Differential crosstalk between epithelial cells, dendritic cells and bacteria in a co-culture model

International Journal of Food Microbiology 131 (2009) 40–51

Contents lists available at ScienceDirect

International Journal of Food Microbiology j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / i j f o o d m i c r o

Differential crosstalk between epithelial cells, dendritic cells and bacteria in a co-culture model Georgia Zoumpopoulou a,b, Effie Tsakalidou b, Joelle Dewulf a, Bruno Pot a, Corinne Grangette a,⁎ a b

Laboratory of Lactic Acid Bacteria and Mucosal Immunity, Institut Pasteur of Lille, 1 rue du Prof. Calmette, 59019 Lille, France Laboratory of Dairy Research, Department of Food Science and Technology, Agricultural University of Athens, Iera Odos 75, 11855 Athens, Greece

a r t i c l e

i n f o

Keywords: Lactic acid bacteria Probiotics E. coli Intestinal epithelial cells m-ICcl2 Dendritic cells Chemokines Cytokines

a b s t r a c t Intestinal epithelial cells (IECs) provide a primary physical barrier against commensal and pathogenic bacteria, but the influence of IECs in the regulation of the associated mucosal immune system remains largely unknown. The network of dendritic cells (DCs) in the vicinity of IECs is known to play a crucial role in the regulation of gut homeostasis. We investigated the cross-talk between murine IECs (m-ICcl2 cell line), bone marrow derived DCs and different bacteria using an in vitro Transwell® co-culture model. IECs responded poorly to different Gram-positive lactic acid bacteria (LAB) and to a Staphylococcus aureus strain. In contrast two Escherichia coli strains, including the probiotic strain Nissle 1917, strongly activated IECs, as evidenced by the induction of different chemokines. While a differential maturation of DCs is observed upon direct stimulation with the various bacteria, DC maturation across the epithelial barrier was only observed upon challenge of the apical surface of the IECs with both E. coli strains and LPS. These results suggested that the capacity of bacteria to induce pro-inflammatory signals through the epithelial barrier is not linked to their pathogenic or commensal status, but seem to be dependent on the presence of specific surface factors. As already reported, we confirmed that m-ICcl2 cells are highly susceptible to LPS. It is highly possible, at least in this model, that free LPS is the “specific factor” key to activate IEC or BMDC. Moreover, IECs are broadly unresponsive to Gram-positive bacterial components, notably TLR-2 ligands, in contrast to Gram-negative bacterial components. These results suggest that the gut epithelium will sense the commensal bacteria in a different way, and seems to be unresponsive to Gram positive bacteria in particular to LAB. © 2009 Elsevier B.V. All rights reserved.

1. Introduction The intestine is a highly sophisticated organ that contains up to 1000 different species of bacteria and has to discriminate between harmful bacteria and commensals. Numerous physical and cellular factors contribute to maintain homeostasis and breakdown of tolerance against the commensal microflora is believed to be a major factor in the pathogenesis of inflammatory bowel disease (IBD) (Sartor, 1997). It is now well accepted that homeostasis versus chronic intestinal inflammation is determined by the presence or absence of appropriate control mechanisms that could be linked to a balance between protective and virulent luminal bacteria. Pursuing this equilibrium, therapeutic approaches which modulate the local microenvironment using probiotics, have been tested both in animal models of colitis (Madsen et al., 1999) as well as in human inflammatory bowel disease (IBD) (Gionchetti et al., 2000; Bibiloni et al., 2005). Wider therapeutic applications are hampered by the limited understanding of the mechanisms by which health beneficial bacteria exert their immune-regulatory effects. We and others have shown that lactobacilli display strain-specific immunemodulation capacities in vitro (Christensen et al., 2002) that have been ⁎ Corresponding author. E-mail address: [email protected] (C. Grangette). 0168-1605/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.ijfoodmicro.2008.12.037

correlated to in vivo anti-inflammatory properties (Foligne et al., 2007a). Recent work reported that selected probiotics were able to protect mice by inducing IL-10 and IL-10 dependent TGFβ-bearing regulatory cells (Di Giacinto et al., 2005). Such regulatory T cells seem to be driven by the modulation of dendritic cell (DC) function (Braat et al., 2004; Smits et al., 2005). Recently, we demonstrated that protective lactobacilli were able to prime regulatory DCs, which conferred a protective effect to mice in a TNBS-induced colitis model (Foligne et al., 2007b). Commensal bacteria seem to prevent or attenuate inflammation by regulating the NFκB activation pathway (Kelly et al., 2004; Petrof et al., 2004). The recognition of the commensal flora by Toll-like receptors is required for intestinal homeostasis and the prevention of epithelial injury, while deregulated interaction between commensal bacteria and TLRs may promote chronic inflammation and tissue damage (Rakoff-Nahoum et al., 2004). The gut epithelium is critically involved in this process, acting as a physical barrier that separates immune cells from luminal bacteria. Recently it has been shown that a NFκB signalling defect in intestinal epithelial cells disrupts immune homeostasis, leading to severe chronic inflammation in mice (Nenci et al., 2007). DCs are generally found in the vicinity of the epithelial barrier and are able to sample luminal antigens by extruding dendrites between epithelial cells (Rescigno et al., 2001). This ‘translocation’ leads either to a protective immune response or to the induction of tolerance, depending on the molecules

G. Zoumpopoulou et al. / International Journal of Food Microbiology 131 (2009) 40–51 Table 1 Bacterial strains used in the present study and their origin. Bacterial species

Strain designation Origin/reference

Lactobacillus acidophilus Lactobacillus fermentum Lactobacillus rhamnosus Lactobacillus salivarius Lactococcus lactis Streptococcus macedonicus Escherichia coli Escherichia coli Staphylococcus aureus

NCFM ACA-DC 179 Lr32 Ls33 MG1363 ACA-DC 198 TG1 Nissle IPL 126

Commercial probiotic stain Kasseri cheese Commercial strain Commercial stain Cheese starter (Gasson, 1983) Kasseri cheese Commensal strain (Sambrook et al., 1989) Probiotic strain (Kruis et al., 1997) IPL collection

encountered and the micro-environmental conditions. The maturation state of DCs within the gastrointestinal tract is currently regarded as a crucial factor governing the type and direction of a mucosal immunological response. Mucosal DCs appear to have unique properties distinguishing them from peripheral DCs (Kelsall and Rescigno, 2004). Interestingly, recent reports showed that this mucosal phenotype could be induced in vitro in the presence of polarized epithelial cells, suggesting that the mucosa environment can educate DCs to mount non-inflammatory responses which maintain gut immune homeostasis (Butler et al., 2006; Rimoldi et al., 2005b). Several reports have shown that non pathogenic commensal bacteria, including probiotics, have the ability to elicit a characteristic cytokine response in in vitro co-culture models of leucocytesensitized epithelial cells and can deliver a distinct signal to underlying immunocompetent cells. Notably, monocytes acquire immunoregulatory functions in the presence of probiotic-activated IEC (Haller et al., 2002; Parlesak et al., 2004). It is therefore interesting to evaluate whether probiotics can modulate DC function through epithelial barrier interaction. As mentioned above we previously showed that probiotic strains have distinct in vitro strain-specific immunomodulatory capacities that closely correlate with their in vivo anti-inflammatory potential (Foligne et al., 2007a). Knowing that selected probiotic strains were able to induce in vitro tolerogenic DCs able to rescue mice from colitis upon adoptive transfer (Foligne et al., 2007b), we investigated further how selected lactic acid bacteria, in comparison to non-pathogenic E. coli strains and to S. aureus, could impact on DC maturation, across a murine epithelial cell line barrier, following controlled apical interaction.

