Bacterial glycobiology: rhamnose-containing cell wall polysaccharides in Gram-positive bacteria

1The composition of the bacterial cell wall is critical for fundamental features such as bacterial cell shape, protection from and interaction with the environment. Carbohydrates are the most abundant molecules in the Gram-positive cell wall with much of it incorporated in the thick layer of peptidoglycan (15–100 nm) that surrounds the cell membrane (Silhavy, Kahne and Walker ). Peptidoglycan is composed of alternating β-1,4-linked N-acetylglucosamine (GlcNAc) and N-acetylmuramic acid (MurNAc) polysaccharide strands that are cross-linked by short peptides to form a three-dimensional network. Individual species tailor important physical properties of their peptidoglycan such as elasticity and porosity through the composition of the peptide bridge, the amount of crosslinking and chemical modifications of the composing glycan residues (Vollmer, Blanot and de Pedro; Vollmer and Seligman ). Peptidoglycan also acts as a scaffold for other critical cell wall structures. For example, proteins containing LPXTG amino acid motifs are covalently attached to the peptidoglycan peptide bridge through the enzymatic action of sortase A (Hendrickx et al. ).

2Cell wall polysaccharides (CWP), such as capsular polysaccharides and wall teichoic acids (WTA), are anchored to peptidoglycan GlcNAc or MurNAc, covering the bacterium with a layer of glycans that is directly exposed to the environment (Deng et al.; Swoboda et al.; Yother ). The predominance of capsulated species among bacterial pathogens instigated studies on the role of capsular polysaccharides in infectious disease pathogenesis. As a result, effective capsule polysaccharide conjugate vaccines were developed against various species including Neisseria meningitidis, Haemophilus influenzae type B and Streptococcus pneumoniae. For some human bacterial pathogens, such as Staphylococcus aureus and Streptococcus pyogenes, the role of polysaccharide capsule in pathogenesis is less pronounced (O'Riordan and Lee; Flores et al. ). In addition, the discovery of host pattern-recognition receptors with specificity for carbohydrates as well as technological advances in the area of complex carbohydrate analysis and synthesis have fueled initiatives to investigate bacterial glycobiology more comprehensively by unraveling the structure, biosynthesis and functions of bacterial polysaccharide structures.

3In this review, we focus on a subclass of secondary CWP that we refer to as rhamnose-containing CWP (RhaCWP). L-rhamnose is commonly found in bacteria but is not used or produced by humans (Maki and Renkonen; Adibekian et al. ). Interestingly, L-rhamnose is often essential for bacterial virulence or even viability (Maki and Renkonen ), making its biosynthesis pathway an attractive therapeutic target. We will therefore review the current knowledge regarding L-rhamnose biosynthesis and functions in bacteria in more detail. Recent insights into the genetic basis and function of RhaCWP in two important human pathogens, S. pyogenes (Group A Streptococcus) and Streptococcus agalactiae (Group B Streptococcus), have emphasized the critical role of these molecules in cell wall biogenesis and pathogenesis (Caliot et al.; van Sorge et al. ). This functional information combined with the localization and abundance of RhaCWP suggests that parallels can be drawn with WTA in other Gram-positive bacteria. WTA structure and function have been reviewed extensively (Weidenmaier and Peschel; Swoboda et al.; Brown, Santa Maria and Walker ). Therefore, we will only highlight specific parallels with WTA biology throughout this review. Finally, from their historic discovery (Lancefield ) and recent insight from bacterial genome sequences, it is apparent that RhaCWP are likely more widespread in Gram-positive cocci within the order Lactobacillales. We will provide an overview of their inferred distribution and review the literature for selected species with regard to structure, genetics, biosynthesis and function. We will end by discussing the potential therapeutic and biotechnological applications of this important class of CWP.

4Most knowledge regarding the architecture and biology of the Gram-positive cell wall is derived from the model organisms Bacillus subtilis and S. aureus. For these and other species, WTA is a major cell wall component representing up to 60% of the total cell wall mass. WTA are anionic glycopolymers that are covalently attached to the peptidoglycan MurNAc residue (Swoboda et al. ). Most commonly, WTA are composed of a poly-ribitolphosphate (-RboP)- or poly-glycerolphosphate (-GroP-) backbone with modifications such as glycosylation and D-alanylation, the latter of which can neutralize the negative charge of the abundant phosphates in the WTA backbone (Brown, Santa Maria and Walker ). However, the exact chemical composition varies among and even within species (Neuhaus and Baddiley; Weidenmaier and Peschel; Winstel, Xia and Peschel ). It has long been recognized that not all Gram-positive bacteria incorporate polyRboP- or a polyGroP-based WTA into their cell wall. Instead, the cell walls of many species within the order Lactobacillales contain peptidoglycan-anchored polysaccharides that are characterized by the presence of L-rhamnose. Probably the first reports of RhaCWP in the genus Streptococcus date back to the late 1920s and early 1930s by Rebecca Lancefield (Lancefield,, ). Her seminal work enabled the development of a new streptococcal classification system—in addition to hemolytic typing—based on the differential antigenic properties of streptococcal CWP, referred to as ‘C-substance’ or ‘Group Antigen’ (Lancefield ). The streptococcal serotyping scheme initially only discriminated Group A–E streptococci but was expanded to comprise as many as 20 serotypes called Groups A–V (excluding I and J) (Facklam; Kohler ). The Lancefield typing system has been instrumental to link mild and severe human and animal diseases to specific bacterial groups within the genus Streptococcus, most notably S. pyogenes, causing more than 700 million infections resulting in over 500 000 deaths worldwide annually (Carapetis et al. ), and S. agalactiae, a pathogen affecting mainly neonates and the elderly (Phares et al.; Edmond et al. ). Furthermore, the Lancefield typing scheme enabled the development of rapid diagnostic tests that aided clinically relevant ‘species identification’ (Lue, Howit and Ellner ). Increased recognition of additional strain characteristics, such as nutrient requirements and later the use of 16S rRNA for classification (and more recently whole genome sequencing), demonstrate that the Lancefield typing system cannot discriminate to the species level such that a single Lancefield Group represents one species. For example, the Group A antigen was long thought to be the exclusive molecular marker of S. pyogenes, yet Streptococcus castoreus is also noted to react with Group A antisera in commercial diagnostic kits (Lawson et al. ). Vice versa, a single species can express different Group antigens; strains of Streptococcus dysgalactiae subsp. equisimilis commonly express either the Group C or G antigen (Broyles et al.; McMillan et al.; Takahashi, Ubukata and Watanabe ) and occasionally Group A carrying S. dysgalactiae subsp. equisimilis strains are identified (Tanaka et al.; McMillan et al. ). Consequently, the taxonomy and nomenclature of the genus Streptococcus has been reevaluated and reclassified over the years (Facklam; Kohler ). This resulted in the split of the genus Streptococcus into three genera, i.e. Enterococcus, Lactococcus and Streptococcus (Schleifer et al. ), as well as subsequent description of several novel members within these genera. Two species covered in this review, Enterococcus faecalis and Lactococcus lactis, were formerly known as Streptococcus faecalis (Lancefield Group D) and Streptococcus lactis (Lancefield Group N). Today, more than 150 different species are known within these three genera, which remain classified within the order Lactobacillales (http://www.bacterio.net/-classifphyla.html#Firmicutes; Price et al. ).

