Mannose-specific interaction of Lactobacillus plantarum with porcine jejunal epithelium

RESEARCH ARTICLE

Mannose-speci¢c interaction of Lactobacillus plantarum with porcine jejunal epithelium Gabriele Gross1,2, Jan van der Meulen1, Johannes Snel2, Roelof van der Meer2, Michiel Kleerebezem2,3, Theo A. Niewold1,4, Marcel M. Hulst1 & Mari A. Smits1 1

Animal Breeding and Genomics Centre, Animal Sciences Group of Wageningen UR, Lelystad, The Netherlands; 2Health and Safety, NIZO Food Research, Ede, The Netherlands; 3Laboratory of Microbiology, Wageningen University, Dreijenplein, Wageningen, The Netherlands; and 4Nutrition and Health, Katholieke Universiteit Leuven, Kasteelpark Arenberg, Heverlee, Belgium

Correspondence: Mari A. Smits, Animal Breeding and Genomics Centre, Animal Sciences Group of Wageningen UR, PO Box 65, 8200 AB Lelystad, The Netherlands. Tel.: 131 320 238270; fax: 131 320 238050; e-mail: [email protected] Received 27 February 2008; revised 9 June 2008; accepted 5 July 2008. First published online 31 July 2008. DOI:10.1111/j.1574-695X.2008.00466.x Editor: Patrik Bavoil Keywords Lactobacillus plantarum ; host–microorganism interaction; PAP/RegIII; msa/srtA ; gene expression; probiotic.

Abstract Host–microorganism interactions in the intestinal tract are complex, and little is known about specific nonpathogenic microbial factors triggering host responses in the gut. In this study, mannose-specific interactions of Lactobacillus plantarum 299v with jejunal epithelium were investigated using an in situ pig Small Intestinal Segment Perfusion model. The effects of L. plantarum 299v wild-type strain were compared with those of two corresponding mutant strains either lacking the gene encoding for the mannose-specific adhesin (msa) or sortase (srtA; responsible for anchoring of cell surface proteins like Msa to the cell wall). A slight enrichment of the wild-type strain associated with the intestinal surface could be observed after 8 h of perfusion when a mixture of wild-type and msa-mutant strain had been applied. In contrast to the mutant strains, the L. plantarum wild-type strain tended to induce a decrease in jejunal net fluid absorption compared with control conditions. Furthermore, after 8 h of perfusion expression of the host gene encoding pancreatitis-associated protein, a protein with proposed bactericidal properties, was found to be upregulated by the wild-type strain only. These observations suggest a role of Msa in the induction of host responses in the pig intestine.

Introduction The intestinal microbial community contains a complex system of bacterial species, maintaining a variety of interactions with the host via the intestinal epithelium. For example, commensal microorganisms have been recognized to influence the intestinal immune system and to induce the expression of bactericidal proteins that can shape the intestinal microbial community (Hooper et al., 2003; Macpherson & Harris, 2004; Cash & Hooper, 2005). Probiotic lactic acid bacteria are intended to have a beneficial health effect for the host and to improve resistance against intestinal pathogens (Cross, 2002; Reid & Burton, 2002). Currently, transcriptome profiles are increasingly used to gain global views on host responses to bacterial stimuli reflected in gene expression levels. Colonization of germ-free mice with commensal and probiotic bacteria such as Bacteroides thetaiotaomicron and Bifidobacterium longum has been shown to cause FEMS Immunol Med Microbiol 54 (2008) 215–223

substantial changes in gene expression of the host, including the induction of host genes involved in innate immunity, mucosal barrier fortification, and intestinal maturation (Hooper et al., 2001; Sonnenburg et al., 2006). Intestinal carbohydrate moieties have been demonstrated to play an important role in the interaction between numerous bacterial species and their hosts (Neeser et al., 2000; Karlsson, 2001). For example, it has been demonstrated for certain Lactobacillus plantarum strains and type 1-fimbriated pathogens such as Salmonella enterica and enterotoxigenic Escherichia coli that they have the capacity to bind to mannose-containing sugar moieties on the surface of epithelial cells (Wold et al., 1988; Aslanzadeh & Paulissen, 1992; Adlerberth et al., 1996). Lactobacillus plantarum is a member of the human intestinal microbiota and specific strains are marketed as probiotics in functional foods with a proposed beneficial health effect (for a review, see De Vries et al., 2006). Furthermore, the mannosespecific adhesin-encoding gene of L. plantarum, msa, has been 2008 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved


c

216

identified recently (Pretzer et al., 2005). This gene is of interest because it represents a potential probiotic function of L. plantarum that might be relevant for increased resistance against intestinal infections by competitive exclusion of pathogens (Adlerberth et al., 1996; Lee & Puong, 2002). The aim of this study was to gain first insights into the role of mannose-specific interactions of L. plantarum for its capacity to trigger responses of the host epithelial cells. Therefore, the effects of L. plantarum 299v wild-type strain in the intestine were compared with those of two gene-specific mutant strains, lacking either the functional msa or the srtA (encoding sortase) gene. Because Msa belongs to the sortase target proteins of L. plantarum, it cannot be anchored properly to the cell wall in an srtA-deficient strain that concomitantly has lost its mannose-adhesion capacity, analogous to the msa mutant (Pretzer et al., 2005). Bacterial association with the epithelial surface, intestinal net fluid absorption, and the induction of differential gene expression of the host by the wild-type and mutant strains were studied in a pig in situ jejunal loop model, the Small Intestinal Segment Perfusion model (SISP). This model was chosen because of the similarities between pig and human intestinal physiology and anatomy and the possibility to study the very early intestinal response to various food components in situ in a single animal providing an isogenic background under nearly physiological conditions (Bruins et al., 2006; Kiers et al., 2006). Samples were taken after 4 and 8 h of perfusion of jejunal loops with bacterial suspensions, representing an early response to the experimental treatments. The results of this study provide first important insights into the molecular interaction mechanisms of probiotics with the host.

