OUP user menu

Molecular-phylogenetic characterization of microbial communities imbalances in the small intestine of dogs with inflammatory bowel disease

Panagiotis G. Xenoulis , Blake Palculict , Karin Allenspach , Jörg M. Steiner , Angela M. Van House , Jan S. Suchodolski
DOI: http://dx.doi.org/10.1111/j.1574-6941.2008.00556.x 579-589 First published online: 1 December 2008

Abstract

An association between luminal commensal bacteria and inflammatory bowel disease (IBD) has been suggested in humans, but studies investigating the intestinal microbial communities of dogs with IBD have not been published. The aim of this study was to characterize differences of the small intestinal microbial communities between dogs with IBD and healthy control dogs. Duodenal brush cytology samples were endoscopically collected from 10 dogs with IBD and nine healthy control dogs. DNA was extracted and 16S rRNA gene was amplified using universal bacterial primers. Constructed 16S rRNA gene clone libraries were compared between groups. From a total of 1240 selected clones, 156 unique 16S rRNA gene sequences were identified, belonging to six phyla: Firmicutes (53.4%), Proteobacteria (28.4%), Bacteroidetes (7.0%), Spirochaetes (5.2%), Fusobacteria (3.4%), Actinobacteria (1.1%), and Incertae sedis (1.5%). Species richness was significantly lower in the IBD group (P=0.038). Principal component analysis indicated that the small intestinal microbial communities of IBD and control dogs are composed of distinct microbial communities. The most profound difference involved enrichment of the IBD dogs with members of the Enterobacteriaceae family. However, differences involving members of other families, such as Clostridiaceae, Bacteroidetes and Spirochaetes, were also identified. In conclusion, canine IBD is associated with altered duodenal microbial communities compared with healthy controls.

Keywords
  • phylogenetic
  • microbial communities
  • intestine
  • dog
  • inflammatory bowel disease
  • IBD

Introduction

There is growing evidence that inflammatory bowel disease (IBD) develops as a result of an abnormal interaction between luminal commensal bacteria and the immune system in genetically predisposed individuals (Riouxet al., 2005). The first requirement for the development of IBD is most likely a genetic susceptibility to the disease (Riouxet al., 2005). Indeed, studies have shown that in human patients there is a significant association between IBD and certain genetic defects (e.g. mutations in the CARD15 or TLR-4 genes) that interfere mainly with the normal recognition and clearance of invasive bacteria and the innate immune response (Oguraet al., 2001). It has been suggested that in genetically susceptible individuals, the resident bacterial flora is the most important factor in the development of intestinal inflammation (Riouxet al., 2005).

The importance of the resident bacterial flora in the pathogenesis of IBD is illustrated by a large number of studies in both animals and humans. Results of studies performed in animal models that spontaneously develop IBD (e.g. IL-2 or IL-10-deficient mice), indicate that enteric bacteria are required for spontaneous induction of intestinal inflammation, regardless of the underlying immunological defect (Sellonet al., 1998; Schuppleret al., 2004; Elsonet al., 2005; Riouxet al., 2005). In these studies, members of the normal bacterial flora (e.g. Escherichia coli, Enterococcus faecalis) have been associated with inflammation (Sellonet al., 1998; Schuppleret al., 2004). Likewise, studies in human patients have shown that IBD is commonly associated with a disturbed intestinal bacterial microbial communities and, in some cases, with certain bacterial species, including E. coli (Masseretet al., 2001; Neutet al., 2002; Swidsinskiet al., 2002; Darfeuille-Michaudet al., 2004; Ryanet al., 2004; Martinez-Medinaet al., 2006; Kotlowskiet al., 2007). In addition, several studies in humans with IBD have demonstrated increased antibody responses to intestinal microbial antigens (Matsudaet al., 2000; Mowet al., 2004). Finally, there is evidence that patients with IBD respond to an alteration of the enteric microbial communities through antibiotic (Burkeet al., 1990; Sutherlandet al., 1991) and/or probiotic (Gionchettiet al., 2003; Riouxet al., 2005) therapy. Collectively, these findings suggest that the intestinal bacterial flora plays a fundamental role in the pathogenesis of human IBD. Also, a recent study suggested that the intestinal bacterial flora might be indicated in the pathogenesis of IBD in cats (Janeczkoet al., 2008).