5Although the direct correlation between Lancefield serotyping and species clearly no longer persists, bioinformatic analyses of genome sequences indicates that many of the species within these three genera express RhaCWP (see section ‘Distribution of RhaCWP throughout bacteria’). We focus here on the species relevant to food production and human health and with experimental evidence for the presence of RhaCWP: Lancefield Groups A, B, C, E and G Streptococcus represented by S. pyogenes, S. agalactiae, Streptococcus equi subsp. zooepidemicus, Streptococcus mutans and S. dysgalactiae subsp. equisimilis, respectively, as well as E. faecalis and L. lactis.

6The RhaCWP is a major component in streptococcal species comprising about 40%–60% of the cell wall by weight (McCarty; Krause and McCarty; Krause; Doran and Mattingly ). Correspondingly, cell wall thickness is visually reduced by 40%–50% after chemical extraction of the Group-specific carbohydrate (Swanson and Gotschlich; Wagner and Wagner ). Recent mutagenesis studies confirmed the structural importance of streptococcal group antigens; complete loss of RhaCWP expression resulted in severe growth and cell division abnormalities (Fig. 1; Tsukioka et al.; Caliot et al.; van Sorge et al.; van der Beek et al. ).

7RhaCWP are localized on the outermost surface of the cell wall but are likely also intercalated within the mesh-like structure of the peptidoglycan layer since antibodies directed against these structures bind to both sides of isolated cell walls (Swanson and Gotschlich; Wagner and Wagner; Wagner et al. ). Group-specific antigens can also be isolated at high yield in the growth medium (Carey et al.; De Cueninck, Shockman and Swenson; Doran and Mattingly ), possibly as a result of cell wall catabolism during growth. Compared to the cell walls of S. aureus and B. subtilis, streptococcal species appear to lack expression of polyol-based WTA and, typically, lack orthologues of the critical WTA biosynthesis enzymes TagB, TagD and TagF (Sutcliffe, Black and Harrington ) (and unpublished observations). In contrast to streptococci, homologues of WTA biosynthesis genes are found in lactococcal genomes. Consequently, L. lactis strains likely express both polyol-based WTA and RhaCWP as part of their cell wall (Chapot-Chartier and Kulakauskas ). E. faecalis expresses the most extensive surface glycome including WTA, LTA, capsular polysaccharide and a RhaCWP called Enterococcal polysaccharide antigen (Epa) (Hancock and Gilmore; Teng et al.; Thurlow, Thomas and Hancock; Theilacker et al. ). In contrast to RhaCWP in streptococci and L. lactis, Epa appears to be buried in the cell wall precluding interaction with the immune system, at least under laboratory conditions (Hancock and Gilmore ). It should however be noted that sera from patients do contain Epa-specific antibodies (Xu, Murray and Weinstock; Xu et al.; Teng et al. ). Also, the presence of Epa is visible under transmission electron microscopy (TEM) as a separate electron dense outer layer (Rigottier-Gois et al. ), similar to the RhaCWP layer observed in L. lactis and S. agalactiae (Chapot-Chartier et al.; Caliot et al. ) (Fig. 2). In early studies, this electron dense outer layer was known as the outer lamina and was initially described as microcapsule (Baker and Kasper ). Recently, the term pellicle was coined (Chapot-Chartier et al. ) and is preferred since the term does not inherently imply that this layer is either ‘outer’ or of a specific composition (such as capsule). Interestingly, the pellicle is not as impenetrable as TEM images suggest. In S. agalactiae, topographic imaging and atomic force microscopy-based single-molecule mapping on live bacteria revealed that peptidoglycan strands can still be probed in the presence of the GBC (Beaussart et al. ). However, the pellicle does shield peptidoglycan to some extent, since loss of GBC expression through genetic manipulation renders S. agalactiae extremely sensitive to the activity of peptidoglycan-cleaving mutanolysin (Caliot et al. ). Similarly, atomic force microscopy studies on pellicle-deficient L. lactis uncovers peptidoglycan periodic bands orientated parallel to the short axis of the cell (Andre et al. ).

8Studies focusing on the chemical composition of Lancefield Group antigens have demonstrated that rhamnose is the major constituent, along with variable combinations and linkages of Glc, GlcNAc, Gal, GalNAc and phosphate (Pritchard et al. ). Elucidation of Group-specific RhaCWP structures in individual streptococcal species provided structural evidence for the discriminating capacity of the Lancefield typing scheme (Fig. 3A). The serological distinction between Group A and Group C Streptococcus is explained by expression of a terminal β-linked GlcNAc side chain in the Group A Carbohydrate (GAC) versus a (GalNAc)2 side chain in the Group C Carbohydrate (GCC), respectively (McCarty; Krause and McCarty; Coligan, Kindt and Krause ) (Fig. 3A). However, biochemical and immunological characterization of CWP isolated from so-called Group A- and Group C-variant strains also noted that the GAC and GCC are structurally related. These variant strains lost Lancefield serum reactivity and displayed a variant Group antigen comprised of unsubstituted rhamnan with only trace amount of the N-acetylated sugars (McCarty; Krause and McCarty ). Occurrence of such variant strains appears to be a rare event; one Group C-variant strain was isolated as a resistant clone after exposure to virulent Group C bacteriophages (Krause ), whereas Group A-variant strains appear after multiple passages through an unnatural host, such as mice, but have never been isolated from humans (McCarty and Lancefield ). Whether strains are able to vary expression, composition or length of RhaCWP during natural infection is currently unclear. Possibly such variants may be missed in routine diagnostics screening. Alternatively, loss of specific RhaCWP epitopes or loss of the complete structure would severely hamper the ability of the bacterium to colonize or infect the host, allowing rapid eradication by the host immune system as will be described in more detail below.

