EpsA is an essential gene in exopolysaccharide production in L actobacillus johnsonii FI9785

Summary L actobacillus johnsonii FI9785 has an eps gene cluster which is required for the biosynthesis of homopolymeric exopolysaccharides (EPS)‐1 and heteropolymeric EPS‐2 as a capsular layer. The first gene of the cluster, epsA, is the putative transcriptional regulator. In this study we showed the crucial role of epsA in EPS biosynthesis by demonstrating that deletion of epsA resulted in complete loss of both EPS‐1 and EPS‐2 on the cell surface. Plasmid complementation of the epsA gene fully restored EPS production, as confirmed by transmission electron microscopy and nuclear magnetic resonance (NMR) analysis. Furthermore, this complementation resulted in a twofold increase in the expression levels of this gene, which almost doubled amounts of EPS production in comparison with the wild‐type strain. Analysis of EPS by NMR showed an increased ratio of the heteropolysaccharide to homopolysaccharide in the complemented strain and allowed identification of the acetylated residue in EPS‐2 as the (1,4)‐linked βGlcp unit, with the acetyl group located at O‐6. These findings indicate that epsA is a positive regulator of EPS production and that EPS production can be manipulated by altering its expression.


Introduction
Exopolysaccharides (EPS) are one of the key components of the cell surface of lactic acid bacteria (LAB) and have a considerable impact on their surface characteristics (Broadbent et al., 2003). EPS can be either bound to the cell surface or secreted to the environment; structurally, they are classified as homopolysaccharides, which are composed of one type of sugar monomer, and heteropolysaccharides, which contain two or more sugar subunits and sometimes other organic molecules in their structure (Welman and Maddox, 2003;Badel et al., 2011). A single gene defined as gtf is commonly responsible for the production of homopolymeric glucose EPS in LAB (Walter et al., 2008;Badel et al., 2011). The biosynthesis of heteropolymeric EPS is more complex and a cluster of 8-17 genes is required, including conserved genes for regulation, repeat unit biosynthesis, chain length determination, polymerization and export (Cieslewicz et al., 2001;Jolly and Stingele, 2001;Broadbent et al., 2003;Peant et al., 2005;Berger et al., 2007;Lebeer et al., 2009). EPS produced by LAB have a great variety of structures and have many applications in the dairy industry, particularly in improving texture and as thickening and stabilizing agents (Welman and Maddox, 2003;Badel et al., 2011). Consequently the production of novel or modified structures and strategies to increase production in LAB are of significant industrial interest.
Lactobacillus johnsonii FI9785 has been shown to act as a competitive exclusion agent against Clostridium perfringens and other pathogens in poultry (La Ragione et al., 2004). The mechanism of the exclusion process is still undefined, but one of the proposed modes of action is that the bacteria adhere to the gastrointestinal tract and prevent the colonization of pathogenic bacteria (Reid and Burton, 2002), a process which is related to cell surface characteristics (Lebeer et al., 2008). Recently, we demon-strated that L. johnsonii FI9785 produces two different EPS: homopolymer EPS-1 is a branched dextran with every backbone residue substituted with a 2-linked glucose unit, and heteropolymer EPS-2 has a hexasaccharide repeating unit composed of two galactose and four glucose residues with different types of linkages between each sugar residue . The genome of L. johnsonii FI9785 does not contain a classical gtf gene. However, an eps gene cluster with a similar organizational structure to those associated with heteropolymeric EPS production was identified, with 14 putative genes related to EPS biosynthesis (Horn et al., 2013). Nuclear magnetic resonance (NMR) analysis of a mutant where the eps cluster had been deleted showed that this cluster was required for the biosynthesis of both EPS-2 and EPS-1 . Mutations in this cluster resulted in alterations in phenotype, EPS production, surface characteristics and cell-cell interactions (Horn et al., 2013;Dertli et al., 2015). The first gene in the cluster, epsA, has homology to the LytR family of transcriptional regulators; in Streptococcus pneumoniae deletion of a similar gene cpsIaA caused a reduction in capsule production (Cieslewicz et al., 2001).