Materials and methods Animals Four weaned male 5-week-old pigs (Duroc  Topigs 20) were obtained from a commercial Dutch piggery (Varkens-

G. Gross et al.

bedrijf Bert Rijnen, Oirschot, The Netherlands). After transport to the experimental facilities, animals were continued to be fed ad libitum with standard liquid pig feed, prepared at the piggery, and were pretreated with colistine (Eurovet Animal Health, Bladel, The Netherlands) for 4 days according to the manufacturers’ recommendations to standardize background microbiota. Six days after arrival, animals were fasted overnight before the SISP experiment was performed. The animal study was approved by the local Animal Ethics Commission in Lelystad, The Netherlands, in accordance with the Dutch Law on Animal Experimentation.

SISP test The SISP test was performed essentially as described by Nabuurs et al. (1993) and Niewold et al. (2005). Briefly, pigs were sedated with azaperone and inhalation anesthesia was applied with a gas mixture of oxygen, nitrous oxide, and isoflurane. The abdominal cavity was opened and five pairs of segments, each 20 cm long and with inlet tubes at the cranial side and outlet tubes at the caudal side of each segment, were prepared at 30–50% of the jejunum. Before further treatment, samples were taken from the jejunum cranially of segment 1, between segments 5 and 6, and caudally of segment 10. The experiment was started by injecting 20 mL of L. plantarum suspensions [wild-type, msa, or srtA mutant; 1010 CFU mL1 in phosphate-buffered saline (PBS)] or PBS without bacteria as a control into the segments, respectively. Salmonella typhimurium (5 mL suspension, 109 CFU mL1 in PBS) was used as a positive control for gene expression analysis, under the same conditions as used in an earlier experiment (Niewold et al., 2007). Figure 1 gives a schematic overview of the treatment and samplings per segment. In the four animals, the order of positions was shuffled per time point to avoid local effects, except for segments treated with

Fig. 1. Schematic overview of treatments per jejunal segment as applied in the SISP test. PBS, control; Lp wt, Lactobacillus plantarum 299v wild-type strain; Lp msa ko, L. plantarum 299v msa-mutant strain; Lp srtA ko, L. plantarum 299v srtA-mutant strain; St, Salmonella typhimurium DT104 (only sampled at time point 8 h as a positive control for gene expression analysis); Lp wt1msa ko, L. plantarum 299v wild-type strain 50%1msa-mutant strain 50% (only sampled at time point 8 h for microbiological analysis). Because pig 3 died unexpectedly during the SISP procedure, sampling after 8 h was not possible for this animal. Segements treated with Salmonella and the mixture of L. plantarum 299v wild type and msa mutant were sampled after 4 h, in contrast to the other three animals.

2008 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved


c

FEMS Immunol Med Microbiol 54 (2008) 215–223

217

Interaction of L. plantarum with pig intestinal epithelium

S. typhimurium or with a 50–50% mixture of L. plantarum wild-type and msa-mutant strain. These treatments were included as a positive control for gene expression and for microbiological analysis only, respectively. Segments sampled at one time point were arranged in local proximity due to differences in gene expression along the intestine. After 1 h, the segments were perfused with perfusion fluid (physiological salt solution with 0.1% glucose and 0.1% amino acids), 2 mL injected every 15 min for 3 or 7 h. Effluent fluid was collected during this time. After a total of 4 h, some segments were removed and samples were taken (see Fig. 1), and after a total of 8 h the remaining segments were sampled and pigs were euthanized. Fluid remaining in each segment was added to the effluent. The length and diameter of each segment was measured for calculation of net fluid absorption in mL cm2 [difference between in- and outflow divided by the surface area (length  circumference) of a segment]. Mucosal scrapings from the segments were collected, immediately frozen in liquid nitrogen, and stored at  70 1C until further analysis. As described below, RNA was extracted from these scrapings to specifically assess gene expression in epithelial cells, which are supposed to be in direct contact with luminal bacteria. Furthermore, 1 cm2 pieces were cut from the intestinal wall with a stamp, washed in PBS, and stored at 4 1C in PBS until they were homogenized and plated in serial dilutions on appropriate agar plates.

Bacterial strains and culture media Lactobacillus plantarum 299v was initially used as the wildtype strain (Johansson et al., 1993). Isogenic msa- and srtAmutant derivatives of this strain had been constructed previously (G. Gross, J. Snel, J. Boekhorst, M.A. Smits, & M. Kleerebezem, unpublished data). Briefly, msa and srtA deletions had been accomplished by stable double-crossover cat replacement of the target genes in the chromosome of L. plantarum 299v wild type similar to previous construction of msa and srtA mutants in strain WCFS1 (Pretzer et al., 2005). Both wild-type strains, 299v and WCFS1, are capable of mannose adhesion as determined in a yeast agglutination assay, whereas deletion of msa or srtA leads to a complete loss of agglutination ability in each of the mutant strains (Pretzer et al., 2005; G. Gross, J. Snel, J. Boekhorst, M.A. Smits & M. Kleerebezem, unpublished data). Before the SISP experiment, the L. plantarum 299v wild-type, msa– mutant, and srtA-mutant strains were selected for natural rifampicine resistance by culturing on medium with increasing concentrations of that antibiotic (Bron et al., 2004). Salmonella typhimurium DT104A (Hendriksen et al., 2004) served as a positive control in the SISP test because the effects of this strain on host gene expression have already been determined using this model (Niewold et al., 2007). FEMS Immunol Med Microbiol 54 (2008) 215–223