Although an important role of enteric bacteria has been proposed for the pathogenesis of canine IBD (Jergens, 1999), studies investigating this association were, until recently, lacking. In a recent study, adherent and invasive E. coli were found to be associated with granulomatous colitis (also termed histiocytic ulcerative colitis), a specific form of canine IBD that occurs almost exclusively in Boxer dogs, providing the first evidence that the commensal bacterial flora can be associated with the development of idiopathic intestinal inflammation in dogs (Simpsonet al., 2006). The recent finding that TLR2 and TRL4 expression is upregulated in dogs with IBD indirectly suggests a possible association between the intestinal microbial communities and IBD in dogs (McMahonet al., 2007). However, it is currently unknown whether the intestinal microbial communities is reproducibly altered in dogs with IBD, because studies investigating the qualitative and quantitative variation of the intestinal microbial communities in dogs with IBD have not been reported. In the case that similar alterations exist in the intestinal bacterial flora of dogs and humans with IBD, dogs with spontaneous IBD could provide not only clues for further understanding of the pathogenesis of this disease, but also the basis for the development of more effective treatment strategies in both species.

Our hypothesis was that IBD in dogs is associated with altered intestinal microbial communities. Thus, the aim of the present study was to characterize and compare the small intestinal microbial communities between dogs with spontaneous small intestinal IBD and healthy control dogs, using comparative 16S rRNA gene analysis.

Materials and methods

Animals and histopathology

The protocol was reviewed and approved by the Clinical Research Review Committee at Texas A&M University and the Ethics Committee of the Royal Veterinary College, and informed consent was obtained from the owners of each dog participating in the study. Samples were collected from the duodenum of a total of 19 dogs during endoscopy using sterile cytology brushes. The IBD group consisted of 10 dogs (age range: 3–11.3 years; median age: 5.7 years; five males, five females; one Cocker Spaniel, one Cavalier King Charles Spaniel, one Bearded Collie, one Border Collie, one English Springer Spaniel, one German Shepherd, one Great Dane, one Bull Terrier, one German Shorthaired Pointer, and one Boxer) with a diagnosis of IBD. The control group consisted of nine healthy dogs (age range: 2.7–6 years; median age: 5 years; six females and three males; three Greyhounds and six Beagles). None of the dogs were littermates. Immediately after collection, samples were placed into sterile cryotubes (Cryule 2 mL, Wheaton, Millville, NJ), and stored at −80 °C until further processing.

All dogs of the IBD group underwent a thorough diagnostic evaluation, and diagnosis of IBD was based on previously published criteria (Hall & German, 2005) after other diseases were excluded. Biopsies were collected during diagnostic endoscopy (at the same time when the material for molecular testing was obtained) from the duodenum of the IBD group dogs, fixed in formalin, stained with hematoxylin and eosin (H&E), and examined by a pathologist. Seven dogs had predominately lymphocytic–plasmacytic duodenitis and three dogs had mixed lymphocytic–plasmacytic and neutrophilic duodenitis. Mucosal biopsy specimens were also graded based on severity of histologic lesions (Jergenset al., 2003) and histologic severity scores ranged from 1 to 2 (median: 1). All dogs had clinical signs compatible with IBD, and the canine IBD activity index (CIBDAI; Jergenset al., 2003) ranged from 6 to 12 (median: 8). With the exception of one dog in the IBD group who had received antibiotics 2 weeks before sample collection, none of the dogs had received any antibiotics for at least 4 weeks before endoscopy.

Isolation of DNA

DNA was extracted from intestinal samples using a bead-beating method followed using phenol : chloroform : iso-amyl alcohol extraction as described previously (Suchodolskiet al., 2004). Briefly, 500 μL of cell lysis solution (Puregene, Gentra Systems, Minneapolis, MN), 200 μL of buffer-saturated phenol–chloroform–isoamyl alcohol (25 : 24 : 1, pH 7.2), and 300 μL of 0.1-mm zirconia beads (BioSpec Products Inc., Bartlesville, OK) were added to each sample. The tubes were positioned horizontally on a vortex adapter (Ambion Inc., Austin, TX) mounted on a standard vortexer, and the mixture was vortexed for 5 min at maximum speed. The tubes were centrifuged for 7 min at 12 000 g, and the supernatant transferred to a new, sterile, cryotube. Then 700 μL of phenol–chloroform–isoamyl alcohol was added, and the tube was vortexed for 30 s and centrifuged for 20 min at 12 000 g. The aqueous phase was transferred into a new sterile cryotube. To increase the DNA yield, 200 μL of buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8.5) was added to the remaining phenol and organic phase, the above-described extraction procedure was repeated, and the two aqueous phases were combined. To remove RNA, 5.2 U of RNAse (Puregene RNAse, Gentra Systems) were added to the solution and incubated at 37 °C for 30 min. The RNAse was removed by phenol–chloroform–isoamyl alcohol extraction as described above. The aqueous phase containing DNA was mixed with 0.5 mL of 100% ethanol and applied to commercially available spin columns (GenElute bacterial genomic DNA kit, Sigma Chemicals, St. Louis, MO). Bound DNA was washed and eluted according to the manufacturer's instructions. Purified DNA was stored at −20 °C until further use. A negative control containing H2O instead of sample was purified parallel to each extraction batch to screen for contamination of extraction reagents.

