OUP user menu

Decreased colonization of fecal Clostridium coccoides/Eubacterium rectale species from ulcerative colitis patients in an in vitro dynamic gut model with mucin environment

Joan Vermeiren , Pieter Van den Abbeele , Debby Laukens , Louise Kristine Vigsnæs , Martine De Vos , Nico Boon , Tom Van de Wiele
DOI: http://dx.doi.org/10.1111/j.1574-6941.2011.01252.x 685-696 First published online: 1 March 2012

Abstract

The mucus layer in the colon, acting as a barrier to prevent invasion of pathogens, is thinner and discontinuous in patients with ulcerative colitis (UC). A recent developed in vitro dynamic gut model, the M-SHIME, was used to compare long-term colonization of the mucin layer by the microbiota from six healthy volunteers (HV) and six UC patients and thus distinguish the mucin adhered from the luminal microbiota. Although under the same nutritional conditions, short-chain fatty acid production by the luminal communities from UC patients showed a tendency toward a lower butyrate production. A more in-depth community analysis of those microbial groups known to produce butyrate revealed that the diversity of the Clostridium coccoides/Eubacterium rectale and Clostridium leptum group, and counts of Faecalibacterium prausnitzii were lower in the luminal fractions of the UC samples. Counts of Roseburia spp. were lower in the mucosal fractions of the UC samples. qPCR analysis for butyryl-CoA:acetate CoA transferase, responsible for butyrate production, displayed a lower abundance in both the luminal and mucosal fractions of the UC samples. The M-SHIME model revealed depletion in butyrate producing microbial communities not restricted to the luminal but also in the mucosal samples from UC patients compared to HV.

Keywords
  • inflammatory bowel disease
  • SHIME
  • adhesion
  • Clostridium
  • dysbiosis
  • MAMC

Introduction

The innate immune system is responsible for the recognition of endogenous microorganisms. To maintain immune homeostasis, the host needs to distinguish commensal from pathogenic bacteria. The mucus layer on top of the epithelial cells is a first defense mechanism against which pathogens have developed mechanisms to cross. Throughout the gastrointestinal tract, the mucus layer is composed of an inner and outer mucus layer. While Johansson et al. (2008) found the inner mucus layer to be devoid of bacteria, Schreiber (2010) described the inner and outer mucus layer to be colonized with approximately 105 and 106 colony forming units per mL mucus, respectively. Only specific microbes are able to adhere to and degrade mucins. In addition, these microbes must be able to tolerate higher O2 concentrations that prevail near the epithelium and show resistance toward the antimicrobial peptides and IgA secreted by the host into the mucus layer. This has resulted in a specific mucosa-associated microbial community (MAMC) which is clearly different from the luminal community (Zoetendal et al., 2002; Macfarlane, 2008; Van den Abbeele et al., 2011a).

When the mucus layer is damaged, commensal microorganisms, which are usually only present in the lumen, play a more important role in the direct host/bacteria cross talk at the mucosa and may even contribute to pathology (Barclay et al., 2008; Swidsinski et al., 2008). Such a thinner and more discontinuous mucus layer has been described in patients with an active inflammatory bowel disease (IBD) (Strugala et al., 2008), a group of chronic disorders of the gastrointestinal tract comprising Crohn's disease (CD) and ulcerative colitis (UC). The etiology is not known, but it involves at least in part a loss of tolerance of the host toward the commensal colonic microbiota with a damaged mucus layer facilitating this response (Strober et al., 2007). Studies investigating the luminal microbiota of IBD patients have revealed a decreased diversity in Bacteroidetes and Firmicutes phyla, particularly Lactobacillus and Clostridium species (Manichanh et al., 2006; Scanlan et al., 2006; Marteau, 2009; Sokol et al., 2009). In CD, an increase in Enterobacteriaceae species was reported (Seksik et al., 2003). Studies investigating the MAMC in IBD patients revealed a higher number of bacterial cells compared to healthy individuals (Schultsz et al., 1999; Swidsinski et al., 2002) and shifts in the diversity similar to those in the luminal fraction were reported (Ott et al., 2004; Tamboli et al., 2004; Frank et al., 2007; Nishikawa et al., 2009).