For use in the SISP model, L. plantarum strains were grown overnight at 37 1C in Man Rogosa Sharp (MRS) broth (Merck, Darmstadt, Germany) containing 50 mg mL1 rifampicine for the wild-type strain and additionally 7 mg mL1 chloramphenicol for the msa- and srtA-mutant strains. Salmonella typhimurium was grown overnight aerobically at 37 1C in Luria–Bertani broth (BD Difco, Alphen aan den Rijn, The Netherlands). Bacterial counts of the initial suspensions used in the experiment were standardized by OD measurement and determined by plating of serial dilutions on appropriate agar plates. A mixture of L. plantarum 299v 50% wild-type strain and 50% msa-mutant strain, included for additional microbiological analyses after 8 h of perfusion, was prepared by adding the same volumes of adjusted suspensions of both strains to each other. Serial dilutions of homogenized jejunal samples (1 cm2 pieces) from the SISP test were plated on MRS agar containing 50 mg mL1 rifampicine that were incubated anaerobically at 37 1C for 48 h before CFU of lactobacilli were counted. One hundred CFU cultured from segments treated with the mixture of L. plantarum 299v wild-type and msa-mutant strain were replica plated on MRS agar containing 50 mg mL1 rifampicine and 7 mg mL1 chloramphenicol. Thereby, the two strains could be distinguished, i.e. the wild type being resistant only to rifampicine and the msa mutant being resistant to both rifampicine and chloramphenicol. Using this method, it was also confirmed that the initial bacterial suspension contained equal proportions of each of the two strains.

Isolation of total RNA and microarray analysis Total RNA was isolated from c. 0.5–1 g of frozen mucosal scrapings of the jejunal segments from the SISP test. The tissue was homogenized in 4 mL TRIzol reagent (Invitrogen, Breda, the Netherlands) and the extractions were essentially performed according to the instructions of the manufacturer with additional steps to remove proteoglycan and polysaccharide contaminations and DNAse treatment as described earlier (Niewold et al., 2005). Finally, RNA was dissolved in RNAse-free water and stored at  70 1C until further use. All samples were checked by agarose gel electrophoresis and UV spectrophotometry. For microarray analysis, RNA pools were prepared by mixing equal amounts of RNA isolated from intestinal segments of the different pigs collected at the same time point with the same bacterial treatment. For gene expression analysis, a home-made porcine cDNA microarray was used containing 3072 clones from a jejunum-expressed sequence tag (EST) library (Niewold et al., 2005) plus 3072 clones from a spleen EST library, all spotted in quadruplicate. For production of the spleen ESTs, spleens were collected from the same four 12-week-old pigs 2008 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved


c

218

that provided jejunal mucosal scrapings for preparation of the jejunum array (Niewold et al., 2005). A total of 2688 EST clones generated from total RNA isolated from pooled spleen tissue (n = 4) were spotted on the array along with 192 cDNA clones selected from Marc 1 and Marc 2 libraries (Fahrenkrug et al., 2002) and 192 EST clones prepared from RNA isolated from in vitro ConA/lipopolysaccharide-stimulated spleen cells derived from the abovedescribed pooled spleen tissue. Dual-color hybridization of the slides was performed using the RNA MICROMAX TSA labeling and detection kit (Perkin-Elmer, Zoetermeer, The Netherlands) as described earlier (Niewold et al., 2005). Briefly, labeled cDNA was prepared from 1 mg of RNA template from the pools of the control and of the bacterial treatments of one time point, respectively, using biotin and fluorescein. Labeled cDNAs were simultaneously hybridized to a microarray slide and detected with Cy5 (biotin) and Cy3 (fluorescein), respectively. For each comparison, a dye-swap was performed as a second hybridization with reversed labeling. The slides were scanned for Cy5 and Cy3 fluorescence in a Packard Bioscience BioChip Technologies apparatus (ScanArray Express, Perkin-Elmer). The image was processed with SCANARRAY EXPRESS software (Perkin-Elmer), automatically gridding the spots and measuring spot intensities. Manual elimination of irregular-shaped spots was performed in GenePix Pro 5.0 (Molecular Devices, Apeldoorn, The Netherlands). Data were normalized (Blank-specific background correction; Lowess fit function with a fraction of 0.2) using a customized version of the statistical software package R (Yang et al., 2002; B. Hulsegge, M.F.W. te Pas & M.H. Pool, presented at the 2nd Netherlands Bioinformatics Conference/4th International Symposium on Networks in Bioinformatics, Amsterdam, The Netherlands, 16–19 April 2007). Dye-swaps were simultaneously analyzed using significance analysis of microarrays (SAM; Tusher et al., 2001). Additionally, significantly differential expressed spots with an M value of o  1.58 or 4 1.58 [M = log2 (Cy5/Cy3)] were selected manually from data reports generated after normalization, i.e. with an absolute ratio of the effects of a bacterial treatment vs. control of greater than threefold. The use of the home-made porcine microarray technically implies a threshold level of 4 3  -/ o 0.33  -fold changes of differential gene expression to attain reliable data. For each probe present on the array eight values could be obtained, four spots per slide and per dye-swap. Probes with more than three missing values out of eight were removed from further analysis. The data discussed in this publication have been deposited in NCBIs Gene Expression Omnibus (GEO, http://www. ncbi.nlm.nih.gov/geo/) and are accessible through GEO Series accession number GSE9209. 2008 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved


c

G. Gross et al.

Reverse transcriptase reaction and quantitative real-time PCR cDNA was reverse transcribed from all individual and pooled RNA samples using Superscript II reverse transcriptase (Invitrogen) according to the manufacturer’s recommendations. Pancreatitis-associated protein (PAP) expression levels were quantified in all samples using the LightCycler Real-Time PCR (Roche Diagnostics, Almere, The Netherlands) using SybrGreen as described earlier (Niewold et al., 2007). The concentration of the 18S rRNA gene was determined accordingly by RT-PCR and was used to normalize the relative amount of PAP mRNA in all samples.