9For the Group B (GBC) and G carbohydrate (GGC), rhamnose is the major antigenic determinant (Curtis and Krause,). Species carrying these structures are serologically discriminated based on the presence of either a single rhamnose in GGC versus triterminal rhamnose in GBC (Curtis and Krause ) (Fig. 3A and B). Among the streptococcal group antigens, the GBC is unique since it forms a multiantenna branching structure and is negatively charged, due to the presence of phosphodiester bonds that link different GBC repeat units (Fig. 3B). Similarly, the L. lactis RhaCWP contains phosphodiester bonds that link hexasaccharide or pentasaccharide repeating units (Fig. 3A) (Chapot-Chartier et al.; Ainsworth et al. ). Unfortunately, the structure of E. faecalis Epa has not yet been elucidated but is composed of the monosaccharides glucose, rhamnose, GlcNAc, GalNAc, galactose and probably phosphate (Hancock and Gilmore; Teng et al. ). The presence of phosphates in RhaCWP of S. agalactiae, L. lactis and possibly E. faecalis likely suggests that the functions of these structures more closely resemble functions exerted by WTA in other Gram-positive bacteria (Michon et al.,,; Sutcliffe ).

10CWP can be structurally highly complex due the number and variation in their monosaccharide components, diverse linkage types and chemical substitutions such as (de)acetylation and hydroxylation. Their biosynthesis requires the coordinated action of glycosyltransferases (enzymes that link monosaccharides), transporters and metabolic enzymes for the production of nucleotide-sugar precursors. Often, genes encoding proteins required for glycoconjugate biosynthesis are clustered on the chromosome. Their identification allows subsequent structure–function studies that help to understand the role of glycosylation in bacterial (infection) biology.

11It is surprising that despite the historical and medical importance of the Lancefield Group antigens, no studies were undertaken to decipher the genetic basis of their biosynthesis. By contrast, capsule synthesis was quickly recognized as a major virulence factor in many bacterial species and has consequently been the subject of many genetic, biochemical and functional studies (Llull, Lopez and Garcia; Cress et al. ). Initial predictions for the genetic basis of Lancefield Group carbohydrate biosynthesis were postulated upon availability of the first streptococcal genome sequences (Ferretti et al.; Glaser et al.; Holden et al.; Shimomura et al. ). For S. agalactiae, the availability of the GBC structure and genome sequence enabled a comprehensive and detailed in silico analysis that linked protein-encoding genes to the different glycosidic linkages in the GBC structure (Sutcliffe, Black and Harrington ). That study demonstrated that most, but not all, of the enzymatic activities for synthesis and transport of the mature GBC molecule are present in the predicted 15 kB GBC gene cluster (Sutcliffe, Black and Harrington ); proteins required for lipid carrier activation and peptidoglycan anchoring of GBC seemed to be encoded elsewhere on the genome (Sutcliffe, Black and Harrington ). As will be discussed below, this split in gene organization, correlating with different biosynthetic steps, appears to be a common feature of the streptococcal and lactoccocal RhaCWP biosynthesis pathway. In addition, the biosynthesis of RhaCWP requires the production of appropriate nucleotide sugars precursors. We will only cover dTDP-L-rhamnose biosynthesis here given its characteristic presence in RhaCWP, as well as the possible therapeutic implications of this pathway. We will then discuss the genetic organization and putative biosynthesis pathway of RhaCWP.

12Incorporation of L-rhamnose into polysaccharide structures in both Gram-positive and Gram-negative bacteria requires the formation of the nucleotide sugar precursor dTDP-L-rhamnose. dTDP-L-rhamnose is produced from glucose-1-phophate through a conserved four-step enzymatic reaction that has been characterized both biochemically and structurally (Fig. 4) (Giraud and Naismith; Dong et al. ). In the first step of the pathway, RmlA, a glucose-1-phosphate thymidyltransferase, converts glucose-1-phosphate into dTDP-glucose (Blankenfeldt et al. ), which is subsequently oxidized and dehydrated to form dTDP-4-keto-6-deoxy-D-glucose by the dTDP-D-glucose 4,6-dehydratase RmlB (Beis et al. ). RmlC catalyzes an unusual double epimerization reaction (Giraud et al.; Dong et al.; ), the product of which is finally reduced by RmlD, a dTDP-4-dehydrorhamnosereductase, to form dTDP-L-rhamnose (Blankenfeldt et al.; van der Beek et al. ). Initial structure elucidation of the Rml enzymes from different species, including Pseudomonas aeruginosa (RmlA), Salmonella enterica (RmlC and RmlD) and Streptococcus suis (RmlB and RmlC), demonstrated that rhamnose biosynthesis enzymes require dimerization or even dimerization of dimers to catalyze the respective enzymatic reactions (Blankenfeldt et al.; Giraud et al.; Blankenfeldt et al.; Beis et al.; Dong et al. ). However, recent structural and biochemical characterization of the S. pyogenes RmlD homologue provided surprising insight that, in this species, RmlD is active as a monomer (van der Beek et al. ). Subsequent comprehensive bioinformatics analysis of 213 putative RmlD sequences indicates that the monomeric form of RmlD is more widespread throughout the bacterial kingdom compared to the originally described RmlD from Salmonella (Blankenfeldt et al.; van der Beek et al. ). The benefit of either structure to the enzymatic reaction is currently unknown.

13In many bacteria, such as in E. faecalis (Xu, Murray and Weinstock ), Shigella flexneri (Macpherson, Manning and Morona ), L. lactis (Dupont et al. ), S. pneumoniae (Bentley et al. ) and S. enterica (Jiang et al. ), the RmlA-D proteins are encoded by a single genetic locus, rmlABCD. The rml genes are likely transcribed as an operon, although experimental evidence for that is currently lacking. In contrast, in some bacteria the rml genes display a split architecture, often clustering rmlABC, but excluding rmlD. This is for example the case in Mycobacterium tuberculosis (Ma et al. ) and several streptococcal species (Tsukioka et al.,; van der Beek et al. ). The evolutionary origin or functional benefit of a split versus clustered gene architecture is remains to be determined.