The analysis of the MAMC is mainly performed with biopsies retrieved during colonoscopy; however, these human trials do have some drawbacks: (1) sampling of the mucus layer is difficult and invasive; (2) because of a high variation in the genetic background and environmental conditions of the host, there is high interindividual variability between the samples; (3) the intrinsic capacity of the microbiota to colonize the mucus is influenced by host factors and cannot be studied as such. Therefore, in vitro models offer a good alternative although validation of in vitro results is necessary. The simplest in vitro models make use of commercially available mucins (Kinoshita et al., 2007; Van den Abbeele et al., 2009) or intestinal mucus recovered from feces (Ouwehand et al., 1999). However, these models are limited to short-term colonization studies and do not include interaction with microbial communities. Recently, the in vitro dynamic gut model SHIME was modified to include mucin-covered microcosms (M-SHIME) to study the adherent community and was shown to be a relevant model to study long-term colonization of bacteria in a matrix of luminal bacteria (Van den Abbeele et al., 2011b).

The objective of this study was to investigate the capacity of the microbiota from six healthy volunteers (HV) and six UC patients to colonize a simulated mucus layer under strictly controlled feeding regimes. The M-SHIME was inoculated with fecal samples from the 12 studied individuals. After 48 h of incubation, the production of short-chain fatty acids (SCFA) was measured, and the luminal and mucin-adhered communities were characterized using denaturing gradient gel electrophoresis (DGGE) for several bacterial groups. Furthermore, specific phylogenetic groups of butyrate producers and functional genes for butyrate and propionate production were quantified using qPCR.

Materials and methods

Growth media

Unless stated otherwise, all products were ordered from Sigma Aldrich (Bornem, Belgium).

Feed medium was composed of arabinogalactan (1 g L−1), pectin (2 g L−1), xylan (1 g L−1), potato starch (3 g L−1; Anco, Roeselare, Belgium), glucose (0.4 g L−1), yeast extract (3 g L−1; Oxoid, Aalst, Belgium), special peptone (1 g L−1; Oxoid) and mucin from porcine stomach type II (4 g L−1). l-cysteine (0.5 g L−1) was added to scavenge dissolved oxygen and to lower the redox potential (Molly et al., 1993).

Mucin agar was prepared by boiling autoclaved distilled H2O containing 5% porcine mucin type II and 1% agar. The pH was adjusted to 6.8 with 10 M NaOH.

Patients and fecal suspensions

Fecal samples were obtained from six HV with a median age of 26.5 (ranging between 25 and 34) and six UC patients (UC) with a median age of 40.5 (ranging from 33 to 78). In a recent study, the composition and diversity of the gut microbiota from young adults and 70-year-old adults were found to be highly similar and differences were only found in centenarians (Biagi et al., 2010). The group of HV consisted of five male and one female and the UC group of three male and three female. The individuals did not receive antibiotics at least 1 month prior to sampling. UC patients received immunosuppressives, biologicals, 5-aminosalicylates, or a combination thereof.

Fecal suspensions were collected and prepared within 2 h according to standard procedures (De Boever et al., 2000) and stored at −80 °C until use (Guérin-Danan et al., 1999; Lauber et al., 2010).

M-SHIME

An in vitro dynamic model for the human intestinal tract, which accounts for both the luminal and mucosal microbiota, was designed by Van den Abbeele et al. (2011b). This so-called M-SHIME is based on the existing simulator of the human intestinal microbial ecosystem (SHIME). The SHIME setup represents the different parts of the adult human gut and consists of a succession of five compartments simulating the stomach, small intestine and the three colon regions: ascending, transverse and descending (Molly et al., 1993). In the M-SHIME experiment, six ascending colon (CA) vessels were run in parallel without transversum and descending colon (Fig. ). All CA vessels were modified by incorporating a mucosal environment (M-fraction) in the luminal suspension (L-fraction). The mucosal environment was created by adding 100 mucin type II agar-covered microcosms (AnoxKaldnes K1 carrier; AnoxKaldnes AB, Lund, Sweden) per vessel. Each ascending compartment (500 mL) was inoculated with 10 mL of fecal suspension from one HV or one UC patient (six of each group in total). Three times per day, 140 mL feed and 60 mL pancreatic and bile juice were added to the stomach and small intestinal compartments, respectively (Van den Abbeele et al., 2011b). After 48 h of incubation, samples from the luminal phase were collected and the microcosms covered in mucin agar were washed three times in autoclaved physiological solution (8.5 g NaCl L−1) before DNA extraction.

