Discussion
In this study, seven porcine intestinal L. amylovorus strains, and the type strain of L. amylovorus, which is not of intestinal origin, were characterized in vitro for their abilities 1) to adhere to porcine mucus 2) to bind to epithelial cells of the pig small intestine, 3) to inhibit the adherence of an F4-fimbriated ETEC strain to porcine intestinal epithelial cells, 4) to produce soluble inhibitors against intestinal pathogens and 5) to induce immune signaling in dendritic cells.
None of the eight strains studied exhibited any efficient adherence to porcine gastric or intestinal mucus, as the level of adherence was around 2% or less. In previous studies, lactic acid bacteria or pathogens exhibiting similar levels of adhesion have been among the least adhesive strains, and they are considered to be non-adhering [35,44,45]. In addition, the highly variable adherence of our strains in the different experiments strongly suggests that the binding was non-specific e.g. being due to hydrophobic interactions, a common complication encountered in interpreting the results of mucus binding studies [46,47]. Furthermore, the analysis of the genomic sequences of L. amylovorus GRL 1115 and GRL 1116 revealed the complete absence of genes encoding putative mucus binding proteins in these strains (unpublished). Although the presence of unidentified mucus adhesins cannot be completely excluded, the level of adherence of these strains to mucus can be considered as negligible. Both efficiently and poorly mucus-binding Lactobacillus strains have been isolated from the intestine and milk of swine [48-51], and the lack of a mucus-adhering capability is not uncommon among the widely used human probiotic lactobacilli [52]. Furthermore, growth conditions not tested in this study, such as cultivation on a solid medium or the addition of mucin to the standard culture, might have triggered the mucus-binding capacities of the strains, as described previously for L. reuteri[48]. Other binding functions, such as the capacity to adhere to extracellular matrix components, have previously been described for these GRL-strains [28].
The adherence of the L. amylovorus strain DSM 16698 to IPEC-1 cells and its ability to inhibit the binding of ETEC to IPEC-1 cells in a competition-type assay have been demonstrated previously [26]. In this study, the reported adhesion of DSM 16698 to IPEC-1 cells was confirmed, and the adherence of GRL 1112 and 1115 was found to be at a similar level. These three well-adhering strains were also able to inhibit the adherence of ETEC to IPEC-1 cells in competition and exclusion assays. Surprisingly, the poorly adhesive strain GRL 1118 similarly inhibited ETEC adherence, suggesting that mechanisms other than competitive binding were involved in the inhibition, e.g., secreted inhibitory factors or coaggregation with the pathogen [53]. However, the spent culture supernatant of GRL 1118 did not reduce the growth of F4-positive ETEC more than the culture supernatants of the other GRL strains, and the growth inhibition in all cases was mainly attributable to the production of lactic acid. The production of substances specifically able to inhibit adherence was not tested in this study. Thus, at present we have no clear explanation for the observed inhibition of ETEC binding to IPEC-1 cells by GRL 1118.
One of the main mechanisms of probiotic action in the gastrointestinal tract is the modulation of mucosal and systemic immune responses [54]. These immunomodulatory properties, including immunoregulatory and tolerance-promoting, as well as pro-inflammatory functions, have been suggested to result from the stimulation of mucosal dendritic cells by probiotic bacteria [55,56]. Many different Lactobacillus species have been shown to modulate dendritic cell responses in studies with human or murine DCs [12,55-63]. L. reuteri ASM 20016, L. casei NIZO B255 [55] and L. acidophilus NCFM [12] specifically bind the DC-SIGN molecule (dendritic cell specific C-type lectin intercellular adhesion molecule 3-grabbing non-integrin) on dendritic cells, triggering the differentiation of naïve T cells towards the Treg[55] or Th2[12] functional types. The bacterial component of L. acidophilus NCFM which interacts with DC-SIGN is its S-layer protein SlpA [12]. However, proinflammatory or Th1-polarising effects in DCs have also been described for NCFM [64-66]. These responses have been attributed to either lipoteichoic acid [66] or the S-layer associated protein encoded by the gene in locus Lba-1029 of NCFM [67]. L. amylovorus and L. acidophilus are phylogenetically closely related [68], and the S-layer protein of NCFM shows remarkable amino acid sequence similarity with the major Slp:s of the L. amylovorus strains studied (see Additional file 2). These findings led us to investigate the potential of our S-layer-carrying L. amylovorus isolates to induce cytokine production in human DCs. There were evident strain-specific differences in L. amylovorus immunomodulating capacities, but no clear preference was noted for any of the strains for inducing cytokines to drive the immune response exclusively towards either the Th1 or the Th2 type. Instead, most of the strains induced both pro-inflammatory cytokines (TNF-α and IL-6), IL-12 favouring a Th1 response and IL-10 favouring a Th2-type response, a phenomenon that has also been demonstrated for L. gasseri strains [63]. Although the specific immunomodulating surface molecules of L. amylovorus remained unidentified, the results of this work emphasize the importance of strain-specific differences in the immunomodulating capacities and are thus in line with previous studies [57,59,61,63]. Considering the probiotic potential of lactobacilli, it is clear that for the optimal performance of the complex immune system, Th1, Th2 and Treg responses have to be balanced. The most successful manipulation by probiotics will also depend on the dose and strain combination of probiotic bacteria, the type of pathogen challenge, and the specific environmental conditions [55,57,69].