In the current study, we explored the importance of the epsA gene by demonstrating that deletion of epsA resulted in an acapsular phenotype, while complementation by plasmid expression fully restored EPS production and resulted in an almost twofold increase in EPS compared with the wild type. This demonstrates that epsA controls EPS biosynthesis and can be used to either prevent or increase its production.

Results and discussion
To investigate the importance of epsA, the epsA coding sequence was knocked out by in-frame deletion mutagenesis and this deletion strain (ΔepsA) was also complemented with the wild-type epsA gene on a plasmid under the control of a constitutive promoter (ΔepsA::pepsA, see supporting information). There were no alterations in the growth rate or colony phenotype in comparison with the wild-type strain (Fig. 2C); however, transmission electron microscopy analysis showed the absence of an EPS layer in ΔepsA while the cell walls of the wild type and ΔepsA::pepsA strains appeared thicker and less distinct due to the accumulation of EPS (Fig. 1). Cell surface-associated EPS were isolated from cell pellets of both mutant strains and were subjected to NMR analysis to determine their structures ( Fig. 2A and B). Analysis and quantification of EPS confirmed that ΔepsA was unable to produce either EPS-1 or EPS-2, resulting in an acapsular phenotype. Complementing the epsA gene in this mutant fully restored the biosynthesis of both EPS.
The 1 H NMR spectrum of the ΔepsA::pepsA mutant ( Fig. 2B) showed the presence of both EPS-1 and EPS-2. There was an increased ratio of EPS-2 to EPS-1, which contrasted with the composition of the mixture obtained from the wild type (also isolated from the bacterial cell pellet) in which EPS-1 was the major polysaccharide . 1 H NMR spectra of the wild type (WT) and ΔepsA are provided for comparison with ΔepsA::pepsA as Supporting information (Fig. S1). Previously, acetyl groups were identified but not assigned to specific residues . However it was established that the acetyl group in the hexasaccharide repeating unit of EPS-2 gives rise to a sharp singlet at 2.15 ppm in the 1 H spectrum (Fig. 2B); by integration there is one acetyl group per repeating unit. The other signals at ∼ 1.98 to 2.08 ppm are associated with EPS-1 and although integration shows that there is approximately one acetyl group per disaccharide repeating unit, the presence of multiple peaks means that these groups must be distributed across several locations. The data reported previously  referred to non-acetylated EPS-2, which was the predominant component in the sample analysed from the smooth mutant epsC D88N . EPS-2 from ΔepsA::pepsA was 100% acetylated in addition to being the major polysaccharide, and this, together with comparison with the spectrum of epsC D88N , enabled a more accurate analysis of acetylation. A detailed study of the 2D NMR spectra showed that in this preparation EPS-2 was acetylated at just one position: C6 of the (1,4)-linked β-Glcp residue, g. The 1 H and 13 C chemical shifts of EPS-2 are listed in Table 1. The location of the acetyl group in acetylated EPS-2 is revealed by the downfield displacement, compared with non-acetylated EPS-2, of the g6 signals to δ4.36/4.55 from δ3.84/4.00 ( 1 H) and to δ65.73 from δ63.04 ( 13 C). Shifts of neighbouring atoms (h1, g5) are also affected by the presence of the acetyl group as indicated in Table 1; the remaining shifts are essentially unchanged.