(a) Schematic overview of the M-SHIME with six ascending colon vessels; (b) detailed scheme of mucin type II agar-covered microcosms. - - - -, headspace connection; ——, liquid connection;, pump.

SCFA analysis

SCFA analyses were performed on the luminal and mucosal fraction as described (De Weirdt et al., 2010). Pretreatment of the mucosal fractions consisted of incubating the mucus twice in milli Q water for 2 h at 4 °C. The supernatant was collected by centrifugation at 2300 g and used for the SCFA extraction. Total SCFA is the sum of acetate, propionate, butyrate, isobutyrate, valerate, isovalerate, and isocaproate.

Microbial community analysis

DNA extraction

DNA extraction of the fecal, luminal and mucus samples was performed using the QIAamp DNA Stool Mini kit, and the protocol was followed as described by the manufacturer (Qiagen, Venlo, Netherlands). The DNA concentrations were measured with a Nanodrop ND-1000 Spectrofotometer (Isogen Lifescience, IJsselstein, Nederland).

Denaturing gradient gel electrophoresis

DGGE was applied to characterize the community composition of the luminal and mucosal communities. PCR products of the 16S rRNA genes of the total community were obtained with general bacterial primers (PRBA 338F with a GC clamp of 40 bp and P518R, Table ). A nested PCR approach was used to amplify the 16S rRNA genes of the Bacteroides/Prevotella, bifidobacteria, and lactobacilli (Table ). In the first PCR round, group-specific primer sets were used and in the second PCR round, primers PRBA338f GC and P518r were used. After the first PCR round, one clearly visible band was present on agarose gel, which suggested that no nonspecific amplification was expected in the second round. Before the second round, PCR products were diluted 100 times. For amplification of the 16S rRNA genes of the Clostridium coccoides/Eubacterium rectale group and the Clostridium leptum group, a single PCR approach was used with group-specific primers with GC clamp (Table ). Gels were run on an Ingeny PhorU apparatus (Ingeny International, Goes, The Netherlands), except for the C. leptum gels that were run on a Dcode system apparatus (BioRad, Hercules, CA) at 200 V for 4 h. Further analysis was carried out using bionumerics software version 5.20 (Applied Maths, Sint-Martens-Latem, Belgium). Pearson correlation and UPGMA clustering algorithm were used to calculate dendrograms, taking into account both band position and band density. A detailed analysis of the bands of the DGGE-profiles was performed. Ntotal is the sum of the single bands in both the lumen and the mucus DGGE-profiles, Nlumen and Nmucus indicate the amount of bands in the lumen and mucus DGGE-profiles, respectively, and Nshared/Ntotal indicates the percentage of shared bands in the lumen and mucus DGGE-profiles over Ntotal. Bands were considered different when their normalized position differed more than 0.3%.

View this table:

Overview of the primers, denaturing gradients [% acrylamide (AA) and denaturing formamide gradient] and references used for DGGE analysis

Target groupPrimersDGGE conditionsReferences
Total bacteriaPRBA 338f GC, P518r8% AA – 45–60% gradientBoon et al. (2003)
Bacteroides/PrevotellaFD1, RbacPre8% AA – 45–60% gradientWood et al. (1998); Possemiers et al. (2004)
PRBA 338f GC, P518r
LactobacilliSGLAB0159f, SGLAB667r8% AA – 45–60% gradientHeilig et al. (2002); Possemiers et al. (2004)
PRBA 338f GC, P518r
BifidobacteriaBIF164f, BIF662r8% AA – 50–65% gradientSatokari et al. (2001)
PRBA 338f GC, P518r
Clostridium coccoides/Eubacterium rectaleCocc-f, Cocc-r GC (30×)6% AA – 45–60% gradientMaukonen et al. (2006)
Clostridium leptumsg-Clept-F, sg-Clept-R3 GC (40×)8% AA – 30–52.5% gradientShen et al. (2006)