14The contribution of Rml enzymes to L-rhamnose biosynthesis has been demonstrated in several species through a genetics approach, i.e. mutation of any of the rml genes results in loss of L-rhamnose in the bacterial cell walls (Tsukioka et al.,; Rahim et al.; Carvalho et al.; van der Beek et al. ). For other species, loss of L-rhamnose in the cell wall was not confirmed by cell wall composition analysis upon mutation of rml genes (Chiang and Mekalanos; Xu et al. ). Importantly, disruption of dTDP-L-rhamnose biosynthesis severely attenuates bacterial fitness and/or virulence (Tsukioka et al.; Chiang and Mekalanos; Rahim et al.; Xu et al.; Carvalho et al.; van der Beek et al. ). The position or percentage of incorporated L-rhamnose likely dictates whether rml genes are essential. For example, in M. tuberculosis, L-rhamnose covalently links arabinogalactan to peptidoglycan, which is critical for the overall architecture of the Mycobacterial cell wall, making L-rhamnose biosynthesis essential (McNeil, Daffe and Brennan; Ma, Pan and McNeil ). Similarly in S. pyogenes, L-rhamnose is incorporated in the GAC that comprises half of the cell wall mass (McCarty ), rendering depletion lethal (Le Breton et al.; van der Beek et al. ). In the case of uropathogenic Escherichia coli and P. aeruginosa, mutation of rmlD results in loss of O-antigen expression but leaves the lipopolysaccharide core and lipid A structure intact, yielding viable bacteria (Burns and Hull; Rahim et al. ). However, loss of rhamnose does come at a certain cost, since rhamnose-deficient E. coli are extremely sensitized to serum-mediated killing (Burns and Hull ).

15Analysis of its genomic location in members of the family Streptococcaceae and in E. faecalis reveal that rmlD is systematically associated with a large cluster of genes encoding glycosyltransferases, polysaccharide transport systems, sugar biosynthesis enzymes and genes of unknown functions (Fig. 5). Experimental evidence in selected streptococcal species, L. lactis and E. faecalis confirms that rmlD-associated loci participate in the biosynthesis of RhaCWP as will be discussed in the next section.

16Biosynthesis of glycosylated surface structures in bacteria displays some common themes, despite the considerable diversity in the chemical composition of glycosylated structures (reviewed in Tytgat and Lebeer ). In general, glycoconjugate biosynthesis proceeds as follows (Fig. 6); (1) initiation of biosynthesis through activation of a lipid carrier, often undecaprenylphosphate, on the cytoplasmic side of the membrane, (2) elongation of the polysaccharide (building block) on the lipid carrier by sequential addition of activated sugar precursors, (3) translocation of lipid-linked precursors, either repeating units or the complete glycoconjugate, across the membrane by ABC transporters or ‘flippases’ (Lazarevic and Karamata ), (4) linkage of the glycoconjugate to peptidoglycan or protein (Kawai et al.; Chan et al. ) and (5) additional modifications to the glycoconjugate that can occur after cell wall anchoring. In some cases, the mature glycoconjugate is formed by polymerization of translocated repeating units between steps (3) and (4) by a dedicated polymerase. Since undecaprenylphosphate serves as a common scaffold to build structurally diverse glycoconjugates including capsules, peptidoglycan, lipopolysaccharides and protein-modifying glycans, the availability of ‘free’ undecaprenylphosphate is essential for bacterial survival (Hartley and Imperiali ).

17In accordance with the general steps of glycoconjugate biosynthesis (Fig. 6), biosynthesis of RhaCWP is likely initiated on the inside of the cytoplasmic membrane on an undecaprenylphosphate lipid carrier, before the structure is transported across the membrane and attached to peptidoglycan. Indeed, most enzymes involved in RhaCWP biosynthesis are predicted to be intracellular or incorporated in the membrane. Furthermore, lysed protoplasts of S. pyogenes incorporate 14C-labeled dTDP-rhamnose into a polysaccharide structure, presumably the GAC (Zeleznick et al. ). The proteins required for elongation and translocation of RhaCWP appear to be encoded by rmlD-associated gene clusters (Fig. 5). Within each cluster, we find genes encoding glycosyltransferases, putative transport systems (either an ABC transport system or putative flippase), sugar modifying enzymes for the production of specific sugar precursors, and hypothetical proteins which can be unique for the species (Sutcliffe, Black and Harrington ). Enzymes for initiation (Fig. 6; step 1) and peptidoglycan anchoring after transport (Fig. 6; step 4) are encoded elsewhere on the genome in most species. It is beyond the scope of this review to discuss the putative role of each gene in detail; instead, we aim to provide genetic and biochemical insights concerning the proposed steps of the RhaCWP biosynthesis pathway.

18The available structures of RhaCWP as well as evidence for their covalent linkage to peptidoglycan MurNAc (Heymann, Manniello and Barkulis; Deng et al. ) implies that activation of the lipid carrier undecaprenylphosphate by the transfer of GlcNAc by a UDP-GlcNAc:lipid phosphate transferase is the first enzymatic step of RhaCWP biosynthesis. This step is similar to the initiating reaction for WTA biosynthesis in B. subtilis and S. aureus, which is catalyzed by integral membrane proteins called TarO/TagO (Swoboda et al. ). A gene encoding a TarO/TagO homologue is not present in the rmlD-associated gene clusters in streptococci and L. lactis (Fig. 5), but can be readily identified elsewhere in the genome through homology searches. Atypically in E. faecalis, the first gene of the putative RhaCWP gene cluster, epaA, may encode the required UDP-GlcNAc:lipid phosphate transferase (Fig. 5). Experimental proof for a role of these enzymes in the biosynthesis of RhaCWP has been provided for S. pyogenes GacO, S. agalactiae GbcO, S. mutans RgpG and E. faecalis EpaA through genetic mutation, bacterial complementation studies or pharmacological inhibition of the transferase by the compound tunicamycin (Yamashita et al.; Teng et al.; Campbell et al.; Caliot et al.; van Sorge et al. ). Indeed, interference with expression or enzymatic function of these transferases depleted the cell wall of L-rhamnose, attenuated bacterial growth and induced aberrant morphology and cell division resulting in an increased chain length (Yamashita et al.; Teng et al.; Caliot et al.; van Sorge et al. ). Correspondingly, the S. agalactiae gbcO mutant lost expression of the pellicle (Fig. 2) (Caliot et al. ), confirming previous observations in L. lactis that this outer layer correlates with the presence of RhaCWP (Fig. 2) (Chapot-Chartier et al. ). For streptococci, this phenotype mimics the defects observed in rml mutants, which also lose expression of RhaCWP and display attenuated growth, aberrant morphology and growth in long chains (Fig. 1) (Tsukioka et al.; van der Beek et al. ).

19Following lipid carrier activation by UDP-GlcNAc:lipid phosphate transferases, biosynthesis of the actual RhaCWP structure occurs through the step-wise addition of monosaccharides by specific glycosyltransferases. In most cases, each glycosidic bond requires a dedicated enzyme. Such knowledge informs detailed structure–function studies of RhaCWP as will be discussed below.

