Impact of 4 Lactobacillus plantarum capsular polysaccharide clusters on surface glycan composition and host cell signaling

Background Bacterial cell surface-associated polysaccharides are involved in the interactions of bacteria with their environment and play an important role in the communication between pathogenic bacteria and their host organisms. Cell surface polysaccharides of probiotic species are far less well described. Therefore, improved knowledge on these molecules is potentially of great importance to understand the strain-specific and proposed beneficial modes of probiotic action. Results The Lactobacillus plantarum WCFS1 genome encodes 4 clusters of genes that are associated with surface polysaccharide production. Two of these clusters appear to encode all functions required for capsular polysaccharide formation (cps2A-J and cps4A-J), while the remaining clusters are predicted to lack genes encoding chain-length control functions and a priming glycosyl-transferase (cps1A-I and cps3A-J). We constructed L. plantarum WCFS1 gene deletion mutants that lack individual (Δcps1A-I, Δcps2A-J, Δcps3A-J and Δcps4A-J) or combinations of cps clusters (Δcps1A-3J and Δcps1A-3I, Δcps4A-J) and assessed the genome wide impact of these mutations by transcriptome analysis. The cps cluster deletions influenced the expression of variable gene sets in the individual cps cluster mutants, but also considerable numbers of up- and down-regulated genes were shared between mutants in cps cluster 1 and 2, as well as between mutant in cps clusters 3 and 4. Additionally, the composition of overall cell surface polysaccharide fractions was altered in each mutant strain, implying that despite the apparent incompleteness of cps1A-I and cps3A-J, all clusters are active and functional in L. plantarum. The Δcps1A-I strain produced surface polysaccharides in equal amounts as compared to the wild-type strain, while the polysaccharides were characterized by a reduced molar mass and the lack of rhamnose. The mutants that lacked functional copies of cps2A-J, cps3A-J or cps4A-J produced decreased levels of surface polysaccharides, whereas the molar mass and the composition of polysaccharides was not affected by these cluster mutations. In the quadruple mutant, the amount of surface polysaccharides was strongly reduced. The impact of the cps cluster mutations on toll-like receptor (TLR)-mediated human nuclear factor (NF)-κB activation in host cells was evaluated using a TLR2 reporter cell line. In comparison to a L. plantarum wild-type derivative, TLR2 activation remained unaffected by the Δcps1A-I and Δcps3A-J mutants but appeared slightly increased after stimulation with the Δcps2A-J and Δcps4A-J mutants, while the Δcps1A-3J and Δcps1A-3J, Δcps4A-J mutants elicited the strongest responses and clearly displayed enhanced TLR2 signaling. Conclusions Our study reveals that modulation of surface glycan characteristics in L. plantarum highlights the role of these molecules in shielding of cell envelope embedded host receptor ligands. Although the apparently complete cps clusters (cps2A-J and cps4A-J) contributed individually to this shielding, the removal of all cps clusters led to the strongest signaling enhancement. Our findings provide new insights into cell surface glycan biosynthesis in L. plantarum, which bears relevance in the context of host-cell signaling by probiotic bacteria.


Background
The Gram-positive bacterial cell envelope is a multilayered structure, which is mainly composed of peptidoglycan with embedded teichoic acids, proteins, and polysaccharides and which is essential to maintain cellular integrity and shape [1]. The molecules of the cell envelope collectively decorate the bacterial surface in a strain-and species-specific manner and facilitate important bacterial processes such as stress-and environmental-adaptation, surface colonization, and adhesion [2].