Real-time PCR quantification

Triplicate samples of 10- or 100- fold diluted genomic DNA were analyzed for total bacteria, specific phylogenetic groups (i.e., Firmicutes, Bacteroides spp., lactobacilli, bifidobacteria, Roseburia spp., Faecalibacterium prausnitzii) and functional genes (i.e., butyryl-CoA:acetate CoA transferase and propionate kinase). An ABI Prism SDS 7000 instrument (Applied Biosystems, Foster City, CA) and CFX96 instrument (BioRad) (for butyryl-CoA:acetate CoA transferase) were used. Bacteroides spp. and bifidobacteria were quantified using the qPCR® core kit for SYBR® Green I (Eurogentec, Liege, Belgium), and butyryl-CoA:acetate CoA transferase using the SensiMix SYBR No-ROX kit (Bioline Ltd., London, UK). All other protocols were performed using the Power SYBR® Green PCR Master Mix (Applied Biosystems). All primers with references are described in Table . Primers for propionate kinase were developed based on available sequences at NCBI. qPCR program for propionate kinase was as follows: 2 min at 50 °C, 5 min at 94 °C followed by 40 cycles of 30 s at 95 °C, 30 s at 58 °C, and 30 s at 60 °C. The qPCR data were normalized based on the DNA concentrations.

View this table:

Overview of the primers and references used for qPCR analysis to target specific phylogenetic groups and functional genes

Target gene/groupPrimersSequence (5′ → 3′)References
Total bacteriaPRBA 338fACTCCTACGGGAGGCAGCAGBoon et al. (2003)
P518rATTACCGCGGCTGCTGG
FirmicutesFirm934FGGAGYATGTGGTTTAATTCGAAGCAGuo et al. (2008)
Firm1060RAGCTGACGACAACCATGCAC
Bacteroides spp.Bacter 140fGGTGTCGGCTTAAGTGCCATRinttilä et al. (2004)
Bacter 140rCGGAYGTAAGGGCCGTGC
LactobacilliR_Lacto_04AGCAGTAGGGAATCTTCCAFuret et al. (2009)
F_Lacto_05CGCCACTGGTGT TCYTCCATATA
BifidobacteriaBif 243rTCGCGTCYGGTGTGAAAGPossemiers et al. (2006)
Bif 243fCCACATCCAGCRTCCAC
Roseburia spp.Ros-F1GCGGTRCGGCAAGTCTGARamirez-Farias et al. (2009)
Ros-R1CCTCCGACACTCTAGTMCGAC
Faecalibacterium prausnitziiFprau 07CCATGAATTGCCTTCAAAACTGTTSokol et al. (2009)
Fprau 02GAGCCTCAGCGTCAGTTGGT
Butyryl-CoA:acetateCoA transferaseBCoATscrFGCIGAICATTTCACITGGAAYWSITGGCAYATGLouis & Flint (2007)
BCoATscrRCCTGCCTTTGCAATRTCIACRAANGC
Propionate kinasePk-F5ACVTCRATGGGMATGACGCCThis study
Pk-R6GARTTCTCKCCWATKCCSCC

Statistical analyses

spss 16.0 (SPSS Inc., Chicago, IL) was used to carry out all statistical analyses. Normality of the data and homogeneity of variances were assessed using the Kolmogorov–Smirnov and the Levene test, respectively. Comparison of normally distributed data was performed with the Student's t-test. Otherwise, a nonparametric Kruskal–Wallis test was performed, when this indicated significant differences, means were compared using the Mann–Whitney comparison test. P-values of < 0.05 were considered significant.