Statistical analysis When appropriate, results are indicated as  SE of the mean (SEM) and the significance of the difference between the results of the treatment and the control conditions was calculated using Student’s t-test (two sided; considered statistically significant when P is  0.05; trends are indicated when P is  0.1). Changes in the ratio of wild type and msa-mutant strain CFU when applied together in a 50–50% mixture were assessed per animal using the sign test (two sided; a = 0.05; considered statistically significant when P is  0.05).

Results Although all animals were healthy at the start of the experiment, animal three died unexpectedly during the SISP procedure after sampling at 4 h. From this pig, segments in which Salmonella and the mixture of L. plantarum 299v wild type and msa mutant had been applied were also sampled additionally at 4 h in contrast to the other three animals. However, further inspection revealed no abnormalities of this pig and, therefore, it was not excluded from data analysis of time point 4 h (i.e. 4 h n = 4, 8 h n = 3).

Microbiological analyses Bacterial counts of L. plantarum associated with the intestinal surface were determined by serial plating of jejunal samples, cut from the intestinal wall, on MRS agar containing 50 mg mL1 rifampicine (Fig. 2). In the control segments, background values of apparently naturally rifampicineresistant lactobacilli were detected. Numbers of CFU were statistically significantly higher in all jejunal segments in which L. plantarum was applied compared with the control segments after 4 and 8 h of perfusion (P o 0.05 for all treatments vs. control). No difference in bacterial counts could be observed between the treatments with L. plantarum 299v wild type, both msa- and srtA-deletion strains, and with the mixture of wild-type and msa-deletion strain. FEMS Immunol Med Microbiol 54 (2008) 215–223

219

Interaction of L. plantarum with pig intestinal epithelium

1000 net fluid absorption (µL cm–2)

800

#

600

200

*

8.0

*

* *

*

*

7.0

*

6.0

St

ko sa

ko tA

t+ m Lp

w

sa m

sr Lp

t w Lp

Fig. 4. Net fluid absorption per segment expressed in mL cm2  SEM. Data after 4 h of perfusion are depicted with white bars (n = 4), and data after 8 h of perfusion are depicted with gray bars (n = 3). Trends towards statistically significant differences between treatments compared with control conditions are indicated with a number sign (time point 8 h: PBS (control) vs. Lactobacillus plantarum wild type P = 0.087, PBS (control) vs. Salmonella typhimurium P = 0.070).

5.0

ko

ko w

Lp

t+

sr

m sa

tA

ko m sa

Lp

PB

S

Lp

(c

on

Lp

tro

l)

w t

4.0

Fig. 2. Numbers of lactobacilli determined by serial plating of jejunual samples on MRS agar containing 50 mg mL1 rifampicine. Results are calculated as log10 CFU cm2 jejunal segment  SEM. Data after 4 h of perfusion are depicted with white bars (n = 4), and data after 8 h of perfusion are depicted with gray bars (n = 3). Significantly different results (P o 0.05) compared with the control segments sampled at the same time point are indicated with an asterisk.

for all segments, reflecting a direct flow of fluid through the segments in the beginning of the experiment caused by the initial injection of the suspensions. After 8 h of perfusion, a trend towards a statistically significant decrease in net fluid absorption compared with the control conditions could be observed in the segments treated with L. plantarum wildtype strain and S. typhimurium (P = 0.087 and P = 0.070, respectively). Absorption after treatment with both msaand srtA-mutant strains and with the 50–50% mixture of wild-type and msa-mutant strain was at the same level as that of the control segments.

Gene expression analysis

70 *

*

60

% CFU

Lp

on

tro log10 CFU cm−2 jejunal segment

9.0

PB S

Calculated net fluid absorption per segment is illustrated in Fig. 4. The results after 4 h of perfusion are almost identical

ko

0

(c

Net fluid absorption

#

400

l)

However, additional information can be gathered in a competitive setting between bacterial strains that share the same binding sites for attachment to intestinal cells (Larsen et al., 2007). Replica plating discriminating between the two strains in the mixture revealed a modest enrichment of the wild-type strain in three out of four animals compared with the initial bacterial suspension (Fig. 3; statistically significant for animals 3 and 4, P  0.05).

50 40 30

4

3 An

im

al

al im An

An

im

al

1 An

im

al

sp . su al In

iti

2

20

Fig. 3. Percentage of CFU from replica plating of jejunal segments treated with a mixture of Lactobacillus plantarum 299v 50% wild-type and 50% msa-mutant strain, per individual animal. Wild-type L. plantarum is depicted with white bars, and msa mutant with gray bars. Animal 3 was sampled after 4 h of perfusion, and the other three animals after 8 h. Statistically significant changes in the ratio of wild-type and mutant strain as compared with the initial suspension (indicated by ‘Initial susp.’) that had been injected into the segments are indicated with an asterisk (P  0.05).