2. Materials and methods 2.1. Bacterial strains, culture conditions and reagents Bacterial strains used in this study are listed in Table 1. Lactobacilli and Streptococcus macedonicus ACA-DC 198 were grown at 37 °C in MRS broth (Difco, Detroit, USA). Lactococcus lactis MG1363 was grown at 30 °C in M17 broth (Difco) supplemented with 0.5% (w/v) glucose. E. coli TG1 and E. coli Nissle were grown at 37 °C in Luria broth, while Staphylococcus aureus was grown in brain heart infusion medium (Difco). Bacteria were grown overnight, harvested by centrifugation (10 min at 4000 ×g), washed twice with PBS buffer (pH 7.2) and resuspended in the same buffer containing 20% (v/v) glycerol to a final concentration of 1 × 109 cfu/ml. Suspensions were stored at −80 °C until used for stimulation assays.

41

LPS from E. coli serotype 0111:B4 was purchased from Sigma (St. Louis, USA) and used at 10 ng/ml final concentration. 2.2. Animals Animal experiments were performed at the accredited animal facility of the Pasteur Institute of Lille (number A59107, Lille, France) according to the Guidelines of Laboratory Animal Care, published by the French Ethical Committee and the rules of the European Union Normatives (number 86/ 609/EEC). BALB/c mice (female, 7–8 weeks old) purchased from Iffa Credo (St Germain sur l'Arbresle, France), were group-housed (8–10 mice/cage) and had free access to regular rodent chow and tap water. 2.3. Generation of bone marrow-derived dendritic cells (BMDCs) Dendritic cells were generated from bone marrow (BM) progenitor cells isolated from femurs and tibias of BALB/c mice as described by Lutz et al. (1999), with minor modifications, as previously reported (Foligne et al., 2007b). Briefly, after red cell lysis (ammonium chloride 0.14 M, pH 7.2), BM cells were cultured in Petri dishes at 2 × 105 cells/ml in complete Iscove's Modified Dulbecco's Medium (IMDM, Sigma, St. Louis, USA) supplemented with 10% (v/v) heat-inactivated fetal calf serum (FCS, Gibco-BRL, Paisley, Scotland), 50 µM 2-mercaptoethanol, 1 mM glutamine, 50 µg/ml gentamycin and 10% of supernatant from a granulocyte-macrophage colony-stimulating factor-expressing cell line (GM-CSF transfected J588 myeloma cell line). Freshly prepared medium was added every three days and BMDCs were used on day 11 of culture (maximum of CD11c expression as checked by FACS analysis). 2.4. Preparation of epithelial cell monolayers and co-culture The murine intestinal epithelial cell (IECs) line m-ICcl2 (Bens et al., 1996) kindly provided by A. Vandewalle (Inserm U246, Paris, France) was maintained at 37 °C in a 5% CO2/95% air atmosphere in HAMF'12/ DMEM (Gibco, NY, USA, v/v) containing the following reagents (Sigma): insulin (5 µg/ml), dexamethasone (5× 10− 8 M), selenium (60 nM), transferrin (5 µg/ml), triiodothyronine (10− 9 M), EGF (10 ng/ml), hepes 20 mM, glutamine 2 mM, D-glucose (0.22%) and fetal calf serum (2%). For transwell® cultures, cells were seeded (2.5 × 105 cells/well) on a 6well format cell culture insert (3-µM nucleopore size; Costar, Corning Inc., USA). The polycarbonate membranes were pre-treated with rat tail collagen (40 µg/ml in 60% ethanol until complete evaporation). Medium was changed every two days and cells were cultured for 12–13 days to achieve fully differentiated monolayers. Transepithelial electrical resistance (TEER) was determined continuously using a Millicell-ERS voltohmmeter (Millipore, MA). For co-culture experiments, inserts containing fully differentiated m-ICcl2 monolayers were transferred to six well plates containing BMDCs (2× 106 cells/well in 3 ml complete IMDM) or IMDM medium (stimulation alone). In all conditions, gentamicin was added at a final concentration of 150 µg/ml to avoid bacterial overgrowth. In such conditions no change in TEER was noticed after bacterial challenge that ranged between 799 and 893 Ω cm2. 2.5. Cells stimulation The apical surface of m-ICcl2 monolayers was challenged (or not) by addition of 1 × 107 cfu of bacteria or 10 ng/ml LPS in the upper chamber

Fig. 1. Experimental model of co-culture for assessing the crosstalk between bacteria, epithelial and dendritic cells. The murine m-ICcl2 epithelial cell line was differentiated in the upper compartment of a polycarbonate transwell system. IECs were then challenged apically with different bacteria or LPS, either in the absence (A) or presence of BMDCs (B) in the basolateral compartment. BMDCs were also directly stimulated with bacteria (C) or LPS. In all conditions, gentamycin was added at the final concentration of 150 µg/ml.