Results

DGGE-profiles of the HV and UC fecal microbial inocula

Fecal samples from six HV and six UC patients (UC) were screened with DGGE before inoculation. The DGGE-profiles of the total bacteria, Bacteroides/Prevotella group, lactobacilli, bifidobacteria, C. coccoides/E. rectale group and C. leptum group of the 12 individuals were compared. No distinction could be made between the HV and UC group based on these DGGE-profiles except for the Bacteroides/Prevotella group (Fig. ). The dendrogram from the cluster analysis shows two clusters. Cluster I contains all six samples from HV and one UC sample. Cluster II contains four samples from UC, and a single UC sample did not belong to either of the clusters.

Clustering tree based on Pearson and UPGMA correlation of the group-specific DGGE-profiles for Bacteroides/Prevotella of six HV (black) and six UC (gray) samples before inoculation.

Characterization of the luminal and mucin-adhered microbiota from HV and UC patients in the M-SHIME

Production of SCFA by the luminal and mucin-adhered microbiota

The production of SCFA in the luminal content was measured after 48 h of incubation (Fig. ). Although not significant, the butyrate production was 30% lower (P = 0.27) and the acetate production 20% higher (P = 0.11) in the UC samples compared to the HV samples. The mean ratio acetate/butyrate was 2.60 ± 1.06 and 4.11 ± 1.61 for the HV and UC samples, respectively (P = 0.08). The production of isovalerate was significantly lower (85%) in UC samples (P < 0.05), and the production of isobutyrate was 78% lower in UC samples (P = 0.07). In addition, the SCFA levels were determined in the mucin agar of the microcosms. No differences were noted in the SCFA profiles from mucus samples as compared to the luminal samples. An extra incubation without microbiota but with SCFAs confirmed that SCFAs present in the luminal fraction diffuse in the mucin agar (data not shown).

SCFA production in the luminal fraction of six HV (black) and six UC (gray) samples after 48 h of incubation in the M-SHIME. The mean ± SD is shown.

Comparison of the luminal and mucin-adhered communities with DGGE

The DGGE-profiles of the luminal and mucosal communities after 48 h of incubation were compared for different phylogenetic groups. The C. coccoides/E. rectale PCR was negative for two L- and one M-fraction of the UC samples, and the C. leptum PCR was negative for one L- and five M-fractions of the HV samples and five L- and two M-fractions of the UC samples. The difference in composition between the L- and M-communities was characterized by a high interindividual variability. As an example, the clustering of the luminal- (L-) and mucin-adhered (M-) profiles for the lactobacilli is shown in Fig. . The differences in similarity between the L- and M-profiles of all DGGEs were analyzed more in depth by richness analysis of the DGGE-profiles (Table ). The sum of the bands in the lumen and mucus DGGE-profiles (shared bands were counted only once), Ntotal, was significantly lower in the UC patients for the groups C. coccoides/E. rectale and C. leptum. For the C. coccoides/E. rectale group, these lower amounts were seen both in the lumen (Nlumen) (P < 0.05) and mucus (Nmucus) of the UC samples, although the latter was not significant (P = 0.10). For the C. leptum group, the lower amounts were seen only in the lumen (Nlumen) (P < 0.05) of the UC samples. The percentage of shared bands in the L- and M-profiles (Nshared/Ntotal) was significantly higher in UC patients for lactobacilli (P < 0.05) but lower for the Bacteroides/Prevotella group (P = 0.08).

Clustering tree based on Pearson and UPGMA correlation of the group-specific DGGE-profiles for lactobacilli of the HV (black) and UC (gray) samples in the lumen (L-48) or the mucus (M-48) fraction of the M-SHIME.