20The Lancefield GAC, GCC and S. mutans RhaCWP contain a backbone composed of α-1,2-/α-1,3-linked polyrhamnose (Fig. 3A). The first seven genes of the respective RhaCWP gene clusters contain a high sequence identity, suggesting that these genes are required to construct the identical rhamnan backbone (Fig. 5). Indeed, heterologous expression of these seven genes of S. mutans in E. coli results in the production of α-1,2-/α-1,3-linked polyrhamnose (Shibata et al. ). Subsequently, Shibata et al. identified that three glycosyltransferases encoded in this partial cluster, RgpA, RgpB and RgpF, are required for rhamnan biosynthesis, with RgpA adding the first rhamnose to the undecaprenylphosphate-GlcNAc lipid carrier (Shibata et al. ). Homologous glycosyltransferases in S. pyogenes and S. zooepidemicus likely catalyze a similar reaction. Decoration of the rhamnan backbone with specific side chains produces the discriminating epitopes of the GAC, GCC and RhaCWP of S. mutans (Fig. 3A). The remaining glycosyltransferases in the respective gene clusters are likely implicated in these reactions. For S. mutans, RgpE and RgpI add α1,2-linked glucose to RhaCWP (Yamashita et al.; Ozaki et al. ), whereas GacI in S. pyogenes likely adds the characteristic β-linked GlcNAc side chain since mutation of gacI results in loss of side chain expression (van Sorge et al. ). For construction of the GCC side chain, a disaccharide GalNAc (Fig. 3A), glycosyltransferases encoded by Sez_0844 and Sez_0848 likely play a crucial role. Similar to the GAC, GCC and RhaCWP of S. mutans, the Lancefield GGC is also reported to be a linear structure (Pritchard et al. ), albeit with a different disaccharide backbone (Fig. 3A). No studies have experimentally addressed the contribution of specific glycosyltransferases to the biosynthesis of the GGC, but the increased number of glycosidic bonds compared to the GAC, GCC and S. mutans RhaCWP, suggest that the activity of all seven glycosyltransferases encoded in the gene cluster (Fig. 5) are required for its biosynthesis.

21The situation is more complex in the case of the pellicle-forming GBC and RhaCWP of L. lactis strains. The GBC was previously postulated to be composed of four modular subunits (Sutcliffe ), which are ultimately linked together via phosphodiester bonds to form a multiantenna branching structure (Fig. 3B). Despite its structural complexity, it was hypothesized that a minimum of 11 glycosyltransferases would theoretically be required to compose all glycosidic linkages in the GBC, seven of which are encoded within the GBC gene cluster (Sutcliffe, Black and Harrington ). However, upon reexamination of the GBC structure (Michon et al.,, ), we have recognized the presence of two key structural elements, both beginning with GlcNAc: a stem oligosaccharide and a repeating unit (Fig. 3A, B). Following transfer of GlcNAc to the undecaprenylphosphate lipid carrier by the GbcO UDP-GlcNAc:lipid phosphate transferase, cytoplasmic synthesis of both of these subunits could be achieved by the glycosyltransferases encoded in the GBC gene cluster (unpublished observations). Identification of this repeat unit allows for a revised model of GBC assembly (see below) and solves the mystery of the two ‘missing’ GlcNAc transferases (Sutcliffe, Black and Harrington ).

22In contrast to the invariable structure of the GAC, GBC, GCC and GGC in the respective streptococci, L. lactis strains can express structurally diverse CWP structures, correlating to diversity in the responsible gene cluster (Chapot-Chartier et al.; Mahony et al.; Ainsworth et al.; Farenc et al. ). Rhamnose is not incorporated into the CWP of every strain (Ainsworth et al. ). Bioinformatics predict the presence of eight glycosyltransferases in the L. lactis pellicle gene cluster but for none of them a contribution or role has been experimentally addressed. Currently, the structure of the E. faecalis Epa awaits elucidation, making it difficult to speculate about the specific glycosyltransferases involved. However, genetic disruption of epaB and epaN, which encode putative glycosyltransferases (Fig. 5), changes immunoreactivity of Epa, indicating that these enzymes are involved in Epa biosynthesis (Xu, Murray and Weinstock; Teng et al., ). Moreover, the Epa polysaccharide extracted from the ΔepaB mutant was incapable of incorporating rhamnose and instead included mannose, suggesting that epaB encodes a rhamnosyltransferase (Teng et al. ).

23Transport of big sugar complexes across cell membranes is energetically unfavorable and requires one of three mechanisms, i.e. synthase-dependent transporters, ABC transporters, or so-called ‘flippases’ (Cuthbertson, Kos and Whitfield ). The synthase-dependent pathway is least well defined and is involved in the formation of polysaccharides such as chitin, cellulose, hyaluronan and poly-N-GlcNAc. It was recently confirmed that a single protein, the synthase, executes both polymerization and export of the growing polysaccharide chain (Morgan, Strumillo and Zimmer ). ABC transporters translocate longer glycan chains and can be composed of a single protein (for example PglK; Perez et al. ) or of a two-protein complex consisting of a permease protein and an ATP-binding protein (Cuthbertson, Kos and Whitfield ). Finally, classical Wzx flippase systems generally transport oligosaccharide repeat units that are polymerized on the extracytoplasmic face of the membrane by a Wzy-like polymerase to complete the mature polysaccharide (Islam and Lam ). Interestingly, genes consistent with either the Wzx flippase and/or the ABC transport system seem to be present in all RhaCWP gene clusters (Fig. 5). Currently, the only experimental evidence supporting a role for the cognate ABC transport system is for S. mutans (Shibata et al. ). After heterologous expression of the first seven (rmlD-rgpABCDEF) genes in E. coli and subsequent disruption of the ABC transporter-encoding genes (rgpCD), the bacteria were unable to produce rhamnan (Shibata et al. ). This transport mechanism would resemble WTA translocation, which involves the TagG/H ABC transport system (Lazarevic and Karamata ).