PCR for construction of 16S rRNA gene clone libraries

Amplification of the 16S rRNA genes was carried out using the universal primers Univ-27F (5′-AGAGTTTGATCCTGGCTC AG-3′) and Univ-1492R (5′-GGTTACCTTGTTACGACTT-3′). These primers have been extensively used for molecular diversity studies, because they result in a nearly full-length 16S rRNA gene product and are considered universal for the domain Bacteria. Both primers were purchased from Gene Technologies Laboratory (College Station, TX). The PCR reaction mixture contained c. 100 ng of template DNA, reaction buffer (GeneAmp 10 × PCR Gold buffer, Applied Biosystems, Foster City, CA) in a final concentration of 15 mM Tris-HCl and 50 mM KCl (pH, 8.0), 200 μM of each of the dNTPs, 3 mM MgCl2, 0.24 μM of each primer, and 1.25 U of Taq DNA polymerase (Amplitaq Gold LD, Applied Biosystems), in a final volume of 25 μL. PCR was performed in a thermocycler (Mastercycler Gradient, Eppendorf AG, Hamburg, Germany) using the following amplification conditions: an initial denaturation step at 94 °C for 3 min 15 s; 30 cycles (denaturation at 94 °C for 45 s, annealing at 54 °C for 45 s, extension at 72 °C for 3 min 30 s), and a final elongation step at 72 °C for 30 min. In order to maximize the overall DNA yield and to minimize PCR bias (Polz & Cavanaugh, 1998), 8–16 independent PCR reactions were performed for each sample, and PCR products belonging to the same sample were pooled and concentrated using the QIAquick® PCR Purification Kit (Qiagen, Valencia, CA), following the manufacturer's instructions. The purity and correct size of PCR amplicons (c. 1450 bp) were assessed on 1% (w/v) agarose gels prestained with the GelRed (Biotium, Hayward, CA) under UV light.

Cloning of PCR amplicons

The purified PCR amplicons were ligated into plasmid vectors (pCR®4-TOPO, Invitrogen, Carlsbad, CA), and then transformed into competent One Shot DH5αEscherichia coli organisms (Invitrogen) by heat shock (45 s at 42 °C) according to the manufacturer's instructions. Recombinant organisms were grown overnight on Luria–Bertani (LB) medium supplemented with ampicillin (75 μg mL−1) at 37 °C. Up to 72 colonies per sample were randomly picked, transferred to 1.4 mL LB broth in 2-mL well 96-well blocks (Perfectprep® BAC 96, Eppendorf), and grown at 37 °C for 20–24 h sealed with AirPore film (Eppendorf).

Plasmid extraction

Plasmid extraction was performed in a 96-well format using the Perfectprep® BAC 96 plasmid purification kit (Eppendorf) and a vacuum manifold (Eppendorf) as specified by the manufacturer. Plasmid DNA was eluted with 50 μL deionized water and the products were stored at −30 °C until further use.

Sequencing

The 16S rRNA gene inserts were analyzed using automated cycle sequencing. Sequencing reactions were performed using the BigDye Terminator Cycle Sequencing kit (Applied Biosystems, Perkin-Elmer Corporation), as specified by the manufacturer. For provisional grouping of clones, all clones were reamplified initially from the 5′ terminal of the 16S rRNA gene using a single primer (Bac-27F; 5′-AGAGTTTGATCMTGGCTCAG-3′). Amplification was performed using a Mastercycler gradient thermocycler (Eppendorf) with the following conditions: initial denaturation step at 96 °C for 2 min, followed by 30 cycles (denaturation at 96 °C for 30 s, annealing at 48 °C for 15 s, extension at 60 °C for 4 min). Sequencing products were purified using a ZR-96 DNA Sequencing Clean-up Kit (Zymo Research, Orange, CA) according to the manufacturer's instructions. Products were analyzed with an automated sequence analyzer (ABI PRISM 377 DNA Sequencer, Applied Biosystems).