42

G. Zoumpopoulou et al. / International Journal of Food Microbiology 131 (2009) 40–51

Table 2 PCR primers sequence, size of amplicons, annealing temperature and cycle number used for qualitative RT-PCR. Genes

Primers

PCR fragment (bp)

Annealing temperature (°C)

Cycle number

β-actin

F: 5′-GTGGGCCGCCCTAGGCACCA-3′ R: 5-CTCTTTGATGTCACGCACGATTTC-3′ F: 5′-GTGACAACCACGGCCTTCCCTACT-3′ R: 5′-GGTAGCTATGGTACTCCA-3′ F: 5′-GACCCTGCCCATTGAACTGGC-3′ R: 5′-CAACGTTGCATCCTAGGATCG-3′ F: 5′-TGCTGGATTGCAGAGCAGTAA-3′ R: 5′-CTGGAGGAGTTGGCTGAGTC-3′ F: 5′-AGCCCACGTCGTAGCAAACCACCAA-3′ R: 5′-ACACCCATTCCCTTCACAGAGCAAT-3′ F: 5′-ACAGGGCTTTCGATTCAGCGC-3′ R: 5′-CGGGTTGTGTTGGTTGTAGA-3′ F: 5′-ACTGCCCTTGCTGTTCTTCTCTGT-3′ R: 5′-GTCCCTCGATGTGGCTACTTG-3′ F: 5′-ATCCAGAGCTTGAGTGTGACG-3′ R: 5′-ATCAGGTACGATCCAGGCTTCC-3′ F: 5′-TGGTGGAAAAACCTCGTCCA-3′ R: 5′-TGGCTGTTTTGGTAGGCTGTG-3′

540

60

25

313

60

25

415

60

25

410

60

35

446

60

25

403

60

25

402

60

25

253

60

25

252

50

25

IL-6 IL-12 IL-23 TNF-α TGF-β MIP1α/CCL3 MIP-2/CXCL2 COX2

of each well, with or without BMDCs in the lower chamber (Fig.1A and B, respectively). Finally, BMDCs (2 × 106 cells/well in 3 ml complete IMDM) were directly challenged by addition (or not) of the bacteria at 1 × 107 cfu/ well or LPS (10 ng/ml) (Fig. 1C). Gentamycin (150 µg/ml) was added to all conditions tested in order to avoid bacterial overgrowth. All plates were incubated at 37 °C in a 5% CO2/95% air atmosphere for 4 h or 24 h. After 4 h of incubation, the medium in all compartments was removed and cells (IECs and BMDCs) were lysed separately by the addition of 700 µl of RA1 lysis buffer (Nucleospin RNA II kit, MachereyNagel, Düren, Germany), containing 10% β-mercaptoethanol. Lysates were stored at − 80 °C until RNA extraction. After 24 h of incubation, culture supernatants were collected from the basal compartment, clarified by centrifugation and stored at − 20 °C for cytokine and chemokine analysis by ELISA. BMDCs were collected for fluorescenceactivated cell-sorting (FACS) analysis. 2.6. LAL assay The chromogenic Limulus amebocyte lysate (LAL) assay was used to determine the level of endotoxin according to the instructions of the manufacturer (Cape Cod, Inc., USA). Different dilutions of the apical and baso-lateral supernatants (performed with endotoxin-free water) were mixed to equal volumes of the LAL reagent and the mixture was incubated for at least 10 min at 37 °C. The reaction was stopped by adding 25% acetic acid, and the absorbance was read at 405 nm. Endotoxin levels were quantified from standard curves and expressed in EU/ml. 2.7. Immunocytostaining and flow cytometry For FACS analysis (expression of surface markers), BMDCs were collected, centrifuged (300 ×g, 10 min, 4 °C), and resuspended in cold PBS containing 1% (v/v) heat-inactivated FCS, 0.1% (w/v) sodium azide (PBS-FCS-Az) and Fc receptor-blocking mAb anti-CD32 (2.4G2) (vol/ vol). The following Abs were used for staining (all purchased from BD Pharmingen): FITC-conjugated anti-mouse CD11c (HL3); PE-conjugated Table 3 Endotoxin level in the apical and basolateral compartment of the co-culture upon E. coli and LPS stimulation. Bacterial species

Endotoxin level in the apical compartment (EU/ml)

Endotoxin level in the basal compartment (EU/ml)

Escherichia coli TG1 Escherichia coli Nissle LPS

266 247 266

172 167 22

anti-mouse CD86 (GL1); PE-conjugated anti-mouse CD40 (3/23) and appropriate mAb isotypic controls. Cells were incubated with selected mAb for 30 min on ice and at low light exposure. Thereafter, cells were washed with 3 mL PBS-FBS-Az and finally resuspended in 300 µl paraformaldehyde 1% for flow cytometric analysis using a FACS-Calibur flow cytometer and CellQuest software (BD Biosciences, San Jose, CA). 2.8. Cytokine and chemokine quantification in culture supernatants Murine IL-2, IL-6, IL-10, IL-12(p70) and TNF-α were analyzed using the matching Ab pairs purchased from BD Pharmingen (BD Biosciences, San Jose, CA). MIP-1α/CCL3, MIP-2/CXCL2 and MIP-3α/CCL20 were similarly analyzed using commercially available ELISA kits (R&D systems, Minneapolis, MN) following the manufacturer's instructions. 2.9. Reverse-transcriptase (RT) reaction and polymerase chain reaction (PCR) Total RNA was extracted from lysed cells (IECs and BMDCs), obtained after 4 h of stimulation with bacteria as described previously by using the NucleoSpin RNA II kit (Macherey-Nagel) and a DNase treatment according to the manufacturer's recommendations. A 2-µg aliquot of total RNA was reverse-transcribed using random primers (Amersham, Piscataway, NJ, USA) and Superscript II (Invitrogen, Carlsbad, CA, USA). cDNA (5 µl of RT reaction) and cDNA were submitted to PCR amplification in a PTC-100 thermocycler (MJ Research Inc, USA) with a total volume of 25 µl and using Taq polymerase (Promega) and specific primers. Reactions were heat-denatured for 3 min at 94 °C and then amplified with optimized number of PCR cycles (see Table 2), each comprising successive incubations at 94 °C for 1 min, at annealing temperature optimized for each primer pair (see Table 2) for 1 min and at 72 °C for 1 min. The housekeeping gene of β-actin was used as a control. PCR products were subjected to electrophoresis on 1.5% agarose gels and visualized by staining with ethidium bromide. 2.10. Real time RT-PCR (qRT-PCR) Fifty nanograms of template (cDNA as described above) were used in a real-time PCR reaction with a final volume of 25 µl, using the Taq-Man PCR Master Mix (Applied Biosystems, Branchburg, NJ, USA) and the primers and probes designed by Applied-Biosystems (Assays-on-demand) for murine Cox-2/Ptgs2 (Mn00478374), MIP-2/CxCl2 (Mm00436450), MIP1α/CCL3 (Mm00441258), MIP-3α/CCL20 (Mm00444422), and TGFβ (Mm004441724), as suggested by the manufacturer. All reactions were performed in duplicate. The thermal cycling conditions were 10 min at 95 °C, followed by 40 cycles of 15 sec at 95 °C and 1 min at 60 °C, using the Applied Biosystems 7300 real time PCR system. For quantification, we

G. Zoumpopoulou et al. / International Journal of Food Microbiology 131 (2009) 40–51

43

Fig. 2. IECs are unresponsive to Gram-positive bacteria while E. coli strains induce high production of chemokines, either in the absence (A) or presence (B) of 106 BMDCs in the lower compartment. IECs were apically challenged by addition of 1 × 107 cfu of bacteria or 10 ng/ml LPS and after 24 h incubation, cell culture supernatants were harvested from the basal compartment for chemokine measurements (MIP-1α/CCL3, MIP-2/CXCL2 and MIP-3α/CCL20) by ELISA.

compared the amount of target normalized to the β-actin RNA (Mn00607939) amplification by using the 2−ΔΔCt formula representing the n-fold differential expression of the target gene in the treated samples (bacteria-stimulated cells), as compared to the control samples (cells not treated). Ct is the mean of threshold cycle, ΔCt is the difference in the Ct values for the target gene and the reference gene (for each sample), and ΔΔCt represents the difference between the Ct from the control and each sample. Non-Template Control (NTC) and Reverse-Transcription Control (RTC) were included for each quantification experiment.