24The translocation of the GBC structure remains particularly enigmatic, given that assembly of a fully branched polymer before translocation would present a major challenge (Sutcliffe, Black and Harrington ). In contrast to the other RhaCWP gene clusters, the GBC gene cluster only contains a putative Wzx flippase gene (gbs1482) as well as two integral membrane proteins (encoded by gbs1483 and gbs1490) that may act as accessory proteins (Sutcliffe, Black and Harrington ). Recognition that the GBC is composed of two structural elements (stem oligosaccharide and a repeat unit, Fig. 3B) instead of the previously predicted four elements, now allows for a revised route of synthesis. Flipping of these two building blocks (by GBS1482 and, presumptively, a second flippase) could allow assembly of the mature GBC on the extracytoplasmic face of the membrane. The integral membrane proteins (GBS1484, GBS1489 or GBS1490) can be proposed as the second flippase and/or assembly proteins. Assembly of repeating units requires formation of linkages between the GlcNAc and C3 of a terminal rhamnose in the growing chain, whilst the branch points are formed by repeating unit linkage to the C4 of the penultimate rhamnose. An attractive aspect of this revised proposed biosynthesis is that it removes the need for as yet unidentified cytoplasmic GlcNAc transferases. In addition, it would allow flipping of smaller units instead of the fully branched GBC, which is likely energetically more favorable. Thus, complete synthesis, translocation and assembly of the GBC can be predicted to occur through the action of GbcO and the proteins encoded in GBC gene cluster. Experimental evidence to support these hypotheses should now be sought.

25Detailed biochemical analysis has demonstrated that the GAC and GBC are attached to the MurNAc moiety of peptidoglycan, similar to WTA in other Gram-positive bacteria (Swoboda et al. ). GAC is presumably connected to MurNAc through a phosphate containing bridge composed of one or more units of glycerol (Heymann, Manniello and Barkulis ), which would concur with the linkage of WTA to peptidoglycan in S. aureus (Swoboda et al. ). The transfer of nascent RhaCWP from the flipped lipid carrier onto the peptidoglycan is likely catalyzed by a member of the LytR-CpsA-Psr family. These proteins were initially identified to catalyze the attachment of WTA to peptidoglycan in B. subtilis (Kawai et al. ). This family of proteins is widespread throughout Gram-positive bacteria, with at least two family members present in the genomes of streptococci up to five in E. faecalis (Hubscher et al. ). The LytR-CpsA-Psr proteins appear to be highly redundant since lack of peptidoglycan-attached glycopolymers, such as capsule and WTA, only becomes apparent upon genetic mutation of all encoded enzymes (Kawai et al.; Eberhardt et al.; Chan et al., ). The LytR-CpsA-Psr phosphotransferases typically hydrolyze the phosphodiester linkage between the lipid-carrier and the first GlcNAc at the stem base of the polysaccharide (i.e. they hydrolyze the linkage created by the UDP-GlcNAc:lipid phosphate transferase such as GacO and GbcO) and attach the polymers to peptidoglycan via a phosphate ester linkage (Kawai et al.; Eberhardt et al.; Chan et al. ). Overall, it seems likely that LytR-CpsA-Psr proteins are involved in anchoring RhaCWP to the cell wall, but experimental proof is currently lacking.

26As mentioned above, the Lancefield typing scheme is unable to discriminate bacteria up to the species level. The availability of genome sequences, as well as knowledge regarding RhaCWP gene clusters, provides an opportunity to gain insight into the distribution and potential structure of RhaCWP among Gram-positive bacteria. For example, S. castoreus was noted to react with Group A-specific antisera and indeed its draft genome contains a biosynthetic locus syntenous with that of the GAC of S. pyogenes (Table S1, Supporting Information). However, the presence of an additional glycosyltransferase compared to the GAC gene cluster in S. pyogenes suggests some fine structural variation. Likewise, the GBC has been reported to be expressed by different streptococcal species most notably Streptococcus porcinus, Streptococcus pseudoporcinus, Streptococcus troglodytidis and Streptococcus plurextorum. Correspondingly, the genomes of S. pseudoporcinus and S. porcinus contain a fully syntenous GBC biosynthetic gene cluster except for the lack of a gbs1485 orthologue (unpublished observations). Presumably, the expressed structures only lack one of the monosaccharide rhamnose side-branches present in the GBC repeating unit (Fig. 3B) (Sutcliffe, Black and Harrington ), but can still make the linear trirhamnosyl immunodominant epitope with a terminal rhamnose that is detected by Group B-specific serotyping (Curtis and Krause ). In contrast, Streptococcus thoraltensis contains a GBC-variant gene cluster that lacks orthologues of two of the predicted rhamnosyltransferases, gbs1481 and gbs1485, present in S. agalactiae (Table S2, Supporting Information) and this species is non-groupable by Lancefield serotyping assays. Instead of gbs1481 and gbs1485, the S. thoraltensis locus harbors two additional glycosyltransferases absent in the otherwise syntenous locus of S. agalactiae (Table S2, Supporting Information). Absence of a gbs1481 orthologue is likely responsible for abrogated cross-reactivity in the Group B antigen serotyping tests due to the loss of the terminal rhamnose from the dominant trirhamnosyl epitope. This analysis therefore suggests that S. thoraltensis is capable of synthesizing a RhaCWP that is a structural variant of the GBC. Increased availability of genome sequences for streptococci and related species will help identify additional rmlD-linked gene clusters for RhaCWP biosynthesis that are consistent with either known Lancefield serotyping reactions (as exemplified here by S. castoreus) or from which variant or novel RhaCWP structures can be predicted.

27Since their identification and structural characterization from the 1930s onwards, the biological roles of the Lancefield Group antigens or other RhaCWP have received little attention. It was long thought that Group-specific antigens were only of structural importance (McCarty ). Indeed, complete loss of RhaCWP expression through genetic mutation results in severe growth and cell division abnormalities (Fig. 1) and can be essential under competing conditions (McCarty; Tsukioka et al.; Caliot et al.; van der Beek et al. ). Caliot et al. targeted the UDP-GlcNAc:lipid phosphate transferase gbcO (gbs0136) of S. agalactiae to initiate functional studies on the GBC. The resulting GBC-negative S. agalactiae strain was devoid of cell wall rhamnose and phosphate and lost expression of the pellicle structure (Caliot et al. ) (Fig. 2). The loss of GBC was associated with major morphological and cell growth defects, including mislocated septa and defects in cell division and separation, which resulted in the formation of very long chains (Caliot et al. ). This phenotype corresponds to previous descriptions of a stable opacity variant of S. agalactiae that had lost GBC expression and displayed growth and morphological defects (Pincus et al., ). For this spontaneous opaque S. agalactiae mutant strain the underlying genetic defect has never been clarified. For the gbcO mutant strain, the defects result from reduced levels of highly cross-linked peptidoglycan and mislocalization of the important peptidoglycan hydrolase PcsB (Caliot et al. ), a protein required for streptococci cell wall separation (Reinscheid et al.; Sham et al. ). The latter observation is reminiscent of studies in S. aureus, where the preferential localization of the major autolysin Atl is lost in absence of WTA (Schlag et al. ). Collectively, these observations support a role of GBC in cell wall homeostasis of S. agalactiae. In line with observations in the S. agalactiae gbcO mutant (Caliot et al. ), pharmacological inhibition of the S. pyogenes enzyme GacO (encoded by M5005_Spy0240) by tunicamycin resulted in depletion of GAC from the cell wall, increased mutanolysin susceptibility and increased chain length as a result of cell separation defects (van Sorge et al. ). However, interpretations from these studies may be obscured by the interconnection between biosynthesis of Group-specific antigens and that of other cell wall glycopolymers, including peptidoglycan, since most glycoconjugates use undecaprenylphosphate as a carrier for biosynthesis. Coordinated regulation between biosynthesis of different glycoconjugates is further supported by the observation that the S. agalactiae gbcO mutant increases capsule production suggesting a coordinated regulation between the two glycoconjugates (Beaussart et al. ). Knowledge regarding the interrelatedness of glycoconjugate biosynthesis pathways is relevant for the design of new antibiotics since interference of such connected pathways may have synergistic effects as recently demonstrated for WTA and peptidoglycan biosynthesis (Sewell and Brown ).

