the proteins in the Wzy-dependent pathway.The reactions catalyzed by each GT gene product are indicated in the upper-left corner gray box.These reactions occur in the cytoplasm CM, cytoplasmic membrane.

Figure 6 .
Figure 6.Priming glycosyltransferases in LAB and their characteristics.All LAB priming GTs functionally characterized so far fall into the hexose-1-phosphate transferase category.The WciI present in S. pneumoniae serotype 4 (TIGR4) is shown as an example of a hexosamine-1-phosphate transferase, but its activity as an Nacetylgalactosamine-1-phosphate transferase has not been demonstrated experimentally.WciI can presumably catalyze the transfer of GalNAc and/or GlcNAc (Aanensen et al. 2007).Abbreviations as in Fig. 4. FucNAc, N-acetylfucosamine.Rods represent the four protein types.The shaded area indicates the bacterial transfer domain of family PF02397.

Figure 7 .
Figure 7. Schematic overview of exocellular polysaccharide screening methods.Traditional methods used in the past 15-30 years are shown at the top, and emerging microtiter plate-based methods are shown at the bottom.The traditional methods often rely on visual inspection of samples, or manual measurement of either run through of liquid samples in inverted capillaries or gravimetric mass determination of EPS.The automated methods are high throughput and quantitative.See the text for further details.UHPLC-UV-ESI-MS: ultrahigh-performance liquid chromatography with ultra violet and electrospray ionization ion trap mass spectrometer.

Table 4 .
Overview of polysaccharide-producing strains and their effects on physicochemical properties of fermented milk products.