Lactic acid bacteria (LAB) are Gram-positive bacteria that can be encountered in a wide range of environmental niches, including a range of dairy and other food-raw material fermentations as well as the gastrointestinal (GI) tract of humans and animals. LAB have a clear industrial relevance, based on their role in food fermentation whereby they contribute to the preservation of raw-materials in foods, but also to the product's flavor and texture [3]. Additionally, specific LAB are marketed as health-promoting organisms or probiotics [4]. Most LAB are able to synthesize extracellular polysaccharides, many of which consist of heteropolysaccharides built up from regular repeating oligosaccharide units that commonly contain the monosaccharides galactose, glucose, and in several cases also encompass rhamnose, N-acetylglucosamine, N-acetyl-galactosamine, mannose, and non-carbohydrate substitutions [5][6][7][8]. The repeating units are synthesized in the cytoplasm and assembled on the lipid carrier undecaprenyl phosphate by sequential transfer of monosaccharides from nucleotide sugars by specific glycosyltransferases. Membrane-associated and -assembled repeating-unit oligosaccharides are thought to be translocated across the cytoplasmic membrane and polymerized to polysaccharides by a dedicated transport and polymerization machinery [9,10]. The overall capacity for this type of polysaccharide synthesis is encoded by clusters of genes that are readily detectable by their conserved structural composition (Wzy-dependent polymer gene cluster) and are commonly transcribed as a single transcript [11], although exceptions have been reported [12]. These clusters are found in many Gram-positive bacteria and are probably best documented for Streptococcus pneumonia [13,14]. The first genes in these clusters (wzd, wze, wzh) are involved in modulation of capsule synthesis and form a tyrosine phosphoregulatory circuit that controls polymer production [14]. These regulatory gene-cassettes are commonly followed by a variable number of genes encoding glycosyltransferases, which are involved in repeat unit synthesis on the cytoplasmic face of the cell membrane [14]. Once synthesized, the repeat unit is flipped across the membrane by a dedicated flippase (Wzx) and subsequently polymerized (Wzy) to form the extracellular polysaccharide [14].
EPS produced by LAB has received a lot of attention based on their role as thickening agents that are naturally produced during fermentation and influence the textural properties of fermented dairy and non-dairy products [9,[15][16][17][18]. Besides the industrial relevance of polysaccharides produced by LAB in product characteristics, they may also play a role in the interaction between microbes and the host intestinal mucosa. While in pathogenic bacteria such as Streptococcus pneumoniae, polysaccharide capsules are extensively studied and were shown to play important roles in virulence by inhibiting opsonization and phagocytosis and are employed frequently as important serotyping antigens for epidemiological studies [19,20], the role of cell surface polysaccharides of commensal or probiotic bacteria is far less understood. In Lactobacillus rhamnosus GG, biosynthesis of the high-molecular-weight, galactose-rich surface polysaccharide molecules negatively affects the bacterial capacity to bind to intestinal epithelial cells, which may be due to the shielding of adhesins on the bacterial cell surface [21]. Additionally, surface polysaccharides may also contribute to protection against intestinal innate immune factors such as the antimicrobial peptide LL-37 [22]. A direct role in host signaling has been proposed for purified surface polysaccharides of Lactobacillus casei Shirota that were shown to mediate the suppression of pro-inflammatory responses in macrophages [23]. Moreover, several polysaccharide biosynthesis-related genes of Lactobacillus plantarum were up-regulated in vivo in the GI-tract of mice and humans [24], and a recent study tentatively correlated elevated expression of cps genes to enhanced survival under GItract mimicking conditions in this species (van Bokhorst-van de Veen, unpublished observations).
Here, we characterize the 4 cps gene clusters encoded by L. plantarum WCFS1, an extensively studied model organism for probiotic lactobacilli, aiming to identify bacterial molecules involved in microbe-host interactions with relevance for probiotic functions [25]. We constructed various cps-cluster deficient strains, including single (Δcps1A-I, Δcps2A-J, Δcps3A-J, Δcps4A-J), triple (Δcps1A-3J), and quadruple (Δcps1A-3J, Δcps4A-J) deletion mutants. The consequences of cps cluster mutation were studied at the level of whole genome transcriptome profiles, and by determining the amount, molar mass and chemical composition of the surface polysaccharide fraction of the mutants in comparison to the wild-type strain. Finally, the impact of the cps cluster mutations on toll-like receptor (TLR)-dependent human nuclear factor (NF)-κB activation in host cells was evaluated using a TLR2 reporter cell line.