Analysis of DNA sequences

All obtained sequences were individually inspected using commercially available software (chromaspro, Technelysium Pty Ltd, Eden Prairie, MN) in order to identify and exclude sequences of bad quality, and to remove vector sequence regions and PCR primer-binding sites. Only sequences with unambiguous nucleotide positions were included in the analysis. All sequences were scanned for possible chimeric artifacts using the programs chimera check and bellerophon, available through the internet portals Ribosomal Database Project (RDP) and Greengenes, respectively (Coleet al., 2003). Putative chimeras were excluded from further analysis. Sequences were then aligned using the clustal-w program included in the bioedit software package. The alignment was inspected and corrected manually if necessary. A distance matrix was calculated using dnadist (Jukes–Cantor correction for multiple substitutions) and used as the input file for the dotur software to determine operational taxonomical units (OTUs) (Schloss & Handelsman, 2005). An OTU was defined as sequences with <2% difference to each other based on the furthest-neighbor algorithm of the software program dotur.

The UniFrac distance metric, a method for computing differences between microbial communities based on phylogenetic information, was used to determine whether the bacterial communities of dogs with IBD and control dogs were significantly different (Lozupone, 2005). The RDP Library Compare tool, based on a naïve Bayesian classifier, was used to classify the 16S rRNA gene sequences into the new higher-order taxonomy proposed in Bergey's Taxonomic Outline of the Prokaryotes and to compare the clone libraries obtained from dogs with IBD and healthy control dogs.

The coverage of the clone library (i.e. the probability that any additional analyzed clone is different from any previously identified unique clone) was calculated according to Good (1953) using the formula Cx=[1−(Nx/n)] × 100, where Nx is the number of phylotypes represented by one clone and n is the total number of clones. The coverage was first calculated for each dog individually and then the mean coverage was calculated for each of the two groups. Information about species diversity within the bacterial communities was obtained using the Shannon–Weaver diversity index.

Data were tested for normal distribution using the Kolmogorov–Smirnov test (prism5, GraphPad Software Inc., San Diego, CA). A Mann–Whitney test was used to compare the coverage and also the number of clones in each family between the two groups. Fisher's exact tests [including odds ratios and 95% confidence intervals (CI)] were used to compare proportions of dogs between groups (prism5). Also, Spearman's r correlations were used to evaluate relationships between numbers of clones and histologic severity scores and CIBDAI. The RDP classifier was used to compare proportions of clones between groups. An t-test was used to compare species richness (i.e. the number of species identified in each of the two groups) between the two groups (prism5).

Phylogenetic analysis of sequences belonging to the genera Escherichia

One representative of each OTU that was classified in the family Enterobacteriaceae was nearly fully sequenced using primers Bac-27F, Univ-1492R, and Univ-1054R (5′-ACGAGCTGACGACAGCCATG-3′) in order to determine the genera or species that these sequences belonged to.

Nucleotide sequences accession numbers

One representative of each OTU was nearly fully sequenced (using primers Bac-27F, Univ-1492R, and Univ-1054R), and resulting sequences were submitted to GenBank under accession numbers EU681966EU682004.

Results

A total of 1240 clones were selected and their 16S rRNA gene amplicons were sequenced. Sequences of bad quality or inadequate length (345 sequences total) were excluded from analysis. Of the remaining 895 sequences, 71 (7.9%) were determined to be putative chimeras and these were excluded from further analysis, resulting in a total of 824 sequences that were subjected to final analysis. The majority of these 824 sequences had a length between 400 and 700 bases, and the average-determined length of the sequences analyzed was 625 bases. These sequences were compared with sequences listed in GenBank. Of the 824 sequences, 418 belonged to the control group and 406 belonged to the IBD group. There were a total of 156 unique 16S rRNA gene sequences, 55 (35.3%) of which represented bacterial organisms that had <98% similarity to previously described sequences, and some of them might represent new phylotypes.

The coverage of the clone library was 83% for the control group and 92% for the IBD group. In other words, the probability that the next selected clone sequence belonging to an OTU not yet observed in our study was 17% for the control group and 8% for the IBD group. These findings suggest that the clones analyzed adequately describe the dominant biodiversity in the samples evaluated. The difference in coverage between the two groups was not significant (P=0.06), thus permitting meaningful comparison between the clone libraries.