FEMS Immunol Med Microbiol 54 (2008) 215–223

For microarray analysis, pooled samples of segments with the same bacterial treatments were compared with control segments sampled at the same time point, and a threshold level of 4 3  -/ o 0.33  -fold changes of differential gene expression was applied for all comparisons because a lower cut-off value is technically not appropriate for the home-made microarray used in this study. Detected differences in host gene expression are presented in Table 1. No greater than threefold differences in host gene expression were revealed between control and any of the L. plantarum strains at time point 4 h. However, after 8 h of perfusion only the expression of one gene – encoding PAP – was detected to be greater than threefold (10.6  ) upregulated by L. plantarum 299v wild type. In contrast, treatment with L. plantarum 299v msa- and srtA-mutant strains did not cause any greater than threefold differential gene expression after 8 h perfusion as compared with the control treatment. No downregulation of gene expression (ratio o 0.33) was observed for any bacterial treatment. 2008 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved


c

220

Table 1. Differential gene expression induced after 8 h of perfusion of jejunal segments with bacterial suspensions Comparison L. plantarum 299v wild type vs. control L. plantarum 299v msa mutant vs. control L. plantarum 299v srt mutant vs. control S. typhimurium vs. control

Samples taken at the beginning of the experiment (t = 0) vs. control

Differentially expressed genes at time point 8 h Pancreatitis-associated protein (PAP; 10.6  ) No differential gene expression No differential gene expression Pancreatitis-associated protein (PAP; 11.2  ) Matrix-metalloproteinase-1 (MMP-1; 3.4  ) Chemokine CXC-like protein (IP-10; 4.1  ) Interleukin 8 (IL-8; 2.0  ) b-Globin (0.2  ) Cytochrome P450 (CYP3A29) (0.3  ) Calbindin (0.4  )

No up- or downregulation of gene expression was found after 4 h of perfusion. Results from microarray analysis of pooled samples from experimental treatment segments vs. control segments at the same time point are indicated as differentially expressed genes (ratio 4 3  / o 0.33  ;  -fold change). Pools from samples of segments perfused with Salmonella typhimurium and from samples collected at the beginning of the experiment were only compared vs. control segments at time point 8 h, the latter serving as a control for the effect of the perfusion procedure itself.

The results of microarray comparisons of gene expression in segments perfused with S. typhimurium for 8 h vs. control segments are displayed in Table 1 and are in agreement with previous findings (Niewold et al., 2007). Data analysis using SAM indicated IL-8 mRNA and calbindin mRNA to be differentially expressed although these did not fulfill the selection criterion of greater than threefold differential regulation. An explanation for this is that SAM processes data in a different way compared with the manual selection from normalized data, e.g. regarding averaging of M-values and exclusion of spots. Slight changes in gene expression caused by the perfusion process itself, assessed by comparison of pooled samples taken at the beginning of the experiment vs. control segments at time point 8 h, were also similar to earlier results using the SISP model (Niewold et al., 2007). Results obtained by microarray analysis indicating an induction of PAP mRNA expression by the L. plantarum 299v wild-type strain, but not by the msa- and srtA-mutant strains after 8 h of perfusion were confirmed using quantitative RT-PCR (Fig. 5). Analysis of individual samples of the animals demonstrated that PAP mRNA expression displayed a large interindividual variation among the animals; 2008 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved


c

PAP expression normalized for 18S (arbitrary units)

G. Gross et al.

25 20 15 10 5 0 124P

124P

124P

124P

124P

PBS (control)

Lp wt

Lp msa ko

Lp srt ko

St

Fig. 5. Relative expression of PAP mRNA after 8 h of perfusion in pools and per individual sample as assessed by quantitative RT-PCR normalized for the amount of 18S rRNA gene. Individual animals are depicted with white bars (animals 1, 2, and 4 indicated with 1, 2, 4 beneath bars) and pooled samples with gray bars (indicated with P beneath bars).

especially, expression levels in animal 2 were low under all experimental conditions.

Discussion In this study, a bacterial wild type and two gene-specific mutant strains of L. plantarum were used to study specific host–microorganism interaction mechanisms. The role of mannose-specific host–microorganism interactions was explored, including the effect of these bacterial strains on host responses such as fluid absorption and host gene expression. The findings of this study indicate that in this model, L. plantarum can interact with the host by inducing mRNA expression of the innate immune factor PAP and compromise fluid absorption of the intestine in a manner resembling that of the pathogen S. typhimurium. Importantly, the results imply that the mannose-specific adhesin of L. plantarum mediates these host–microorganism interactions. As various bacterial species are capable of mannosespecific interactions, the observations presented here might be of interest for several other probiotic, commensal, or pathogenic microorganisms. Intriguingly, the proposed mannose-specific bacterial induction of PAP mRNA expression suggests a newly identified molecular mechanism of host–microorganism communication. Previously, it had been concluded from an in vitro yeast agglutination assay that the L. plantarum gene msa is directly involved in mannose adhesion (Pretzer et al., 2005). In contrast to L. plantarum 299v wild-type, genespecific msa- or srtA-mutant strains are not capable of mannose-binding in vitro (G. Gross, J. Snel, J. Boekhorst, M.A. Smits & M. Kleerebezem, unpublished data). In the current study, a competitive advantage of L. plantarum 299v wild type compared with the mutant strain lacking msa was demonstrated in jejunal segments perfused with a 50–50% mixture of both strains. This enrichment of the wild-type FEMS Immunol Med Microbiol 54 (2008) 215–223