Results and discussion
CPS biosynthesis cluster organization in L. plantarum WCFS1 The L. plantarum WCFS1 genome contains two regions with surface-associated polysaccharide biosynthesis genes that are designated cps genes ( Figure 1A and B). One region of 49 kb contains three gene clusters separated by transposase genes and has been identified as a genomic life-style island with high variability between L. plantarum strains [26]. Of these three gene clusters, cps1 and cps2 are identified only in strain WCFS1 and not in the other completely sequenced strains of the same species, while gene cluster cps3 is conserved in ST-III and ATCC 14917 but not in JDM1 (Figure 1; Additional file 1). A second region of 14 kb comprises the WCFS1 cps4 gene cluster and is conserved in other L. plantarum strains (Additional file 1). Clusters 2 and 4 have a typical structure of a Wzy-dependent polymer gene cluster ( Figure 1; Additional file 1). The first three genes (cps2ABC and cps4ABC) are homologous to the typical components of the tyrosine kinase phosphoregulatory circuit involved in control of capsule synthesis [14]. The fourth gene in both gene clusters is predicted to encode an UDP-N-acetylglucosamine 4-epimerase catalyzing the interconversion between UDP-N-acetyl-D-glucosamine and UDP-N-acetyl-Dgalactosamine (cps2D, and cps4D). The fifth gene is predicted to encode the priming glycosyltransferase catalyzing the transfer of a sugar-1-phosphate from a UDP-sugar to the undecaprenyl-phosphate, the first step in the synthesis of the repeat unit (cps2E and cps4E). The cps clusters 2 and 4 contain three additional genes with high sequence similarity to glycosyltransferase genes (cps2FGJ and cps4FGI), indicating that the encoded polysaccharides would be made up of quatro-saccharide repeat units. The 3' regions of both clusters encode homologues of the typical flippase (cps2I and cps4J) and polymerase (cps2H and cps4H) functions required for capsule synthesis [14].
The organization of clusters 1 and 3 is different from that of clusters 2 and 4. Cluster 3 appears to lack clear homologues of the chain-length modulator genes wzd, wze, and wzh. However, it is tempting to speculate that cps3D, cps3E, and cps3H serve such function, as their gene products show low sequence similarities to several polysaccharide biosynthesis proteins of unknown functions and Cps3E shares distant homology (26% identity) with the N-terminus of Clavibacter michiganensis subsp. michiganensis NCPPB 382 Wzc tyrosine kinase (functional homologues wzd and wze; NCBI accession number YP_001221462). The cps3 wzy homologue is split in two by a frame-shift caused by a single nucleotide insertion immediately upstream of the cps3F stop codon. It is unclear if a functional Wzy protein can be composed of Cps3F and Cps3G. The cluster contains three predicted glycosyltransferase genes but no priming glycosyltransferase gene. The presence of an acetyltransferase gene indicates that acetylation of the repeat units might take place. A set of genes encoding two transcriptional regulators and a mannose-specific adhesin protein separates the cps3 gene cluster from a polysaccharide polymeraselike (lp_1231) and priming glycosyltransferase (lp_1233) gene ( Figure 1). These genes could complete the polysaccharide synthesis machinery of cps3, which would then be predicted to be involved in the synthesis of a polysaccharide made up of acetylated quatro-saccharide repeat units. Notably, in Lactobacillus rhamnosus, exopolysaccharide gene clusters the priming glycosyltransferase genes are also separated from the body of the polysaccharide gene cluster [12].
The glf gene of cps cluster 1 is predicted to encode an UDP-galactopyranose mutase catalyzing the interconversion of UDP-galactopyranose and UDP-galacto-1,4-furanose. The glf gene encodes a protein with 86% identity to the protein encoded by the cluster 3 glf gene. The cps1 cluster contains 5 predicted glycosyltransferase genes but like the cps3 cluster, it appears to lack a priming glycosyltransferase gene. In addition, it contains a predicted acetyltransferase gene, wzx, wzy, and wzd homologues but no wze and wzh homologues. Expression of the cps1 cluster is predicted to produce a polysaccharide comprising an acetylated hexa-saccharide repeat unit.
A functional role of the cps 1 and 3 clusters in production of surface glycans in L. plantarum WCFS1 is supported by the impact of individual deletion of either the cps1 or cps3 genecluster on the surface glycan characteristics (see below). Notably, the lack of a predicted priming glycosyltransferase and in both the cps3 and cps1 clusters could imply that the glycosyltransferases encoded by these clusters interact with the undecaprenylphosphate-sugars derived from other pathways (e.g., cps clusters 2 and 4, or more remotely encoded priming glycosyltransferases like the lp_1233encoded function). However, an additional or alternative role of the cps geneclusters 1 and 3 in glycosylation of other surface molecules like teichoic acids and/or proteins as has been hypothesized before [2] can't be excluded at this stage. . Analysis of the cps cluster-related gene expression patterns revealed that the cps1A-I cluster-associated genes were highly expressed under laboratory conditions, whereas the genes of cps clusters 2A-K, 3A-I and 4A-J were expressed at much lower levels ( Figure 2).