The acetylation of bacterial EPS and plant polysaccharides has been shown to contribute to the gelling properties of these biopolymers (Robijn et al., Wild type epsA epsA::pepsA PM CW EPS Fig. 1. Transmission electron microscopy of wild type and mutant strains. L. johnsonii FI9785 (wild type) and its derivatives ΔepsA, made by deletion mutagenesis using the thermosensitive pG + host9 vector system, and complemented strain ΔepsA::pepsA (see supporting information) were grown and visualized as described previously . The bar represents 100 nm. CW, cell wall, PM, plasma membrane. 1995; Huang et al., 2002) and it was also reported that acetylation of the alginate produced by Pseudomonas aeruginosa increases the in vivo adherence of this pathogen to lung epithelium in cystic fibrosis patients (Riley et al., 2013). Although some functions have been attributed to the acetylation of the polysaccharides in both Gram-positive and Gram-negative bacteria, such as the provision of additional protection against the many types of hydrolases produced by gut bacteria, the acetylation mechanism of EPS has not been identified in detail. It was shown that an inner membrane protein WecH was responsible as an acetyltransferase for the O-acetylation of the cell surface polysaccharide of Escherichia coli K12, and the gene encoding WecH was not located in the gene cluster responsible for the production of this polysaccharide (Kajimura et al., 2006). More recently a gene within the algD operon of the Gram-negative P. aeruginosa responsible for alginate production, designated as Algx, was shown to be an acetyltransferase responsible for the acetylation of alginate (Riley et al., 2013). From Gram-positive bacteria, a gene required for the O-acetylation of EPS from Staphylococcus aureus (cap5h) was identified and shown to have a positive effect on colonization of mice (Bhasin et al., 1998), and epsH from the eps gene cluster of Streptococcus thermophilus Sfi6 showed homology to this gene (Stingele et al., 1999). None of the predicted gene products from the L. johnsonii FI9785 eps cluster show amino acid homology to acetyltransferases, but the genome harbours several putative acetyltransferase genes (Wegmann et al., 2009). B. 600 MHz 1 H NMR spectrum (338°K, D2O) of EPS extracted from the ΔepsA::pepsA mutant. EPS were isolated from 500 ml cultures as previously described (Tallon et al., 2003); lyophilized samples were dissolved in D2O and 1D and 2D NMR spectra were obtained using the same pulse sequences as described previously . C. Growth of strains to stationary phase. Optical density (OD600) readings represent the mean of triplicate samples +/− standard deviation. ■, wild type; □, ΔepsA; ▲, ΔepsA::pepsA. D. Increase in EPS production after complementation. EPS were quantified using the phenol-sulfuric acid method (DuBois et al., 1956) with glucose as a standard and expressed as a quantity of EPS production per 10 7 cfu (mean of triplicate samples +/− standard deviation). Black, wild type; white, ΔepsA::pepsA. E. Aggregation of wild type and mutant strains. The aggregation percentage was measured after overnight incubation (16 h) by flow cytometry (Dertli et al., 2015). Results are the mean of triplicates +/− standard deviation. Black, wild type; grey, ΔepsA; white, ΔepsA::pepsA. Most of these are associated with the acetylation of core cell wall components, but a novel gene/s within this set could be responsible for the acetylation of EPS; the translated product of one locus, FI9785_RS04565, has homology to similar domains to those found in EpsH and Cap5H.
Having established that the L. johnsonii ΔepsA::pepsA strain was able to produce both EPS types, we subsequently investigated the EPS production levels in this mutant. Quantitative polymerase chain reaction (qPCR) analysis was performed to compare the epsA gene expression levels in wild type, ΔepsA::pepsA and ΔepsA mutant cells (see Supporting information). The expression of the epsA gene in the ΔepsA mutant could not be detected by qPCR analysis as expected (data not shown). In contrast, complementation of the epsA gene in the ΔepsA::pepsA mutant resulted in a 1.99 ± 0.34-fold increase in the epsA gene expression level compared with that in the wild type. This was reflected in the accumulation of EPS -overexpression of epsA resulted in clearly increased levels of total EPS production, giving 817.6 +/− 108.4 μg/10 7 colony-forming units (cfu) compared with the wild type (441.2 +/− 73.5 μg/10 7 cfu) (Fig. 2D).