221

Interaction of L. plantarum with pig intestinal epithelium

strain in a competitive setting with the msa mutant is the first indication of a contribution of Msa to the attachment of L. plantarum to the host-intestinal surface in situ. The enrichment is not likely to be caused by growth differences between the wild-type and the mutant strain because a corresponding effect was not observed after up to 24 h of in vitro coculturing of both strains (data not shown). It should be noted that adhesion of lactobacilli to the intestinal surface is a result of multifactorial interactions, involving e.g. surface-layer proteins, lipoteichoic acids, and aggregation ability (Granato et al., 1999; Kos et al., 2003; JohnsonHenry et al., 2007). Therefore, mechanisms other than mannose-specific ones could have facilitated bacterial adherence in this study. As demonstrated here, a competitive setting is appropriate to detect differences between bacterial strains that recognize the same binding sites for attachment to the intestinal epithelium because these slight differences could not be detected in segments perfused with the individual bacterial strains. Net fluid absorption in segments perfused with L. plantarum 299v wild-type strain and S. typhimurium was found to be decreased as compared with the control segments and those treated with the L. plantarum-mutant strains. Reduced net fluid absorption can be caused by inhibition of electrolyte absorption or by the physiological situation of secretory diarrhea. These findings are known to be caused by a pathogen like Salmonella, but are not anticipated for probiotic microorganisms like lactobacilli. Metabolic activity of bacteria and subsequent accumulation of organic acids such as lactic acid have also been reported to induce net fluid excretion (Saunders & Sillery, 1982; Argenzio & Meuten, 1991). However, this is not likely to be an explanation for the effect observed in the present study because a decrease of fluid absorption was only caused by L. plantarum wild type, but not by equal numbers of the two mutant strains, suggesting a particular Msa-mediated host response. It can be hypothesized that Msa triggers decreased net fluid absorption as an initial innate reaction of the host against increased bacterial penetration. Possibly, the exposure to the lower bacterial load with Msa-expressing lactobacilli in the injected bacterial suspension can explain why the mixture of wild-type and msa-mutant strain did not affect net fluid absorption. In this model, an increased expression of host PAP mRNA was observed in two of the three individual animals after treatment with the L. plantarum 299v wild-type strain. Intriguingly, in the same animals the msa- and srtA-mutant strains did not induce a corresponding upregulation of PAP mRNA expression. This remarkable finding suggests that PAP mRNA induction is related to the functional characteristics of Msa. PAP/RegIII is a C-type lectin that has originally been described to be induced in acute pancreatitis and inflammatory bowel disease (Iovanna et al., 1993; Unno FEMS Immunol Med Microbiol 54 (2008) 215–223

et al., 1993; Dieckgraefe et al., 2002). PAP/RegIII plays a role in the promotion of epithelial cell proliferation, growth and regeneration of intestinal tissue, and in the inflammatory response (Moucadel et al., 2001; Vasseur et al., 2004; Breikers et al., 2006). Recently, it has been shown that PAP/ RegIII binds to peptidoglycans on the surface of grampositive bacteria and is directly antimicrobial (Cash et al., 2006). Upregulation of PAP mRNA by L. plantarum wild type as observed in the present study corresponds with the assumption that PAP/RegIII is induced by increased bacterial–epithelial contact to limit potential microbial penetration and to maintain mucosal integrity. This is augmented by its antimicrobial action and its contribution to epithelial repair (Keilbaugh et al., 2005; Cash et al., 2006). Indeed, elevated expression of PAP mRNA is also part of the reaction to different pathogens such as enterotoxigenic E. coli and Salmonella species as demonstrated in different animal models (this study; Niewold et al., 2005, 2007; Rodenburg et al., 2007). Furthermore, an increase in PAP/RegIII expression was detected in germ-free mice after colonization with B. thetaiotaomicron, segmented filamentous bacteria, and Schaedler’s E. coli, whereas Listeria innocua had no effect on PAP/RegIIIg mRNA levels, and it was downregulated by B. longum (Keilbaugh et al., 2005; Cash et al., 2006; Sonnenburg et al., 2006). Remarkably, in pig 2 no enrichment of the wild-type L. plantarum strain was observed after application of the mixture of a wild-type and an msa-mutant strain (Fig. 3), and hardly any PAP mRNA induction could be detected, even after treatment with S. typhimurium (Fig. 5). Individual variation of intestinal PAP expression in pigs has been observed under various experimental conditions (M. Hulst, pers. commun.) and might be influenced by the overall bacterial colonization level of each animal. This suggests that the responsiveness on this parameter differs substantially per animal and that future studies will have to take into account a proportion of ‘non-/late-responders’. In the present study, very early host responses in the pig intestine were analyzed upon challenge with L. plantarum. Most other studies in this area focus on long-term changes in gene expression induced by commensal or probiotic microorganisms in humans and mice (Hooper et al., 2001; Di Caro et al., 2005; Sonnenburg et al., 2006; Troost et al., 2006). Therefore, in these studies substantially higher numbers of genes involved in various biological functions were detected to be differentially expressed. In contrast to the current study, these effects could not be linked to one specific bacterial characteristic. Here, microarray analysis revealed no downregulation of gene expression by the bacterial treatments, and only one gene was found to be induced by L. plantarum early after challenge. This is in line with the expectation that the effects of commensal or probiotic bacterial strains on gut response are more 2008 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved


c

222

subtle compared with those induced during infections with pathogens such as Salmonella that stimulate inflammatory pathways. Gene expression data after perfusion with S. typhimurium as presented here are in agreement with this and correspond well with earlier findings of Salmonellainduced host responses using the SISP model and the same microarray platform (Niewold et al., 2007). Lower numbers of genes found to be differentially expressed as compared with other studies might be explained by a variety of parameters, e.g. model host organism, differently chosen threshold values for differential gene expression, differing host cell/microorganism ratios, and technical aspects of the home-made array platform in comparison with only recently available commercial oligonucleotide arrays for pigs. In conclusion, the results of the current study suggest a role of Msa in bacterial adherence and induction of host responses in the intestine, as reflected in decreased fluid absorption and enhanced expression of PAP mRNA as an innate immune parameter. Possibly, Msa-mediated host responses might not necessarily be dependent on bacterial adhesion properties. The preliminary indications that mannose-specific interactions contribute to early host– microorganism relationships and the proposed role of Msa will be evaluated in further in vivo experiments using larger numbers of animals.