Genome-wide transcriptional profiles of the single cps cluster deletion strains were generated, confirming the complete abolishment of expression of the genes affected by the mutation, and thus establishing the integrity of the mutants. These transcriptome analyses also revealed that cps cluster deletion affected the expression of variable gene sets in the individual cps cluster mutants as compared to the wild-type strain. Deletion of cps cluster 1 or 2 affected the largest group of genes (Table 1), whereas the transcriptome impact of deletion of cluster 4 was considerably smaller, and deletion of cluster 3 appeared the smallest (Table 1). Intriguingly, the transcriptome changes elicited upon deletion of cps clusters 1 or 2 as compared to the wild-type strain, shared a substantial number of downregulated genes (87 shared repressed genes) for which the encoded functions are associated with transport and metabolism of certain amino acids, including serine, glutamate, aspartate, asparagines, methionine, cysteine, and the ABCtransporter for the branch chain amino acids (Additional file 2). Notably, genes that were consistently upregulated upon deletion of cps clusters 1 and 2 (30 upregulated genes shared, Additional file 2) included some that encoded amino acid related functions, including a diand tri-peptide transport system as well as specific glutamine and a facilitator-family branched-chain amino acid transport systems, but also several genes with predicted cell-envelope associated functions like several extracellular proteins, and genes involved in aminosugar synthesis and teichoic acid decoration. Remarkably, of the 20 upregulated genes observed in the cps cluster 3 deletion mutant relative to the wild-type strain, 11 were shared with the upregulated gene set found in the cps cluster 4 deletion mutant (Additional file 2). These genes included those encoding metabolic functions associated with pyruvate discipation or its control like, L-lactate dehydrogenase (ldhL2), pyruvate dehydrogenase (pdhA, pdhB; also activated in the cps 2 deletion strain), NADH oxidase (nox5), but also several carbohydrate degradation and transport functions like an α-amylase (malS), and the genetically linked and correlated transport functions for maltodextrins (mdxG), as well as two 6-phospho-beta-glucosidases (pbg5, and pbg4). Among these 6-phospho-beta-glucosidase encoding genes, the pgb4 gene appeared to be induced in all cps cluster deletion strains, a feature shared by only two additional genes that appeared to be consistently downregulated in all mutants, and encode a prephenate dehydrogenase (tyrA), and a sodium-coupled N-acetylneuramidate transporter (lp_3563). In addition, expression of a FAD/FMN-containing dehydrogenase encoding gene (lp_0291) appeared to be differentially affected in the 4 cps deletion strains, i.e., downregulated upon deletion of cps cluster 1 or 2, but upregulated when cps cluster 3 or 4 is deleted. Importantly, compensatory activation of one of the alternative cps clusters was not observed in any of the cps cluster deletion mutants. These results show that especially deletion of cps cluster 1 or 2 elicited pleiotropic transcriptome consequences, which appeared to be centered on genes with functions associated with metabolism and transport of various amino acids, but also included genes encoding several other functions. The observation that deletion of the capacity to produce CPS affected the expression of genes encoding specific transport and metabolism functions, may suggest that the presence of polysaccharides in the cell envelope plays a role in the access that the bacteria have to nutrients from the environment. Polysaccharides may function as macromolecules that sequester nutrients and thereby facilitate their import [28,29], or alternatively they may form a capsular structure that surrounds the cell and retards or inhibits diffusion or transport of nutrients towards the membrane surface and thereby reduces trans-membrane transport efficiencies.