3. Results 3.1. IECs are poorly responsive to lactic acid bacteria in comparison to E. coli strains The capacity of the different bacterial strains to stimulate IECs in the presence (co-culture Transwell® model) or absence of BMDCs was investigated. Bacterial strains tested (see Table 1) included strains belonging to different lactic acid bacteria strains and species previously

44

G. Zoumpopoulou et al. / International Journal of Food Microbiology 131 (2009) 40–51

G. Zoumpopoulou et al. / International Journal of Food Microbiology 131 (2009) 40–51

45

Fig. 4. The expression of genes encoding chemokines (MIP-2/CXCL2 and MIP-3a/CCL20) by IECs is only enhanced upon apical challenge of E. coli strains, while TGF-β gene expression was not up-regulated by any bacterial strain. IECs were apically challenged by addition of 1 × 107 cfu of bacteria in the absence (A) or presence (B) of 106 BMDCs in the lower compartment. After 4 h of bacterial stimulation, total RNA was extracted from IECs and gene expression was analyzed by real time RT-PCR using TaqMan probes. β-actin was used as reference gene. Each reaction was performed in duplicate.

shown to exhibit (L. rhamnosus Lr32, L. salivarius Ls33, L. fermentum ACADC 179) or not (L. acidophilus NCFM, Lc. lactis MG1363, S. macedonicus ACA-DC 198), an in vitro and in vivo anti-inflammatory profile (Foligne et al., 2007a; Zoumpopoulou et al., 2008). In addition, two Gram-negative bacteria were studied, namely a non-pathogenic E. coli strain (TG1), previously shown to exhibit in vitro a pro-inflammatory potential (Foligne et al., 2007a), as well as a known probiotic E. coli strain (strain Nissle 1917) (Kruis et al., 1997). Lastly, a pathogenic strain of S. aureus (IPL 126) was added. We also included LPS stimulation as control. We compared the concentration of endotoxin released after 24 h stimulation with the two E. coli strains and with LPS, both in the apical and basal

compartments (Table 3). The level of endotoxin released in the apical compartments was similar upon E. coli strains and LPS stimulation (247– 266 EU/ml), while more endotoxin was released in the basal compartment upon E. coli stimulation (172 and 167 EU/ml for TG1 and Nissle, respectively), in comparison to LPS stimulation (22 EU/ml), indicating that LPS/endotoxin could translocate through the epithelial monolayer and that this translocation was higher with the bacteria. No detectable endotoxin was found in co-cultures performed with Gram (+) bacteria (data not shown). In all experiments, the bacterial strains tested were added at the apical surface of the IECs in the presence of gentamicin (150 µg/ml) in

Fig. 3. Cytokine and chemokine production is only observed in the co-culture model upon E. coli challenge. IECs were apically challenged by addition of 1 × 107 cfu of bacteria or 10 ng/ml LPS in the presence of 106 BMDCs in the lower compartment. Cell culture supernatants were harvested from the basal compartment after 24 h incubation, for measurement of cytokines (IL-2, IL-6, IL-10, IL-12, TNF-α) and chemokines (MIP-1α/CCL3, MIP-2/CXCL2) by ELISA.

46

G. Zoumpopoulou et al. / International Journal of Food Microbiology 131 (2009) 40–51

Fig. 5. Differential cytokine and chemokine production is observed when BMDCs (106 cells) were directly challenged with bacteria (1 × 107 cfu) or LPS (10 ng/ml). Cell culture supernatants were harvested from the basal compartment after 24 h incubation, for measurement of cytokines (IL-2, IL-6, IL-10, IL-12, TNF-α) and chemokines (MIP-1α/CCL3, MIP-2/CXCL2) by ELISA.

G. Zoumpopoulou et al. / International Journal of Food Microbiology 131 (2009) 40–51

order to avoid bacterial overgrowth. In such condition, no change in TEER was noticed as reported in the Material and Methods section. Moreover, we also checked using two non pathogenic strains of L. rhamnosus and L. acidophilus that no bacterial translocation appeared after 5 h of co-culture in the absence of gentamycin, in order to be sure about the integrity of the epithelial monolayer (data not shown). Cytokine and chemokine productions were measured by ELISA in the basal compartments as preliminary experiments indicated similar profiles in the apical medium but predominant intensity in the basal compartment (data not shown). When IECs were stimulated in the absence of BMDCs, the production of two of the three chemokines tested, namely MIP-2/CXCL2 and MIP-3α/CCL20 seemed to be enhanced only upon stimulation with E. coli TG1, E. coli Nissle and LPS (Fig. 2A). These two strains as well as LPS were also able to induce additionally MIP-1α production in co-culture with BMDCs (Fig. 2B). IECs did not produce cytokines in the absence of BMDCs (independent of the stains tested; data not shown), while in the presence of BMDCs (co-culture system), the production of all cytokines tested was largely enhanced following the interaction with the E. coli strains, as opposed to the LAB or the S. aureus strains (Fig. 3). These results were corroborated by qualitative RT-PCR analysis that revealed no gene expression in the IECs for most of the cytokines tested (IL-6, IL-10, IL-12, TNF-α), while in contrast to the LAB and S. aureus, the expression of genes coding for MIP-2/CXCL2 and MIP-3α/CCL20 was largely enhanced after E. coli stimulation (data not shown). No expression of MIP-1α/CCL3 was detected in bacteria-treated IECs in the absence of BMDCs (data not shown), indicating that the production of this cytokine observed in co-culture experiments was induced by BMDCs through the epithelial monolayer. A constitutive expression of TGF-β was also observed, which was not modulated by bacterial challenges (data not shown). These observations were again confirmed by qRT-PCR experiments, which showed a great over-expression of the genes coding for MIP-2/CXCL2 and MIP-3α/CCL20 in IECs which were challenged with E. coli only, in the presence (Fig. 4A) or absence (Fig. 4B) of BMDCs. No significant over-expression of TGF-β was measured in treated cells. 3.2. LAB differentially activated BMDCs upon direct interaction while only E. coli strains were able to activate BMDCs through the IEC monolayer We previously showed that LAB exhibit differential immunomodulation effects both on PBMC (Foligne et al., 2007a) or BMDCs (Foligne et al., 2007b). In the present work, we extended that study to the selected LAB strains, the two E. coli and the S. aureus strains and to LPS. As illustrated in Fig. 5, the three strains Lr32, Ls33 and ACA-DC 179, previously shown to exhibit considerable in vivo and in vitro antiinflammatory potential, were found to be the lowest inducers of almost all cytokines and chemokines tested, as compared to the other LAB stains (NCFM, MG1363 and ACA-DC 198). Remarkably, the Lc. lactis strain (MG1363) exhibited the highest capacity to induce the cytokines or chemokines tested, except IL-2 that was mainly induced by E. coli strains. Interestingly, for all cytokines and chemokines tested, the probiotic E. coli (strain Nissle 1917) was not found to have a different impact on the activation of BMDCs, as compared to the non-pathogenic E. coli (strain TG1). As expected, LPS was able to induce the production of all the cytokines and chemokines. We also examined the ability of the bacterial strains to activate DCs through the epithelial monolayer (co-culture system), as compared to direct stimulation. We first verified by cytometric analysis the capacity of the bacteria to activate the co-stimulatory functions in BMDC. As expected, when BMDCs were directly stimulated with selected bacteria or LPS, different levels of maturation were observed (Fig. 6A). The three anti-inflammatory LAB (strains Lr32, Ls33 and ACA-DC 179) induced very low expression levels of both CD40 and CD86, as compared to untreated DCs, while strong up-regulation was