Acknowledgements The authors thank Johan Meijer and Arie Hoogendoorn for assistance with the animal experiment and Agnes de Wit for advice on performing microarray analyses and RT-PCRs. This work was supported by the Centre for Human Nutrigenomics, The Netherlands, and EADGENE (EU contract no.: FOOD-CT-2004-506416).

References Adlerberth I, Ahrn´e S, Johansson ML, Molin G, Hanson LA˚ & Wold AE (1996) A mannose-specific adherence mechanism in Lactobacillus plantarum conferring binding to the human colonic cell line HT-29. Appl Environ Microbiol 62: 2244–2251. Argenzio RA & Meuten DJ (1991) Short-chain fatty acids induce reversible injury of porcine colon. Dig Dis Sci 36: 1459–1468. Aslanzadeh J & Paulissen LJ (1992) Role of type 1 and type 3 fimbriae on the adherence and pathogenesis of Salmonella enteritidis in mice. Microbiol Immunol 36: 351–359. Breikers G, van Breda SG, Bouwman FG, van Herwijnen MH, Renes J, Mariman EC, Kleinjans JC & van Delft JH (2006) Potential protein markers for nutritional health effects on colorectal cancer in the mouse as revealed by proteomics analysis. Proteomics 6: 2844–2852. Bron PA, Grangette C, Mercenier A, de Vos WM & Kleerebezem M (2004) Identification of Lactobacillus plantarum genes that

2008 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved


c

G. Gross et al.

are induced in the gastrointestinal tract of mice. J Bacteriol 186: 5721–5729. Bruins MJ, Cermak R, Kiers JL, van der Meulen J, van Amelsvoort JM & van Klinken BJ (2006) In vivo and in vitro effects of tea extracts on enterotoxigenic Escherichia coli-induced intestinal fluid loss in animal models. J Pediatr Gastroenterol Nutr 43: 459–469. Cash HL & Hooper LV (2005) Commensal bacteria shape intestinal immune system development. ASM News 71: 77–83. Cash HL, Whitham CV, Behrendt CL & Hooper LV (2006) Symbiotic bacteria direct expression of an intestinal bactericidal lectin. Science 313: 1126–1130. Cross ML (2002) Microbes versus microbes: immune signals generated by probiotic lactobacilli and their role in protection against microbial pathogens. FEMS Immunol Med Microbiol 34: 245–253. De Vries MC, Vaughan EE, Kleerebezem M & de Vos WM (2006) Lactobacillus plantarum-survival, functional and potential probiotic properties in the human intestinal tract. Int Dairy J 16: 1018–1028. Di Caro S, Tao H, Grillo A, Elia C, Gasbarrini G, Sepulveda AR & Gasbarrini A (2005) Effects of Lactobacillus GG on genes expression pattern in small bowel mucosa. Dig Liver Dis 37: 320–329. Dieckgraefe BK, Crimmins DL, Landt V, Houchen C, Anant S, Porche-Sorbet R & Ladenson JH (2002) Expression of the regenerating gene family in inflammatory bowel disease mucosa: Reg Ia upregulation, processing, and antiapoptotic activity. J Investig Med 50: 421–434. Fahrenkrug SC, Smith TP, Freking BA et al. (2002) Porcine gene discovery by normalized cDNA-library sequencing and EST cluster assembly. Mamm Genome 13: 475–478. Granato D, Perotti F, Masserey I, Rouvet M, Golliard M, Servin A & Brassart D (1999) Cell surface-associated lipoteichoic acid acts as an adhesion factor for attachment of Lactobacillus johnsonii La1 to human enterocyte-like Caco-2 cells. Appl Environ Microbiol 65: 1071–1077. Hendriksen SW, Orsel K, Wagenaar JA, Miko A & van Duijkeren E (2004) Animal-to-human transmission of Salmonella Typhimurium DT104A variant. Emerg Infect Dis 10: 2225–2227. Hooper LV, Wong MH, Thelin A, Hansson L, Falk PG & Gordon JI (2001) Molecular analysis of commensal host-microbial relationships in the intestine. Science 291: 881–884. Hooper LV, Stappenbeck TS, Hong CV & Gordon JI (2003) Angiogenins: a new class of microbicidal proteins involved in innate immunity. Nat Immunol 4: 269–273. Iovanna JL, Keim V, Bosshard A, Orelle B, Frigerio JM, Dusetti N & Dagorn JC (1993) PAP, a pancreatic secretory protein induced during acute pancreatitis, is expressed in rat intestine. Am J Physiol 265: G611–G618. Johansson ML, Molin G, Jeppsson B, Nobaek S, Ahrn´e S & Bengmark S (1993) Administration of different Lactobacillus strains in fermented oatmeal soup: in vivo colonization of