Impact of individual cps cluster mutations on L. plantarum surface glycan characteristics
Surface polysaccharides of L. plantarum WCFS1 and the cps cluster deletion mutants were isolated to assess their monosaccharide composition ( Table 2). The results clearly demonstrated that the wild-type strain produces significant amounts of surface polysaccharides. Deletion of cluster 1 did not impact on the total amount of surface polysaccharides produced, which is remarkable in  the light of the relatively high expression of the cps1A-I gene cluster in the wild-type ( Figure 2) and the lack of compensatory expression changes of clusters 2, 3, or 4 in the cps1A-I mutant. However, the deletion of cluster 1 led to a decreased molar mass of the isolated polysaccharides and influenced the monosaccharide composition, i.e. led to a reduced relative amount of galactose and to a complete lack of rhamnose. This observation establishes that the cps1 cluster is functional and leads to production of a specific polysaccharide in L. plantarum WCFS1. In addition, the lack of rhamnose in the surface polysaccharide fraction obtained for strain NZ3548Cm (Δcps1A-I) is in agreement with the annotation of cps1H as the only rhamnosyltransferase found in all cps clusters. As the expression of the rfb genes (genetically linked to the cps1 cluster) was not abolished in response to cps1A-I deletion (even slightly increased expression), it is likely that indeed the deletion of cps1H has led to the loss of rhamnose in the surface polysaccharide fraction. Despite the low expression levels of clusters 2, 3 and 4 (Figure 2), deletion of each cluster reduced the production of surface polysaccharides, indicating their respective contributions to the overall surface polysaccharides produced by the wild-type strain ( Table 2). The polysaccharides isolated from these mutants had a similar molecular mass as the polysaccharides isolated from the wild-type strain, indicating a particular role of the cps cluster 1 in polysaccharide chain length determination. Deletion of cluster 2 reduced the relative abundance of galactose in the surface polysaccharides, indicating the presence of this sugar in the cps2-encoded polysaccharide. In contrast, deletion of clusters 3 or 4 did not significantly affect the monosaccharide compositions of the polysaccharides produced. Notably, all polysaccharide molecules detected in the cps2A-J, cps3A-J and cps4A-J mutant strains contained rhamnose, whereby the presence of this monosaccharide in the surface polysaccharide fraction appears to be exclusively dependent on the cps1A-I cluster.
Deletion of all 4 cps clusters (Δcps1A-3J, Δcps4A-J) led to a substantial reduction (approximately 90% reduction in comparison to the wild-type) of the overall amount of surface polysaccharides isolated, which completely lacked rhamnose and showed a reduced relative amount of galactose, analogous to the Δcps1A-I strain. As this mutant is expected to lack the genetic capacity to produce these typical wzy-dependent polysaccharides, the surface glycans isolated from this strain may derive from other surface polymers that are synthesized via other mechanisms. These remaining polysaccharides contain glucose, glucosamine, and galacturonic acid and might derive from teichoic acids, glycosylated proteins, and/or peptidoglycan. Remarkably, mutation of the individual cps clusters did not lead to an apparent phenotypic change in terms of growth or cell morphology, while the deletion of all four clusters (Δcps1A-3J, Δcps4A-J) caused aggregation of cells and rapid sedimentation (data not shown), which may be explained by increased cell surface-hydrophobicity due to reduced amounts of the typically hydrophilic surface polysaccharides [30].

Deletion of cps clusters influences L. plantarum TLR2-mediated NF-κB activation
The effects of cps cluster deletion on host cell signaling were examined in HEK-293 reporter cell lines stably expressing human Toll-like receptor (TLR)2, which harbor a reporter plasmid containing firefly luciferase under the control of the human nuclear factor (NF)-κB promoter. HEK-293 cells do not normally produce TLRs and have been previously shown to be unresponsive to several microbial-associated molecular patterns (MAMPs) [31,32] HEK-293 cells transiently transfected with only the NF-κB reporter plasmid (pNiFTY) were unresponsive to TLR2 agonists of L. plantarum demonstrating the requirement of human TLR2 in this signaling pathway (data not shown). An L plantarum wild-type like derivative (NZ3400Cm) activated TLR2dependent NF-κB ( Figure 3). Overall, activation of NF-κB remained unaffected by Δcps1A-I and Δcps3A-J but was slightly increased after stimulation by the Δcps2A-J and Δcps4A-J mutants, which are the strains that displayed the largest reduction in the amount of surface polysaccharides produced. Notably, exposure of TLR2expressing HEK-293 cells to Δcps1A-3J and Δcps1A-3J, Δcps4A-J elicited the strongest responses and clearly activated NF-κB. Although cps4A-J appears to be expressed at very low levels, its deletion in the Δcps1A-3J background led to a substantial increase of TLR2-mediated NF-κB activity of the resulting strain, demonstrating that deletion of all cps clusters and the concomitant severe reduction of surface polysaccharide production leads to enhanced release and/or exposure of TLR2-agonists. These results suggest a role of surface polysaccharides in the shielding of other L. plantarum cell envelope MAMPs such as lipoproteins and teichoic acids that could activate TLR signaling, as was previously proposed for Lactobacillus rhamnosus GG [21].