47

observed when BMDCs were stimulated with the non anti-inflammatory LAB strains (NCFM, MG1363, ACA-DC 198), the two E. coli strains TG1 and Nissle as well as LPS. Surprisingly, only partial maturation was observed when BMDCs were stimulated with S. aureus. Interestingly, when bacterial stimulation was performed at the apical surface of the epithelial monolayer (co-culture system), the upregulation of the two co-stimulatory molecules (CD40 and CD86) was enhanced only with the E. coli strains and LPS, while no BMDC activation was detected upon apical challenge of the IECs with any of the Gram positive bacteria (Fig. 6B). In BMDCs stimulated either directly or through the epithelial cells, we furthermore evaluated the expression of a number of genes encoding pro-inflammatory or regulatory mediators or factors involved in cell recruitment (chemokines). Most of the genes were differently up-regulated upon direct bacteria interaction (Fig. 7A). Notably, the expression of IL-23 and IL-6 was enhanced when BMDCs were stimulated with non anti-inflammatory strains (NCFM, MG1363, E. coli TG1 and Nissle 1917) and LPS, while slight up-regulation was observed with Lr32, Ls33 and ACA-DC 179, confirming previous measurements of cytokine and chemokine productions by ELISA (Fig. 5). These results were partially confirmed by qRT-PCR which revealed differential up-regulation of genes encoding MIP-2/ CXCL2 and Cox2/ ptgs2. The lowest gene expression was obtained when BMDCs were directly stimulated with the three anti-inflammatory strains (Lr32, Ls33, ACA-DC 179). No modulation of the expression of TGF-β was observed, in comparison to the unstimulated control. When BMDCs were in coculture with IECs but had no direct contact with bacteria, PCR results were totally different from those obtained after direct stimulation of BMDCs (Fig. 7B). Indeed, under these conditions, only the two E. coli strains and to a lesser extend LPS, were able to strongly activate the expression of all genes investigated, except for TGF-β, that was constitutively expressed and MIP-1α/CCL3 that was also slightly expressed upon challenge with Gram positive bacteria (Fig. 7B). These results confirmed the cytokine and chemokine induction profile observed in the co-culture system upon stimulation with the two E. coli strains and LPS only (Fig. 3). qRT-PCR analysis for MIP-2/CXCL2 and Cox2/ptgs2 genes confirmed these results, showing differential upregulation of the two genes in BMDCs after direct bacterial interaction (Fig. 8A), while only E. coli strains were able to strongly up-regulate the expression of the two genes across the IEC monolayer (Fig. 8B). The expression of TGF-β was not modulated in both conditions. Consequently, when the BMDCs were not in direct contact with the bacteria, only the two Gram-negative E. coli strains and LPS were capable to induce the production of cytokines and chemokines as well as the maturation of the BMDCs. 4. Discussion It has recently been shown that intestinal homeostasis is regulated by a tight crosstalk between epithelial cells and immune cells, in particular with DCs. Notably, recent papers demonstrated that the typical phenotype of mucosal DCs can be obtained after co-culture with polarized epithelial cells, creating a “tolerogenic” environment (Butler et al., 2006; Rimoldi et al., 2005a,b). We previously showed that selected probiotic strains exhibit both in vitro and in vivo antiinflammatory capacities (Foligne et al., 2007a) and are able to induce regulatory DCs which could transfer a protective effect in a murine model of colitis (Foligne et al., 2007b). In the present paper, we developed a reductionist transwell® murine co-culture model mimicking the intestinal barrier to investigate enteric bacteria signaling both on IECs and adjacent DCs. We particularly compared the effect of different probiotic strains (which were found to be very different in their anti-inflammatory properties), with two E. coli strains, including the probiotic strain E. coli Nissle 1917, with a pathogenic S. aureus strain and LPS. We observed that IECs responded differently to the selected bacteria, being poorly responsive to the LAB and S. aureus as compared

48

G. Zoumpopoulou et al. / International Journal of Food Microbiology 131 (2009) 40–51

Fig. 6. LAB differentially activated BMDCs upon direct interaction while only E. coli strains and LPS were able to induce BMDC maturation through the IEC monolayer. BMDCs (106 cells) were challenged by addition of 1 × 107 cfu of bacteria or 10 ng/ml LPS, directly (A) or indirectly through the epithelial monolayer (B). After 24 h stimulation, BMDCs were harvested and the expression of co-stimulatory molecules (CD40 and CD86) was quantified by FACs analysis. Plots show flow cytometric profiles of CD40 and CD86 (thick lines) in comparison to isotypic control labeling (thin lines).