View this table:

Richness analysis (amount of bands in the DGGE profile) of the HV and UC samples after 48 h of incubation in the M-SHIME

NtotalNlumenNmucusNshared/Ntotal
HVUCHVUCHVUCHVUC
Total bacteria25.7 ± 5.1a26.5 ± 2.8a15.5 ± 4.2a16.8 ± 2.6a17.0 ± 3.7a17.0 ± 0.0a35.2 ± 13.2a47.6 ± 12.2a
Lactobacilli18.8 ± 5.0a18.7 ± 3.0a14.2 ± 5.1a18.7 ± 3.0a14.7 ± 4.2a15.5 ± 2.0a43.8 ± 17.1a64.9 ± 10.9b
Bifidobacteria14.0 ± 4.9a15.5 ± 1.9a11.2 ± 3.4a12.3 ± 2.1a11.3 ± 3.2a13.3 ± 1.9a69.8 ± 20.8a59.1 ± 13.8a
Bacteroides/Prevotella7.7 ± 2.7a7.0 ± 3.9a5.5 ± 1.4a4.5 ± 3.3a5.8 ± 1.7a4.7 ± 1.4a51.5 ± 21.1a29.6 ± 19.0a
Clostridium coccoides/Eubacterium rectale11.2 ± 3.2a4.3 ± 4.3b7.2 ± 2.8a2.3 ± 2.4b6.7 ± 2.2a4.2 ± 2.6a24.5 ± 12.1a34.6 ± 10.8a
Clostridium leptum17.5 ± 4.8a3.7 ± 5.1b15.5 ± 8.5a2.0 ± 4.9b2.0 ± 4.9a4.5 ± 4.4a0a20.8 ± 41.6a
  • Ntotal is the sum of the bands in the lumen and the mucus DGGE-profiles, Nlumen and Nmucus indicate the amount of bands in the lumen and mucus DGGE-profiles, respectively, and Nshared/Ntotal indicates the percentage of shared bands in the lumen and mucus DGGE-profiles over Ntotal. For each analysis, values with different superscripts (a or b) indicate significant differences between the HV and UC samples.

Abundance of bacterial groups in the luminal and mucin-adhered communities with qPCR

To quantify specific phylogenetic groups, qPCR analysis was performed (Fig. ). After 48 h of incubation, the total amount of bacteria in the L- and M-samples from HV and UC patients was not significantly different (Fig. a). Counts of Firmicutes were 0.4 log units higher in UC samples compared to HV samples (P = 0.05) (Fig. b). No differences were found for Bacteroides spp. and lactobacilli (Fig. c and d). Representative groups of butyrate producers, Roseburia spp. and F. prausnitzii, were found to be significantly lower in the UC samples (Fig. e and f). For both Roseburia spp. and F. prausnitzii, only one of the six luminal UC samples was above the detection limit in contrast to four of six and even five of six luminal HV samples. In the mucosal UC samples, mainly Roseburia spp. were depleted (P < 0.05), while the difference in F. prausnitzii was less pronounced with three of six and four of the six samples being negative for the HV and UC group, respectively. qPCR for the bifidobacteria did not show any difference between the HV and UC samples in the L- and M-communities (data not shown).

Phylogenetic and functional qPCR analyses of the fecal communities from six HV and six UC patients inoculated in the M-SHIME for 48 h. Luminal samples are indicated by L-48 and mucus samples by M-48. (a) Total bacteria (log copies mL−1), (b) Firmicutes (log copies mL−1), (c) Bacteroides spp. (log copies mL−1), (d) lactobacilli (log copies mL−1), (e) Roseburia spp. (log copies mL−1), (f) Faecalibacterium prausnitzii (log copies mL−1), (g) butyryl-CoA:acetate CoA transferase (log copies mL−1), and (h) propionate kinase gene (log copies mL−1). Each data point represents an individual sample. The median of a group is indicated by the black line. Not detected (ND) indicates the lower detection limit of the analyses.

Abundance of functional genes in the luminal and mucin-adhered communities with qPCR

Finally, two genes of interest for microbial functionalities in UC were quantified. The gene for butyryl-CoA:acetate CoA transferase was detected in only one of the luminal and mucosal UC samples, while in HV samples, it had a median concentration of 6.42 log copies mL−1 (5.87–6.50 log copies mL−1) and 6.49 (6.18–7.13 log copies mL−1) in the luminal and mucosal fraction, respectively (Fig. g). Propionate kinase was found in similar concentrations in the lumen and mucus of HV and UC samples (Fig. h).