Conclusions
In L. plantarum, surface polysaccharide biosynthesis is encoded by 4 gene clusters, which independently contribute to the overall surface polysaccharides produced in this bacterium. Transcriptome analysis revealed that cps cluster deletion affected different sets of genes in the individual cps cluster mutants, whereas substantial numbers of regulated genes were shared between cps cluster mutant 1 and 2. Surface polysaccharide analysis revealed that the individual clusters influenced specific features of the polysaccharides produced, i.e. the amount, composition and molar mass. Although some of the individual mutants moderately affected TLR2-mediated NF-κB signaling, deletion of all clusters elicited a drastically increased NF-κB activation. In conclusion, the cps cluster encoded surface polysaccharides contribute to the L. plantarum cell surface architecture, and probably reduce release and/or exposure of TLR2-activating bacterial molecules.

Bacterial culture conditions
Escherichia coli strain TOP-10 (Invitrogen, Carlsbad, USA), which was used as intermediate cloning host, was grown aerobically in TY medium. L. plantarum WCFS1 [33,34] and its derivatives (Additional file 3) were either cultivated in Mann-Rogosa Sharpe (MRS; Merck, Darmstadt, Germany) or in 2-fold concentrated (2 ×) chemical defined media (CDM) [35] supplemented with 1.5% (wt/vol) glucose without agitation. All bacteria were grown at 37°C and when appropriate, the media were supplemented with antibiotics; for E. coli and L. plantarum chloramphenicol was added at 10 μg/mL, and during replica plating of L. plantarum chloramphenicol and erythromycin were added at final concentrations of 10 μg/mL and 30 μg/mL, respectively.

DNA manipulation techniques
Plasmid DNA was isolated from E. coli using Jetstar columns following the manufacturer's instructions (Genomed GmbH, Bad Oberhausen, Germany). Restriction endonucleases (Fermentas, St. Leon-Rot, Germany), KOD-DNA polymerase (Toyobo, Osaka, Japan) and T4 DNA ligase (Invitrogen, Carlsbad CA, USA) were used as recommended by the manufacturers. Primers were obtained from Sigma Aldrich (Zwijndrecht, The Netherlands), and DNA sequencing reactions were performed at BaseClear (Leiden, The Netherlands). Chromosomal DNA isolation, preparation of electrocompetent cells and DNA transformation of L. plantarum were performed as described previously [36,37].

Construction of cps cluster deletion mutants
Plasmids, primers and strains used in this study are listed in Additional file 3. The cps deletion mutants were constructed according to previously described methods [38], by which the target cps clusters were replaced by a chloramphenicol acetyltransferase (cat) gene cassette. In this study a derivative of the commonly used mutagenesis vector pNZ5319 [38], designated pNZ5319TAG (Bron et al., unpublished observations) was used that introduces a unique DNA-tag into the chromosome during gene deletion, which can be used for detection were subsequently joined by a second PCR to the tag-lox66-F3/tag-lox71-R3 or tag-lox66-F2/tag-lox71-catR2 cassette, respectively. The resulting amplicons were cloned into SwaI-Ecl136II digested pNZ5319TAG. The obtained mutagenesis plasmids were integrated into the L. plantarum WCFS1 chromosome by double cross over replacement of the target gene (clusters) by the cat cassette, yielding the deletion mutant strains NZ3548Cm (Δcps1A-I), NZ5333ACm (Δcps2A-J), NZ3549Cm (Δcps3A-J), and NZ3550Cm (Δcps1A-3J), as well as the tagged strain NZ3400Cm (lp_2681-P 32 -cat-lp_2683). For construction of the quadruple mutant (Δcps1A-3I, Δcps4A-J), the mutagenesis plasmid pNZ3550 (Additional file 3) was integrated into the Δcps4A-J chromosome, in which the cat cassette was removed by the temporal expression of the Cre recombinase [38], yielding the deletion mutant strain NZ3534Cm (Δcps1A-3I, Δcps4A-J) [39]. The anticipated genotype of all mutants was confirmed by PCR using primers flanking the sites of recombination.