FEMS Immunol Med Microbiol 54 (2008) 215–223

223

Interaction of L. plantarum with pig intestinal epithelium

human intestinal mucosa and effect on the indigenous flora. Appl Environ Microbiol 59: 15–20. Johnson-Henry KC, Hagen KE, Gordonpour M, Tompkins TA & Sherman PM (2007) Surface-layer protein extracts from Lactobacillus helveticus inhibit enterohaemorrhagic Escherichia coli O157: H7 adhesion to epithelial cells. Cell Microbiol 9: 356–367. Karlsson KA (2001) Pathogen-host protein–carbohydrate interactions as the basis of important infections. Adv Exp Med Biol 491: 431–443. Keilbaugh SA, Shin ME, Banchereau RF et al. (2005) Activation of RegIIIb/g and interferon g expression in the intestinal tract of SCID mice: an innate response to bacterial colonisation of the gut. Gut 54: 623–629. Kiers JL, Nout MJ, Rombouts FM, van Andel EE, Nabuurs MJ & van der Meulen J (2006) Effect of processed and fermented soyabeans on net absorption in enterotoxigenic Escherichia coli-infected piglet small intestine. Br J Nutr 95: 1193–1198. Kos B, Suskovic J, Vukovic S, Simpraga M, Frece J & Matosic S (2003) Adhesion and aggregation ability of probiotic strain Lactobacillus acidophilus M92. J Appl Microbiol 94: 981–987. Larsen N, Nissen P & Willats WG (2007) The effect of calcium ions on adhesion and competitive exclusion of Lactobacillus ssp. and E. coli O138. Int J Food Microbiol 114: 113–119. Lee YK & Puong KY (2002) Competition for adhesion between probiotics and human gastrointestinal pathogens in the presence of carbohydrate. Br J Nutr 88 (suppl 1), S101–S108. Macpherson AJ & Harris NL (2004) Interactions between commensal intestinal bacteria and the immune system. Nat Rev Immunol 4: 478–485. Moucadel V, Soubeyran P, Vasseur S, Dusetti NJ, Dagorn JC & Iovanna JL (2001) Cdx1 promotes cellular growth of epithelial intestinal cells through induction of the secretory protein PAP I. Eur J Cell Biol 80: 156–163. Nabuurs MJ, Hoogendoorn A, van Zijderveld FG & van der Klis JD (1993) A long-term perfusion test to measure net absorption in the small intestine of weaned pigs. Res Vet Sci 55: 108–114. Neeser JR, Granato D, Rouvet M, Servin A, Teneberg S & Karlsson KA (2000) Lactobacillus johnsonii La1 shares carbohydratebinding specificities with several enteropathogenic bacteria. Glycobiology 10: 1193–1199. Niewold TA, Kerstens HH, van der Meulen J, Smits MA & Hulst MM (2005) Development of a porcine small intestinal cDNA micro-array: characterization and functional analysis of the

FEMS Immunol Med Microbiol 54 (2008) 215–223

response to enterotoxigenic E. coli. Vet Immunol Immunopathol 105: 317–329. Niewold TA, Veldhuizen EJ, van der Meulen J, Haagsman HP, de Wit AA, Smits MA, Tersteeg MH & Hulst MM (2007) The early transcriptional response of pig small intestinal mucosa to invasion by Salmonella enterica serovar typhimurium DT104. Mol Immunol 44: 1316–1322. Pretzer G, Snel J, Molenaar D, Wiersma A, Bron PA, Lambert J, de Vos WM, van der Meer R, Smits MA & Kleerebezem M (2005) Biodiversity-based identification and functional characterization of the mannose-specific adhesin of Lactobacillus plantarum. J Bacteriol 187: 6128–6136. Reid G & Burton J (2002) Use of Lactobacillus to prevent infection by pathogenic bacteria. Microbes Infect 4: 319–324. Rodenburg W, Bovee-Oudenhoven IM, Kramer E, van der Meer R & Keijer J (2007) Gene expression response of the rat small intestine following oral Salmonella infection. Physiol Genomics 30: 123–133. Saunders DR & Sillery J (1982) Effect of lactate and H1 on structure and function of rat intestine. Dig Dis Sci 27: 33–41. Sonnenburg JL, Chen CT & Gordon JI (2006) Genomic and metabolic studies of the impact of probiotics on a model gut symbiont and host. PLoS Biol 4: 2213–2226. Troost F, Kleerebezem M, De Vos WM & Brummer R-JM (2006) Identification of the acute effects of Lactobacillus plantarum WCFS1 on transcriptional responses in human intestinal mucosa. Gastroenterology 130: A610. Tusher VG, Tibshirani R & Chu G (2001) Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci USA 98: 5116–5121. Unno M, Yonekura H, Nakagawara K et al. (1993) Structure, chromosomal localization, and expression of mouse reg genes, reg I and reg II. A novel type of reg gene, reg II, exists in the mouse genome. J Biol Chem 268: 15974–15982. Vasseur S, Folch-Puy E, Hlouschek V et al. (2004) p8 Improves pancreatic response to acute pancreatitis by enhancing the expression of the anti-inflammatory protein pancreatitisassociated protein I. J Biol Chem 279: 7199–7207. Wold AE, Thorss´en M, Hull S & Ed´en CS (1988) Attachment of Escherichia coli via mannose- or Gala1 ! 4Galb-containing receptors to human colonic epithelial cells. Infect Immun 56: 2531–2537. Yang YH, Dudoit S, Luu P, Lin DM, Peng V, Ngai J & Speed TP (2002) Normalization for cDNA microarray data: a robust composite method addressing single and multiple slide systematic variation. Nucleic Acids Res 30: e15.

2008 Federation of European Microbiological Societies Published by Blackwell Publishing Ltd. All rights reserved


c