G. Zoumpopoulou et al. / International Journal of Food Microbiology 131 (2009) 40–51

49

Fig. 7. Differential gene expression is observed when BMDCs (106 cells) are directly challenged with bacteria (1 × 107 cfu), while only the E. coli strains and LPS could up-regulate proinflammatory genes across the IEC monolayer. BMDCs (106 cells) were challenged by addition of 1 × 107 cfu of bacteria or 10 ng/ml LPS, directly (A) or indirectly through the epithelial monolayer (B). After 4 h of bacterial stimulation, total RNA was extracted from BMDCs and gene expression was analyzed by qualitative RT-PCR.

to the E. coli strains and LPS which strongly induced chemokines (MIP2/CXCL2 and MIP-3α/CCL20) upon apical challenge of the IECs. Moreover, when BMDCs were added to the basolateral compartment, MIP-1α/ CCL3 and the pro-inflammatory/ Th1 cytokines IL-2, IL-6, IL12, TNF-α were induced in the co-culture only upon E. coli stimulation, suggesting that only E. coli was able to activate DCs across the epithelial barrier. We therefore analyzed the DC activation, comparing direct and indirect (through the IECs) stimulation. We could confirm previous results which showed differential activation of DCs upon direct interaction with LAB (Foligne et al., 2007b), since a partial maturation was obtained with the three LAB strains (Lr32, Ls33, ACADC 179) exhibiting anti-inflammatory capacities. These three strains could hardly or not at all trigger the expression of cytokines and chemokines, of co-stimulatory molecules and pro-inflammatory genes, while the other LAB strains, in particular Lc. lactis MG1363 as well as the two E. coli strains and LPS, did induce maturation of BMDCs. Surprisingly, S. aureus induced only partial DC maturation. When evaluating indirect interactions with DCs through the epithelial monolayer, only the E. coli strains and LPS were able to maturate DCs, as evidenced by a strong upregulation of co-stimulatory molecules and strong induction of all genes investigated in BMDCs. These results are in agreement with previous work showing that a Bifidobacterium infantis and a L. salivarius strain did not induce proinflammatory response in human IECs as compared to Salmonella typhimurium, suggesting that IECs display immunological unresponsiveness when exposed to LAB (O'Hara et al., 2006). Using a co-culture model of CaCO-2 IECs and PBMC, Haller et al. (2000, 2002) observed also discriminative IEC activation between E. coli and LAB strains (L. johnsonii or L. gasseri). Rimoldi et al. (2005a,b) reported in a human co-culture system, that the release of pro-inflammatory mediators by IECs in response to bacteria is dependent on bacterial invasiveness and on the presence of flagella. Indeed, a non-flagellated E. coli was not able to induce chemokine production, nor were L. plantarum and Bacillus subtilis strains. Similar results were reported by Bambou et al. (2004)

demonstrating that flagellin is required to induce a pro-inflammatory response in enterocytes, both in vitro and in vivo. By transcriptomic analysis using microarrays, E. coli Nissle 1917 was indeed shown to upregulate a certain number of genes in IECs, including chemokines such as MCP-1 and MIP-2α (Ukena et al., 2005). Although this probiotic strain has been successfully used for the treatment of IBD (Kruis et al., 1997), the precise mechanism remains unclear. It has been recently suggested that the strain Nissle 1917 enhances mucosal integrity by mediating upregulation of tight junction proteins (Ukena et al., 2007) or by stimulating the expression of beta-defensins through a flagellin stimulatorydependent effect (Schlee et al., 2007). It can also be speculated that the strain will adapt to the gut environment in vivo, leading to a modulation of regulator genes, possibly including the ones controlling flagellin expression (Giraud et al., 2008). The epithelial cell line used in this study (m-ICcl2) was previously shown to be highly susceptible to LPS, notably because it expresses CD14, TLR4 and MD-2 (Hornef et al., 2003). Moreover, LPS seems to be able to translocate through the epithelial monolayer, as previously reported by Haller et al. (2004). Indeed, we observed similar levels of endotoxin in the apical compartment after E. coli and LPS stimulation as well as significant levels in all the basal compartments. The fact that greater levels were detected with both E. coli strains could possibly explain the differences of chemokine and cytokines concentrations detected. Conversely, as reported by Gewirtz et al. (2001), the expression of TLR5, the receptor for flagellin seems to be restricted to the basolateral surface membrane of polarized IECs. In the m-ICcl2 cell line, a strong chemokine induction has been reported in response to LPS while this epithelial cell line did not respond to other TLRs agonists such as CpG, Pam3Cys-SK4 and S. enterica flagellin (Sterzenbach et al., 2007). IECs, and in particular m-ICcl2, seem to respond poorly to Gram-positive bacterial cell wall components and to a variety of TLR-2 specific ligands, suggesting that the intestinal epithelium has developed a careful system by which only selected bacterial pathogens induce an inflammatory response and not the non-pathogenic commensal organisms (Melmed

50

G. Zoumpopoulou et al. / International Journal of Food Microbiology 131 (2009) 40–51

Fig. 8. Differential chemokine gene expression is observed when BMDCs (106 cells) are challenged directly (A) with bacteria (1 × 107 cfu), while only E. coli strains were able to upregulate these genes across the epithelial monolayer. BMDCs (106 cells) were challenged by addition of 1 × 107 cfu of bacteria, directly (A) or indirectly through the epithelial monolayer (B). After 4 h of bacterial stimulation, total RNA was extracted from BMDCs and gene expression (TGF-β, CXCL2 and Cox-2) was analyzed by real time RT-PCR using TaqMan probes. β-actin was used as reference gene. Each reaction was performed in duplicate.

et al., 2003). This could explain the unresponsiveness of IECs and the absence of consequent activation of adjacent DCs to the selected LAB strains as well as to the S. aureus strain. Our results seem to indicate that the capacity of the strain to activate IECs and to deliver proinflammatory signals to adjacent immune cells is not linked to the pathogenic status of the bacteria, but seem to depend on the presence of specific surface factors. It is highly possible, at least in this model, that free LPS is the “specific factor” key by which E. coli strains activate IEC or BMDC. It is possible that the signal induced in the baso-lateral level by whole bacteria could be due to free LPS translocating through the epithelial monolayer. We could also speculate that LPS crosses the

epithelial layer better than Gram (+) microbial associated molecular patterns (MAMPs). Nevertheless, it is often observed (Crane-Godreau and Wira, 2005; Strandberg et al., 2005) that larger concentrations (1– 10 µg/ml) of Gram (+) MAMPs (Peptidoglycan, LTA, lipoproteins…) are generally needed to stimulate epithelial cells while LPS is efficient in a fairly low dose (1–100 ng/ml). Another explanation for the unresponsiveness of Gram (+) bacteria, as suggested by Crane-Godreau and Wira (2005), may be the fact that their strong cell wall is less prone to fragmentation than cell walls from Gram (−) bacteria such as E. coli. Although we previously observed differential DC maturation amongst different LAB as well as the potential of anti-inflammatory LAB strains to