RNA isolation
L. plantarum and its cps-cluster deficient derivatives were grown in 2 × CDM and RNA was isolated according to previous described methods [40,41]. In short, following methanol quenching [42], cells were harvested by centrifugation (6000 × g, 20 min, 4°C), resuspended in 400 μL ice-cold CDM medium and transferred to tubes containing 500 μL phenol/chloroform solution (4:1 [v/v]), 30 μL 10% sodium dodecyl sulfate, 30 μL 3 M sodium acetate (pH 5.2), and 0.5 g zirconium beads. Cells were disrupted by bead beating using a Savant FastPrep FP120 instrument (Qbiogen Inc., Illkirch, France), and RNA was purified from the aqueous phase using the High Pure Isolation Kit (Roche Diagnostics, Germany). RNA concentration and purity were determined using A260 and A280 measurements using a ND-1000 spectrometer (NanoDrop Technologies Inc., Wilmington, United States), and RNA quality was verified with a 2100 Bioanalyzer (Agilent Technologies, Amstelveen, the Netherlands). Samples that displayed a 23S/16S RNA ratio equal or superior to 1.6 were used for labeling.

Transcriptome analysis and interpretation
3 μg RNA was used for cDNA synthesis. Cyanine-3 (Cy3) and cyanine-5 (Cy5) cDNA labeling was performed as described previously [43], using the CyScribe Post-Labeling and Purification kits according to the manufacturer's instructions (Amsersham Biosciences, Buckinghamshire, UK). Cy-dye-labeled cDNAs (0.5 μg each) were hybridized to L. plantarum WCFS1 printedoligonucleotide DNA microarrays (Agilent Technologies, Amstelveen, the Netherlands). The array design and transcriptome data were deposited under platform GPL13984 and accession number GSE34690 in NCBI's Gene Expression Omnibus (GEO) [44,45] at http://www. ncbi.nlm.nih.gov/geo/. Hybridization and scanning procedures were performed as previously described [43]. Slide scanning was carried out at several photo multiplier tube (pmt) values, and the optimal scan of each individual microarray was selected on the basis of signal distribution (combination of a low number of saturated spots and a low number of low signal spots). The data were normalized using the Lowess normalization as available in MicroPrep [46]. For statistical significance, Benjamini and Hochberg's False Discovery Rate (FDR) was used [47], with a FDR-adjusted p-value cutoff of 0.05, employed for genes showing at least 2-fold altered expression.

Surface polysaccharide isolation and determination of glycan composition
Surface polysaccharides were isolated and characterized according to previously described methods [48,49]. L. plantarum WCFS1 and its mutant derivatives were grown in 2 × CDM until late stationary phase. After growth, cultures were incubated at 55°C for 1 h, followed by pelleting of the bacterial cells (6000 x g, 15 min, room temperature). The supernatants were supplemented with erythromycin (30 μg/mL), transferred to dialysis tubes (molecular weight cutoff of 12-14000 Da, Fisher Scientific, Landsmeer, The Netherland) and dialyzed overnight against running tap water followed by dialysis for 4 h against deionized water. The dialyzed samples were freeze-dried and stored at −20°C until further analysis. Samples were dissolved in eluent (100 mM NaNO 3 + 0.02% NaN 3 ), and polysaccharides were separated by size exclusion chromatography (SEC), light scattering was measured at 632.8 nm, UV-adsorption of proteins was measured at 280 nm, viscosity was measured with a viscosity detector (ViscoStar, Wyatt Technologies, Santa Barbara, USA), and sample concentrations were measured by determining their refraction index. During SEC, polysaccharide peaks were collected and hydrolyzed with 2 M trifluoroacetic acid (TFA), dried and dissolved in water. The quantitative monosaccharide composition of the polysaccharide fractions was analyzed using High Performance Anion Exchange Chromatography with Pulsed Amperometric Detection (HPAEC-PAD) equipped with a gold electrode. The monosaccharides were eluted isocratically with 16 mM