Discussion

The colonization of the fecal microbiota from six HV and six UC patients was compared in a modified in vitro SHIME model, the M-SHIME. This model offers the opportunity to study long-term microbial colonization of mucin beads in a background of different microbial populations. Microbe–host interactions are excluded, while microbe–microbe interactions are maintained which allows us to study the intrinsic colonization ability of the microbial communities. Incubation of the fecal microbiota from HV and UC patients in the M-SHIME showed that depletion of butyrate-producing species, particularly from the C. coccoides/E. rectale group, in UC patients is not restricted to the luminal community but is seen for the mucin-adhered community as well.

DGGE analyses showed that the richness of both the L- and M-profiles of the C. coccoides/E. rectale group, also called Clostridium cluster XIVa, was lower in UC individuals compared to healthy individuals. These results confirmed a recent study from Andoh et al. (2011) in which terminal-restriction fragment length polymorphism showed a decrease in the Clostridium cluster XIVa in the fecal microbiota from UC samples. DGGE gave us the opportunity to compare the community composition of the L-and M-profiles for each individual. However, the high interindividual variability made comparison of different individuals difficult and the conventional clustering methods mainly explorative. Therefore, counting the bands of the L- and M-profiles was performed to overcome this limitation. Richness expressed as the amount of bands has been described before by Seksik et al. (2003). However, the number of bands in a DGGE does not necessarily reflect the actual diversity of the microbial community. One band may represent several species with identical partial 16S rRNA gene sequences (Vallaeys et al., 1997), but on the other hand, one species may also produce more than one band because of multiple, heterogeneous rRNA operons (Dahllof et al., 2000). The richness analysis must, therefore, be interpreted as an indication rather than an absolute measure of the microbial diversity.

Roseburia spp., members of the C. coccoides/E. rectale group, showed a decreased abundance especially in the mucosal fraction from the UC patients compared to the HV. Based on prevalence, only one of the L-fractions and 3 of the M-fractions from the UC samples were above the detection limit. Lower amounts of C. coccoides/E. rectale cells have previously been described in fecal samples (Sokol et al., 2006) and in the mucus layer of biopsy specimens from UC patients (Swidsinski et al., 2005). Swidsinski et al. (2005) reported that the adhesion of species from the C. coccoides/E. rectale group was patchy and surprisingly, the species were detected intracellularly in some epithelial cells. Adhesion to the mucosa is a feature that is mainly studied for probiotic strains like lactobacilli and bifidobacteria. One specific probiotic Clostridium strain, that is, Clostridium butyricum CB2 of Clostridium cluster I, was found to have strong adhesion properties (Pan et al., 2008). However, very little information is available about the capacity of Clostridium species to colonize the mucus. Overall, we have demonstrated that Roseburia species are more capable of colonizing mucus than a luminal environment in both healthy and diseased individuals. Yet, we found a decreased presence of the C. coccoides/E. rectale group in the UC samples, which confirms previous observations by Swidsinski et al. (2005). Whether this was caused by a depletion of bacteria capable of colonizing mucus or by a decreased ability to colonize mucus needs further investigation.

DGGE-profiles of the C. leptum group, Clostridium cluster IV, showed that the diversity of the luminal community from UC samples was significantly lower compared to the HV samples. Remarkably, the M-fractions of only one HV sample but four UC samples were positive for the C. leptum group although the diversity of the individual UC samples was lower compared to the HV sample. The abundance of F. prausnitzii, a species of the C. leptum group, was decreased in the L- fraction from the UC samples compared to the HV samples. Both HV and UC samples showed a low prevalence of F. prausnitzii in the M-fraction indicating this species has a lower ability to colonize mucus than a luminal environment under these conditions. Previously, counts of F. prausnitzii were also found to be decreased in both the feces (Sokol et al., 2009) and in the mucus layer (Swidsinski et al., 2005) from UC patients although the latter was not significant. From these results, we can conclude that species from the C. leptum group, in particular F. prausnitzii, are mainly found in the lumen where the diversity and abundance is lower in UC patients.