G. Zoumpopoulou et al. / International Journal of Food Microbiology 131 (2009) 40–51

induce regulatory DCs, the absence of DC maturation after indirect stimulation through the apical surface of IECs by gram-positive bacteria in general, and the absence of a differential induction profile between anti-inflammatory and non-anti-inflammatory strains, support the important ‘immunological filtering’ role of the IECs and makes is more difficult to predict whether selected strains will confer or not to DCs the ability to drive T regulatory cell differentiation. Follow-up studies are ongoing to evaluate whether such DCs could indeed exert in vivo a protective effect in a murine model of colitis and verify whether the same strain-specific profile is obtained. Clearly, further knowledge is needed on the control mechanisms that drive gut homeostasis versus chronic inflammation as to better understand this differential responsiveness of the epithelial barrier towards enteric bacteria. Acknowledgements This work was supported by the Pasteur Institute of Lille and the François Aupetit foundation. Georgia Zoumpopoulou was supported by a grant from the European Community from the Marie Curie fellowship program. We are very grateful to DANISCO for supplying the commercial strains. We also thank Stephane Delhaye for excellent technical assistance. We also thank Denise Goudercourt for technical help in performing the LAL assay. References Bambou, J.C., Giraud, A., Menard, S., Begue, B., Rakotobe, S., Heyman, M., Taddei, F., CerfBensussan, Gaboriau-Routhiau, V., 2004. In vitro and ex vivo activation of the TLR5 signaling pathway in intestinal epithelial cells by a commensal Escherichia coli strain. Journal of Biological Chemistry 279, 42984–42992. Bens, M., Bogdanova, A., Cluzeaud, F., Miquerol, L., Kerneis, S., Kraehenbuhl, J.P., Kahn, A., Pringault, E., Vandewalle, A., 1996. Transimmortalized mouse intestinal cells (m-ICc12) that maintain a crypt phenotype. American Journal of Physiology 270, 1666–1674. Bibiloni, R., Fedorak, R.N., Tannock, G.W., Madsen, K.L., Gionchetti, P., Campieri, M., De Simone, C., Sartor, R.B., 2005. VSL3 probiotic-mixture induces remission in patients with active ulcerative colitis. American Journal of Gastroenterology 100, 1539–1546. Braat, H., van den Brande, J., van Tol, E., Hommes, D., Peppelenbosch, M., van Deventer, S., 2004. Lactobacillus rhamnosus induces peripheral hyporesponsiveness in stimulated CD4+ T cells via modulation of dendritic cell function. American Journal of Clinical Nutrition 80, 1618–1625. Butler, M., Ng, C.Y., van Heel, D.A., Lombardi, G., Lechler, R., Playford, R.J., Ghosh, S., 2006. Modulation of dendritic cell phenotype and function in an in vitro model of the intestinal epithelium. European Journal of Immunology 36, 864–874. Christensen, H.R., Frokiaer, H., Pestka, J.J., 2002. Lactobacilli differentially modulate expression of cytokines and maturation surface markers in murine dendritic cells. Journal of Immunology 168, 171–178. Crane-Godreau, M.A., Wira, C.R., 2005. CCL20/macrophage inflammatory protein 3α and tumor necrosis factor alpha production by primary uterine epithelial cells in response to treatment with lipopolysaccharide or Pam3Cys. Infection and Immunity 73, 476–484. Di Giacinto, C., Marinaro, M., Sanchez, M., Strober, W., Boirivant, M., 2005. Probiotics ameliorate recurrent Th1-mediated murine colitis by inducing IL-10 and IL-10-dependent TGF-beta-bearing regulatory cells. Journal of Immunology 174, 3237–3246. Foligne, B., Nutten, S., Grangette, C., Dennin, V., Goudercourt, D., Poiret, S., Dewulf, J., Brassart, D., Mercenier, A., Pot, B., 2007a. Correlation between in vitro and in vivo immunomodulatory properties of lactic acid bacteria. World Journal of Gastroenterology 13, 236–243. Foligne, B., Zoumpopoulou, G., Dewulf, J., Ben Younes, A., Chareyre, F., Sirard, J.C., Pot, B., Grangette, C., 2007b. A key role of dendritic cells in probiotic functionality. PLoS ONE 2, e313. Gasson, M.J., 1983. Plasmid complements of Streptococcus lactis NCDO 712 and other lactic streptococci after protoplast-induced curing. Journal of Bacteriology 154, 1–9. Gewirtz, A.T., Navas, T.A., Lyons, S., Godowski, P.J., Madara, J.L., 2001. Cutting edge: bacterial flagellin activates basolaterally expressed TLR5 to induce epithelial proinflammatory gene expression. Journal of Immunology 167, 1882–1885. Gionchetti, P., Rizzello, F., Venturi, A., Brigidi, P., Matteuzzi, D., Bazzocchi, G., Poggioli, G., Miglioli, M., Campieri, M., 2000. Oral bacteriotherapy as maintenance treatment in patients with chronic pouchitis: a double-blind, placebo-controlled trial. Gastroenterology 119, 305–309. Giraud, A., Arous, S., Paepe, M.D., Gaboriau-Routhiau, V., Bambou, J.C., Rakotobe, S., Lindner, A.B., Taddei, F., Cerf-Bensussan, N., 2008. Dissecting the genetic components of adaptation of Escherichia coli to the mouse gut. PLoS Genetics 4, e2. Haller, D., Bode, C., Hammes, W.P., Pfeifer, A.M., Schiffrin, E.J., Blum, S., 2000. Nonpathogenic bacteria elicit a differential cytokine response by intestinal epithelial cell/leucocyte co-cultures. Gut 47, 79–87. Haller, D., Serrant, P., Peruisseau, G., Bode, C., Hammes, W.P., Schiffrin, E., Blum, S., 2002. IL-10 producing CD14low monocytes inhibit lymphocyte-dependent activation of intestinal epithelial cells by commensal bacteria. Microbiology and Immunology 46, 195–205.