Deletion of the putative transcriptional regulator of a similar heteropolysaccharide cluster in S. pneumoniae led to a reduction in capsule production, but not to a complete absence (Cieslewicz et al., 2001). Our experiments suggest a simple relationship between epsA gene expression and EPS production under these growth conditions, although expression analysis of the genes of the eps cluster is required to confirm the relationship between epsA and transcriptional regulation of EPS. In other studies, the relationship between transcription and production is not so direct: the EPS production of four Lactobacillus rhamnosus strains varied widely but the transcription levels of genes in eps clusters were similar (Peant et al., 2005), while EPS production in Lactococcus lactis strains varied with sugar source without changes in the activity of the eps promoter (Looijesteijn et al., 1999). Several factors can affect the levels of EPS production including the availability of sugar nucleotides (Looijesteijn et al., 1999;Boels et al., 2001;Levander et al., 2002), the expression level of genes in sugar catabolism pathways (Looijesteijn et al., 1999;Levander et al., 2002) and the carbon source utilized by the bacteria, as well as the transcriptional level of the genes responsible for EPS production (Looijesteijn et al., 1999;Audy et al., 2010). Previously, it was reported that a higher transcription level of the priming glycosyltransferase resulted in higher EPS production in Bifidobacterium longum subsp. longum CRC 002 (Audy et al., 2010). In another study the overexpression of the fructosebiphosphatase, which is a key enzyme in the biosynthesis of sugar nucleotides from fructose, resulted in increased EPS production in L. lactis (Looijesteijn et al., 1999). Similarly, it was shown that an almost threefold increase in the expression of an eps gene cluster in L. lactis gave a nearly fourfold increase in EPS production, although there was a reduction in the cell growth, potentially due to the fact that the sugar nucleotides are used not only in EPS production but also for bacterial metabolic activities including cell wall biosynthesis (Boels et al., 2003). In the current study the increased EPS production in L. johnsonii ΔepsA::pepsA did not significantly change the growth profile. It should be noted that although the eps cluster is required for the biosynthesis of both EPS, overexpression of epsA resulted particularly in an enhanced production of heteropolysaccharide EPS-2, which supports our previous suggestion that a novel gene outside of the eps cluster might be involved in EPS-1 production in conjunction with a gene/s in the eps cluster . a. The residues are listed in the order in which they occur in the linear repeating unit of the acetylated EPS-2, i.e. a is linked to f, f to d etc. and g to a. The repeating unit is acetylated at C6 of residue g. b. Shifts which differ most from the previously reported  non-acetylated EPS-2 are indicated in bold. c,d. 13 C assignments with the same superscript letter may need to be interchanged. The complete loss of an EPS layer in the ΔepsA mutant resulted in significantly higher aggregation levels (almost 100%) compared with the wild type (51.2 +/− 4.5%), while the increased EPS production in the ΔepsA::pepsA complemented strain resulted in a sharp decrease in aggregation to only 13.3 +/− 6.9% (Fig. 2E). These results agree with previous analysis of L. johnsonii mutants with reduced or increased EPS production, which found that a ΔepsE strain which only produced a thin layer of EPS-1 also exhibited nearly 100% aggregation (Horn et al., 2013;Dertli et al., 2015); here, a similar level of aggregation with no EPS indicates that cell surface components revealed by the reduction in the EPS layer are more likely to be the positive agents of aggregation than EPS-1.
Adhesion to HT29 cells, measured as described previously (Horn et al., 2013), was also increased in the acapsular ΔepsA mutant (18.4 +/− 0.3%) compared with the wild type (13.7 +/− 0.3%), while complementation in ΔepsA::pepsA restored the adhesion level to wild type levels (13.5 +/− 1.6%) despite the reduction in aggregation. Previously, increased EPS production in a smooth colony mutant epsC D88N produced a reduction in adhesion, and the finding here that a greater increase in EPS content still gave wild type adhesion levels lends further weight to suggestions that other factors such as chain length, attachment or ratio between EPS types have an impact on cell surface characteristics (Horn et al., 2013).
In conclusion, the current report shows that epsA is required for the biosynthesis of both homopolymeric EPS-1 and heteropolymeric EPS-2. The analysis of EPS production and epsA gene expression in the ΔepsA::pepsA mutant indicates that EPS production can be almost doubled by overexpressing the putative transcriptional regulator without any disruption in cell growth.

Supporting information
Additional Supporting Information may be found in the online version of this article at the publisher's web-site: Fig. S1. 600 MHz 1 H NMR spectra of EPS (300°K, D2O) isolated from WT, ΔepsA and ΔepsA::pepsA strains. Table S1. Primers designed for epsA, 16S and gyrB genes for qPCR analysis.