The amount of copies of the functional gene butyryl-CoA:acetate CoA transferase was above the detection limit in all HV samples but in only one UC sample. These results confirmed the tendency toward a lower butyrate but higher acetate production by the microbial communities from the UC patients. Butyryl-CoA:acetate CoA transferase, in contrast to butyrate kinase, is believed to be responsible for the largest production of butyrate in the colon (Louis & Flint, 2009). The Clostridium family is well known for its capacity to produce butyrate that accounts for up to 10% of the human energy source in the intestine (Sartor, 2008). Butyrate functions as a potent anti-inflammatory factor through inhibition of nuclear factor NF-κB activation (Segain et al., 2000) and an increased production of mucins, antimicrobial peptides and tight junction proteins (Vanhoutvin et al., 2009). Several species can be detected with the primers for butyryl-CoA:acetate CoA transferase. Roseburia spp. and F. prausnitzii make up only around 25% and 4% of the total, respectively (Louis et al., 2010). Overall, the decrease in the functional gene confirms the decreases in Roseburia spp. and F. prausnitzii in UC samples. It suggests that other species may be involved in the decreased butyrate production in both the L- and M- fractions, but this was not further investigated. A decreased butyrate production at the mucosal site may lead to a decreased energy supply in the intestinal epithelium.

Besides the changes in the Clostridium clusters, comparison of the shared bands in the L- and M-profiles showed different similarities between the HV and UC communities for the lactobacilli and Bacteroides/Prevotella group. The Nshared/Ntotal showed that 65% of the total amount of bands in the lactobacilli DGGE-profiles from UC samples is shared between the L- and M-fraction, while this was only 45% in HV samples. The adhesion properties of three specific lactobacilli strains under different intestinal disease conditions have been described. Their capacity to colonize to resected colonic tissue and mucus from IBD patients was higher compared to the healthy tissue and mucus (Ouwehand et al., 2003). A selection of specific lactobacilli strains in the UC host in vivo might explain the higher in vitro colonization capacity of mucus. In contrast, 30% of Bacteroides/Prevotella bands were shared between the L- and M-fraction from the UC samples, while this was around 50% for the HV samples. Decreased colonization capacities for species from the Bacteroides/Prevotella group in UC patients confirm previous reports (Swidsinski et al., 2005; Frank et al., 2007).

It may be possible that shifts in the microbial communities are induced upon in vitro incubation. Yet, previous validation studies have demonstrated that the overall microbial community composition and activity remain fairly stable upon in vitro incubation (Molly et al., 1994; Possemiers et al., 2006) and are reproducible (Van den Abbeele et al., 2011b). Moreover, the differences observed with respect to the C. coccoides/E. rectale and C. leptum groups confirm previous studies (Swidsinski et al., 2005; Sokol et al., 2006, 2009; Andoh et al., 2011).

Taken together, incubation of the fecal microbiota from HV and UC patients in a dynamic in vitro colonization gut model enabled us to show that the diversity of the C. coccoides/E. rectale and C. leptum group, the abundance of F. prausnitzii and the functional gene butyryl-CoA:acetate CoA transferase are decreased in the luminal fractions from UC patients. Moreover, the abundance of Roseburia spp. and butyryl-CoA:acetate CoA transferase was lower in the mucosal fractions from the UC patients. The results obtained with this model confirmed previous in vitro and in vivo studies (Swidsinski et al., 2005; Sokol et al., 2006, 2009; Andoh et al., 2011). Decreased amounts of the butyrate-producing species in the mucosal fraction may lead to decreased energy supplies to the colonocytes and will negatively influence the suppression of the inflammatory processes in the colon. Therefore, we suggest to focus future research on the development of pro and/or prebiotics that are able to colonize and/or stimulate colonization of butyrate-producing strains in the colon of UC patients.

Acknowledgements

We thank Tim Lacoere and Venessa Eeckhaut for assistance with the DGGE and qPCR analysis, respectively. This work was financially supported by a Concerted Research Action of the Flemish Community (GOA) (BOF07/GOA/002) and by a grant of the ‘Strategisch Basisonderzoek – SBO’ of the Institute for the Promotion of Innovation through Science and Technology in Flanders (IWT-Vlaanderen, project nr. 100016).

Footnotes

  • Editor: Julian Marchesi

References

View Abstract