28In addition to their significant role in cell wall architecture, it is appreciated that RhaCWP are important phage receptors for many species. This again highlights a parallel with WTA in other Gram-positive bacteria where WTA is critical to phage-mediated horizontal gene transfer (Baptista, Santos and Sao-Jose; Brown et al.; Winstel et al. ). For several streptococci, the specificity of phage adsorption correlates to the side chain anchored to the rhamnan backbone. Indeed, the (GalNAc)2 side chain of the GCC serves as an attachment site for Group C1 bacteriophage in Group C Streptococcus (Krause; Fischetti and Zabriskie ). Correspondingly, a Group C-variant strain, which completely lacks the immunodominant (GalNAc)2 epitope, was isolated from Group C Streptococcus that survived exposure to Group C1 lytic phages (Krause ). Also for S. pyogenes, the GAC-specific GlcNAc terminal moiety appears to be involved in both lytic and temperate phage adsorption although additional unidentified cell wall factors are also involved (Fischetti and Zabriskie ). Finally, specific phages recognize the α-1,2-linked glucose side chain of serotype c S. mutans strains (Shibata, Yamashita and van der Ploeg ). For L. lactis, selection of phage resistant strains from a random insertional mutagenesis library originally identified the presence and genetic locus of the RhaCWP (Dupont et al. ). Moreover, the precise structure of the RhaCWP dictates bacteriophage sensitivity (Dupont et al.; Mahony et al.; Ainsworth et al. ). Finally, structural changes in E. faecalis Epa through genetic manipulation greatly affect phage sensitivity despite similar adsorption levels (Teng et al. ). Phage dynamics within the bacterial population will impact fitness and may also be important for horizontal gene transfer, even across long phylogenetic distances (Winstel et al. ), affecting virulence or antibiotic resistance of pathogens. Thus the interactions of phages with RhaCWP are likely to impact on bacterial population structure. In the case of L. lactis, knowledge regarding the molecular mechanisms of phage adsorption and infection may benefit the food industry.

29The localization of RhaCWP at the host–pathogen interface suggests that their biological function might be broader than a role in cell wall biogenesis. Similarly for WTA, evidence is accumulating for its role in virulence by increasing adherence and immune evasion (Carvalho et al.; Winstel et al. ). Recent studies in S. pyogenes and E. faecalis now highlight that subtle modifications to the RhaCWP structure, which do not impact bacteria growth, can significantly impact virulence (Xu et al.; Teng et al.; van Sorge et al. ). For E. faecalis, disruption of epaB, epaE, epaM and epaN, which may completely eliminate Epa expression or only modify its structure, all caused significant attenuation in a mouse peritonitis model (Xu et al.; Teng et al. ). Reduced virulence correlated with increased phagocytic uptake and clearance by neutrophils (Teng et al. ). Similarly in L. lactis, loss of the pellicle results in 10-fold more efficient uptake by macrophage cell lines compared to wild-type bacteria (Chapot-Chartier et al. ). These results therefore indicate that the pellicle can exert an anti-phagocytic effect both for L. lactis and E. faecalis. For S. pyogenes, structure–function studies focused on the role of the GAC GlcNAc side chain, which was selectively removed through genetic mutation of the glycosyltransferase-encoding gene gacI (van Sorge et al. ). The gacI mutant bacteria still expressed the rhamnan backbone but did not display apparent cell wall abnormalities. However, bacteria were increasingly susceptible to innate immune clearance by neutrophils and antimicrobial components (van Sorge et al. ). Moreover, virulence of this genetically engineered Group A-variant strain was significantly attenuated in two animal models (van Sorge et al. ). Again, this indicates that specific epitopes of RhaCWP increase bacterial immune resistance, although the mechanism has not been well defined. In addition to increased immune resistance, RhaCWP may modulate host immune responses by targeting specific lectin receptors (carbohydrate-recognizing pattern-recognition receptors) (Sancho and Reis e Sousa ). Human lectins regulate fundamental immunological processes but also directly engage microbial carbohydrates, linking pathogen recognition to appropriate immune responses. Although this often promotes bacterial clearance, lectin targeting can also promote bacterial survival by skewing immune reponses (van Vliet et al. ). Research in this area should provide insight whether the virulence-promoting effect of RhaCWP occurs through interaction with lectin receptors. Alternatively, decoration of the rhamnan backbone or rhamnose moieties by common sugars such as GlcNAc, glucose and GalNAc may be a strategy for microorganisms to avoid immune recognition. This is relevant since the absence of rhamnose in humans makes rhamnose an attractive pattern-associated molecular pattern (PAMP). Indeed, in fish and invertebrates, rhamnose is targeted by the innate immune system through germ-line encoded pattern-recognition receptors called rhamnose-binding lectins (Ogawa et al.; Ng et al. ). Interestingly, rhamnose-binding lectins agglutinate both Gram-positive and Gram-negative bacteria through interaction with glycan structures such as LPS or lipoteichoic acid in the bacterial cell wall (Tateno et al.; Cammarata et al.; Ng et al. ). Rhamnose-binding lectins are also involved in inflammatory responses through the induction of cytokines (Watanabe et al. ). Humans lectins with specificity for rhamnose have not been identified yet, but their existence might be anticipated given the estimated presence of over 150 glycan-binding proteins in humans, many with uncharacterized ligand specificity (Zelensky and Gready; Drickamer ). Overall, the role of RhaCWP in pathogenesis and cell physiology is just starting to be appreciated. Further investigations are likely to unravel new bacterial immune evasion strategies but may also contribute to new insights into innate immune responses and recognition.