51

Haller, D., Holt, L., Parlesak, A., Zanga, J., Bäuerlein, A., Sartor, R.B., Jobin, C., 2004. Differential effect of immune cells on non-pathogenic Gram-negative bacteriainduced nuclear factor-kB activation and pro-inflammatory gene expression in intestinal epithelial cells. Immunology 112, 310–320. Hornef, M.W., Normark, B.H., Vandewalle, A., Normark, S., 2003. Intracellular recognition of lipopolysaccharide by toll-like receptor 4 in intestinal epithelial cells. Journal of Experimental Medicine 198, 1225–1235. Kelly, D., Campbell, J.I., King, T.P., Grant, G., Jansson, E.A., Coutts, A.G., Pettersson, S., Conway, S., 2004. Commensal anaerobic gut bacteria attenuate inflammation by regulating nuclearcytoplasmic shuttling of PPAR-gamma and RelA. Nature Immunology 5, 104–112. Kelsall, B.L., Rescigno, M., 2004. Mucosal dendritic cells in immunity and inflammation. Nature Immunology 5, 1091–1095. Kruis, W., Schutz, E., Fric, P., Fixa, B., Judmaier, G., Stolte, M., 1997. Double-blind comparison of an oral Escherichia coli preparation and mesalazine in maintaining remission of ulcerative colitis. Alimentary Pharmacology and Therapeutics 11, 853–858. Lutz, M.B., Kukutsch, N., Ogilvie, A.L., Rossner, S., Koch, F., Romani, N., Schuler, G., 1999. An advanced culture method for generating large quantities of highly pure dendritic cells from mouse bone marrow. Journal of Immunological Methods 223, 77–92. Madsen, K.L., Doyle, J.S., Jewell, D., Tavernini, M.M., Fedorak, N., 1999. Lactobacillus species prevents colitis in interleukin 10 gene-deficient mice. Gastroenterology 116, 1107–1114. Melmed, G., Thomas, L.S., Lee, N., Tesfay, S.Y., Lukasek, K., Michelsen, K.S., Zhou, Y., Hu, B., Arditi, M., Abreu, M.T., 2003. Human intestinal epithelial cells are broadly unresponsive to Toll-like receptor 2-dependent bacterial ligands: implications for hostmicrobial interactions in the gut. Journal of Immunology 170, 1406–1415. Nenci, A., Becker, C., Wullaert, A., Gareus, R., van Loo, G., Danese, S., Huth, M., Nikolaev, A., Neufert, C., Madison, B., Gumucio, D., Neurath, M.F., Pasparakis, M., 2007. Epithelial NEMO links innate immunity to chronic intestinal inflammation. Nature 446, 557–561. O'Hara, A.M., O'Regan, P., Fanning, A., O'Mahony, C., Macsharry, J., Lyons, A., Bienenstock, J., O'Mahony, L., Shanahan, F., 2006. Functional modulation of human intestinal epithelial cell responses by Bifidobacterium infantis and Lactobacillus salivarius. Immunology 118, 202–215. Parlesak, A., Haller, D., Brinz, S., Baeuerlein, A., Bode, C., 2004. Modulation of cytokine release by differentiated CACO-2 cells in a compartmentalized coculture model with mononuclear leucocytes and nonpathogenic bacteria. Scandinavian Journal of Immunology 60, 477–485. Petrof, E.O., Kojima, K., Ropeleski, M.J., Musch, M.W., Tao, Y., De Simone, C., Chang, E.B., 2004. Probiotics inhibit nuclear factor-kappaB and induce heat shock proteins in colonic epithelial cells through proteasome inhibition. Gastroenterology 127, 1474–1487. Rakoff-Nahoum, S., Paglino, J., Eslami-Varzaneh, F., Edberg, S., Medzhitov, R., 2004. Recognition of commensal microflora by toll-like receptors is required for intestinal homeostasis. Cell 118, 229–241. Rescigno, M., Urbano, M., Valzasina, B., Francolini, M., Rotta, G., Bonasio, R., Granucci, F., Kraehenbuhl, J.P., Ricciardi-Castagnoli, P., 2001. Dendritic cells express tight junction proteins and penetrate gut epithelial monolayers to sample bacteria. Nature Immunology 2, 361–367. Rimoldi, M., Chieppa, M., Larghi, P., Vulcano, M., Allavena, P., Rescigno, M., 2005a. Monocyte-derived dendritic cells activated by bacteria or by bacteria-stimulated epithelial cells are functionally different. Blood 106, 2818–2826. Rimoldi, M., Chieppa, M., Salucci, V., Avogadri, F., Sonzogni, A., Sampietro, G., Nespoli, M.A., Viale, G., Allavena, P., Rescigno, M., 2005b. Intestinal immune homeostasis is regulated by the crosstalk between epithelial cells and dendritic cells. Nature Immunology 6, 507–514. Sambrook, J., Fritsch, E.F., Maniatis, T., 1989. Molecular Cloning: A Laboratory Manual. Cold Spring Harbor, New York. Sartor, R.B., 1997. The influence of normal microbial flora on the development of chronic mucosal inflammation. Research in Immunology 148, 567–576. Schlee, M., Wehkamp, J., Altenhoefer, A., Oelschlaeger, T.A., Stange, E.F., Fellermann, K., 2007. Induction of human beta-defensin 2 by the probiotic Escherichia coli Nissle 1917 is mediated through flagellin. Infection and Immunity 75, 2399–2407. Smits, H.H., Engering, A., van der Kleij, D., de Jong, E.C., Schipper, K., van Capel, T.M., Zaat, B.A., Yazdanbakhsh, M., Wierenga, E.A., van Kooyk, Y., Kapsenberg, M.L., 2005. Selective probiotic bacteria induce IL-10-producing regulatory T cells in vitro by modulating dendritic cell function through dendritic cell-specific intercellular adhesion molecule 3-grabbing nonintegrin. Journal of Allergy and Clinical Immunology 115, 1260–1267. Sterzenbach, T., Lee, S.K., Brenneke, B., von Goetz, F., Schauer, D.B., Fox, J.G., Suerbaum, S., Josenhans, C., 2007. Inhibitory effect of enterohepatic Helicobacter hepaticus on innate immune responses of mouse intestinal epithelial cells. Infection and Immunity 75, 2717–2728. Strandberg, Y., Gray, C., Vuocolo, T., Donaldson, L., Broadway, M., Tellam, R., 2005. Lipopolysaccharide and lipoteichoic acid induce different innate immune responses in bovine mammary epithelial cells. Cytokine 31, 72–86. Ukena, S.N., Singh, A., Dringenberg, U., Engelhardt, R., Seidler, U., Hansen, W., Bleich, A., Bruder, D., Franzke, A., Rogler, G., Suerbaum, S., Buer, J., Gunzer, F., Westendorf, A.M., 2007. Probiotic Escherichia coli Nissle 1917 inhibits leaky gut by enhancing mucosal integrity. PLoS ONE 2, e1308. Ukena, S.N., Westendorf, A.M., Hansen, W., Rohde, M., Geffers, R., Coldewey, S., Suerbaum, S., Buer, J., Gunzer, F., 2005. The host response to the probiotic Escherichia coli strain Nissle 1917: specific up-regulation of the proinflammatory chemokine MCP-1. BioMed Central Genetics 6, 43. Zoumpopoulou, G., Foligne, B., Christodoulou, K., Grangette, C., Pot, B., Tsakalidou, E., 2008. Lactobacillus fermentum ACA-DC 179 displays probiotic potential in vitro and protects against trinitrobenzene sulfonic acid (TNBS)-induced colitis and Salmonella infection in murine models. International Journal of Food Microbiology 121, 18–26.