In the in silico analysis of the L. amylovorus S-layer proteins, we found that the amino acid sequences of the L. amylovorus Slp:s studied, excluding SlpC, were very similar to the amino acid sequence of L. acidophilus NCFM SlpA, especially in the carboxy-terminal region, a phenomenon observed among the S-layer proteins of other L. acidophilus-related lactobacilli as well [70]. The pattern of conservation apparently reflects the well-known role of the carboxy-terminal domains in cell wall binding [70,71], and strongly suggests that the cell-wall binding function also resides in the carboxy-terminal region in L. amylovorus Slp:s. The amino-terminal parts of L. amylovorus Slp:s, apparently facing the environment, are more variable, but the valine-rich regions, which flank the amino-terminal domain in the S-layer protein CbsA of L. crispatus, and which have been shown to be important for the self-assembly of CbsA monomers [70], were however conserved in most of the studied L. amylovorus S-layer proteins.
S-layers typically form the outermost layer of the bacterial cell, making them attractive candidates for being involved in adherence to host cells. However, attempts to create completely S-layer negative Lactobacillus mutants have been unsuccessful [11,72-74], emphasizing the necessity of at least one functional S-layer protein for lactobacilli and compelling us to investigate the role of these proteins by utilizing protein-level methods rather than with knock-out mutants. However, as Slp:s are poorly soluble, the presence of S-layer proteins in an aggregated form in the assays may evoke unspecific effects and compromise the reliability of the results. Indeed, due to the methodological difficulties related to the poor water-solubility of S-layer proteins, the results of most of the previous reports examining the role for Lactobacillus S-layers in adherence have remained ambiguous, and so far only a few Lactobacillus S-layer proteins have been convincingly shown to act as epithelial cell adhesins or to bind to extracellular matrix proteins or immune cells [9-13]. In our experiments, we created a protein presentation system based on the self-assembly of recombinant L. amylovorus S-layer proteins on purified cell wall fragments from each strain. The method of coating the CWF was designed to minimize Slp precipitation being based on the finding that even in an aqueous buffer, a small fraction of Slp molecules remains visibly unprecipitated, and this dilute protein fraction can be separated from the precipitate by centrifugation, as previously described [33]. In the carrier system, the CWF-protein complexes are handled similarly as whole bacterial cells and thus the method largely evades the solubility-related problems. Furthermore, it allowed the use of uncoated CWF as controls and the presentation of the proteins in the native, symmetric organization observed on the bacterial surface, i.e. the obtained results possess real biological relevance.
The results of the adhesion experiments clearly indicated that none of the major S-layer proteins of the L. amylovorus strains on their own mediated bacterial adherence to IPEC-1 cells: when compared to uncoated cell walls, all the proteins exhibited at least a low level of adherence irrespective of the adhesive capacity of the bacterial strain from which the protein had originated. However, the putative co-operative role in adherence of other non-covalently attached cell wall components (e.g. Slp-associated proteins), removed during the preparation of CWF, cannot be completely excluded. On the other hand, the finding that the S-layer protein of the poorly adherent strain GRL 1117 bound highly efficiently to IPEC-1 cells suggests that some component(s) on GRL 1117 shield(s) the S-layer proteins, preventing them from interacting with IPEC-1 cells. An analogous phenomenon has been observed in Lactobacillus rhamnosus GG: the exopolysaccharide component shields the mucus-binding fimbriae, reducing the adhesive capacity of the strain for mucus [75,76]. Genes putatively participating in exopolysaccharide synthesis have been identified in all of the L. amylovorus strains studied (unpublished), but so far no biochemical evidence of their presence has been described.