30Rhamnose is not just incorporated in RhaCWP, but is also present in capsular polysaccharides. We consider this structure distinct from RhaCWP given their localization in the cell wall; RhaCWP are interpolated within the peptidoglycan wall layer, whereas capsular polysaccharides are typically the outermost layers of the cell envelope. Nevertheless, it is worth noting that several clinically relevant Gram-positive bacteria synthesize rhamnose-containing capsules. The significance of the S. pneumoniae capsule as a key virulence factor has been established since the 1928 landmark ‘Griffiths experiment’ and the capsular polysaccharides are major protective antigens utilized in current vaccine formulations (Geno et al. ). Of the 41 (out of 46) serogroups of pneumococcal capsule for which carbohydrate composition or structural information is available, 17 of these (∼40%) contain rhamnose, including those of clinically significant serogroups such as 6, 19 and 23. The capsular biosynthetic loci for these serogroups all contain rmlABCD genes (Bentley et al. ). The presence of these genes in the biosynthetic loci for which capsular polysaccharide structures are not yet available or may be incomplete (e.g. serogroups 21, 40 and 48) suggests additional structures also contain rhamnose. Thus, rhamnose biology is likely of significance in much of the S. pneumoniae population and at least nine of the serotype antigens included in the current 23-valent vaccin are rhamnose-containing polysaccharides (Geno et al. ).

31In addition to S. pneumoniae, S. agalactiae strains belonging to serotype VIII contain rhamnose within the polysaccharide repeating unit (Kogan et al.; Cieslewicz et al. ). This serotype remains relatively rare globally but has been reported to be of significance in some population groups, notably in Japan and the Pacific (Lachenauer et al.; Edmond et al. ).

32Increased knowledge regarding the biosynthesis and function of RhaCWP could aid the development of new antimicrobial agents but may also have applications in metabolic engineering to optimize food production (Chapot-Chartier ) or glycoconjugate production for medical purposes (Jaffe et al. ). Clearly, the L-rhamnose biosynthesis pathway holds promise for antimicrobial drug targeting, given the loss of virulence and/or viability upon rhamnose depletion in a wide range of Gram-positive and Gram-negative bacteria. More importantly, lack of L-rhamnose in humans should preclude off-target effects lowering risks of unwanted side effects. Thus far, several inhibitors screens for Rml enzymes have been initiated, resulting in only one RmlA inhibitor targeting P. aeruginosa with some marginal activity against M. tuberculosis (Alphey et al. ). In addition to targeting rhamnose, inhibition of other steps in the RhaCWP biosynthesis pathway could also be of interest. Although this may not immediately kill the bacterium, it may act as anti-virulence agent, increasing susceptibility to host defense mechanisms such as phagocytic clearance (Nizet ). A similar strategy is currently exploited for S. aureus, where inhibition of the UDP-GlcNAc:lipid phosphate transferase TarO is not detrimental to the bacterium, but render the bacterium avirulent (Sewell and Brown ). An additional effect of TarO inhibition is the re-sensitization of resistant bacteria to β-lactam antibiotics due to an interaction between the WTA and peptidoglycan biosynthesis pathways in S. aureus (Campbell et al. ). Similar synergy may occur between conventional antibiotics and inhibitors of the RhaCWP pathway.

33RhaCWP are also attractive vaccine candidates due to their conserved and constant expression in species of medical importance, such as S. pyogenes and S. agalactiae. Indeed, different strategies are currently explored to develop protective vaccines against these streptococcal species (Dale et al.; Steer, Dale and Carapetis; Nuccitelli, Rinaudo and Maione ). For both pathogens, much research has focused on type-specific vaccine strategies, i.e. a capsule-conjugate vaccine for S. agalactiae (Johri et al.; Nuccitelli, Rinaudo and Maione ) and a multivalent M-protein vaccine for S. pyogenes (Dale et al.; Steer, Dale and Carapetis ). A more elegant and globally effective approach would employ a vaccine antigen that is universally expressed on all strains in an invariant manner. Indeed, the GBC is immunogenic in rabbits but antibodies raised are not protective in a newborn mouse model (Marques et al. ), likely due to shielding of the GBC by the polysaccharide capsule. More encouraging are the results with regard to the GAC as a universal S. pyogenes vaccine antigen. Conjugate vaccines of either isolated or synthetic GAC protect mice from subsequent infection after active and passive immunization (Sabharwal et al.; Kabanova et al. ). However, there is controversy with regards to safety in the use of the native GAC for vaccine purposes, since several groups have indicated a role for anti-GlcNAc antibodies in the pathogenesis of rheumatic fever (Goldstein et al.; Ayoub and Dudding; Kirvan et al. ). Recent elucidation of the molecular pathway for GlcNAc side chain formation allows for the development of an alternate vaccine antigen consisting of the polyrhamnose backbone of the GAC (van Sorge et al. ). Anti-rhamnan antibodies raised against the GlcNAc-deficient GAC enhanced phagocytic killing of multiple M-serotypes in vitro and protected mice from lethal challenge with wild-type S. pyogenes through passive immunization (van Sorge et al. ). Of interest is the observation that some streptococci also release RhaCWP into their surroundings. Although it is unclear whether this involves an active mechanism of release, it possibly has implications for the efficiency of vaccine strategies. Overall, further exploration on the application of RhaCWP for vaccination purposes is warranted.

34Finally, dissection of the RhaCWP biosynthesis pathway could benefit food production, most notably the dairy industry that uses lactic acid bacteria for food fermentations (Chapot-Chartier ). Phage infection of these fermenting cultures results in product variations but can also lead to huge economic losses (Samson and Moineau ). The selection or engineering of phage-resistant strains is therefore of considerable interest. The role of RhaCWP in phage–host interaction has opened up new possibilities for the development of bacteriophage insensitive mutants for food production purposes. In addition to food preparation, polysaccharides often have cosmetic, pharmaceutical and biomedical applications. Elucidation of specific transferase activities could be used towards metabolic engineering of new materials or compounds with interesting biological or physical properties. In particular, the incorporation of rare sugars such as rhamnose and uronic acids is a rather unexplored area but may be of interest to different areas including the biomedical field (Roca et al. ).

35Despite their long history in streptococcal diagnostics, investigations on the biological roles and possible applications of RhaCWP have lagged behind. Genome sequencing has initiated genetic studies to elucidate structure–function relationships of RhaCWP, highlighting their critical importance in proper cell wall architecture and pathogenesis. Their indispensable nature identifies the RhaCWP biosynthesis pathway as an attractive therapeutic target for antimicrobial drug development. Spin offs will likely find applications in the area of metabolic engineering for food production and biomedical purposes.