The 16S rRNA gene sequences obtained in this study, regardless of group, belonged to organisms from six phyla of the Bacteria domain (Fig. 1): Firmicutes (53.4%), Proteobacteria (28.4%), Bacteroidetes (7.0%), Spirochaetes (5.2%), Fusobacteria (3.4%), and Actinobateria (1.1%). 1.5% of the sequences belonged to Genera incertae sedis. In the control group, representatives of all six phyla were present: Firmicutes (46.4%), Proteobacteria (26.6%), Bacteroidetes (11.2%), Spirochaetes (10.3%), Fusobacteria (3.6%) and Actinobateria (1.0%). In the IBD group, sequences belonging to five phyla were identified: Firmicutes (60.6%), Proteobacteria (30.3%), Bacteroidetes (2.7%), Fusobacteria (3.2%) and Actinobateria (1.2%).

1

Phylum-level composition and comparison of the duodenal bacterial flora of healthy dogs (control dogs) and dogs with IBD (IBD dogs) as determined by sequencing of 824 clones. The shade of the scale bars depicts percentages of cloned sequences in all samples of each group that belong to one of the six phyla of Bacteria identified in the present study.

Sequences belonging to the phylum Bacteroidetes were found to be significantly less common in dogs of the IBD group (2.7%) than in dogs of the control group (11.2%; P<0.001; odds ratio, 4.5; 95% CI, 2.3–8.9), but the difference in the number of dogs in each group that harbored organisms belonging to the phylum Bacteroidetes was not significant (P=0.37). Sequences of the phylum Spirochaetes were found in dogs of the control group (10.3%) but not in dogs of the IBD group (P<0.001; odds ratio, 94.2; 95% CI, 5.8–1536). The difference in the number of dogs in each group that harbored organisms belonging to the Spirochaetes phylum was significant (P<0.001).

The overall phylogenetic distribution of the sequences identified in dogs of the control group are shown inTable 1. Nearly, half of the sequences (47.6%) of the control group belonged to three orders: Clostridiales (19.6%), Lactobacillales (14.1%) and Campylobacteriales (13.9%). The majority of the sequences in these orders belonged to the families Clostridiaceae (70.7%), Carnobacteriaceae (52.5%) and Helicobacteriaceae (94.8%), respectively.

View this table:
1

Overall phylogenetic distribution of the sequences identified in the group of healthy control dogs (n=9) and in the group of dogs with IBD (n=10)

PhylumClassOrderFamilyHealthy dogs % (n=418)IBD dogs % (n=406)
Firmicutes46.460.6
Clostridia19.640.4
Clostridiales19.640.4
Clostridiaceae13.934.2
Peptostreptococaceae1.90.2
Acidaminococaceae0.2
Peptococaceae1.2
Lachnospiraceae1.75.7
Eubacteriaceae0.2
Thermosulfodiaceae0.7
Bacilli26.119.5
Lactobacillales14.117.5
Carnobacteriaceae7.40.5
Lactobacillaceae1.90.5
Streptococaceae2.67.2
Aerococaceae2.24.5
Enterococaceae5
Bacillales122
Turicibacteriaceae8.9
Staphylococcaceae3.12
Mollicutes0.70.7
Mycoplasmatales0.70.7
Mycoplasmataceae0.70.7
Proteobacteria26.630.3
Alphaproteobacteria0.2
Rhizobiales0.2
Hyphomicrobiaceae0.2
Betaproteobacteria2.20.7
Burkholderiales1.20.2
Alcaligenaceae0.5
Burkholderiaceae0.7
Comamonadaceae0.2
Neisseriales10.5
Neisseriaceae10.5
Gammaproteobacteria10.324.4
Pasteurellales6.53
Pasterelaceae6.53
Pseudomonadales1.9
Moraxellaceae1.9
Cardiobacteriales0.20.5
Cardiobacteriaceae0.20.2
Enterobacteriales0.720.9
Enterobacteriaceae0.720.9
Aeromonadales0.9
Succinivibrionaceae0.9
Epsilonproteobacteria13.95.2
Campylobacteriales13.95.2
Helicobacteriaceae13.23.7
Campylobacteriaceae0.71.5
Bacteroidetes11.22.7
Bacteroidetes10.32.2
Bacteroidales10.32.2
Porphyromonadaceae7.20.7
Prevotellaceae2.2
Bacteroidaceae11.5
Flavobacteria0.90.5
Flavobacteriales0.90.5
Flavobacteriaceae0.90.5
Actinobacteria11.2
Actinobacteria11.2
Actinomycetales11.2
Microbacteriaceae0.70.2
Micrococaceae0.2
Actinomycetaceae0.2
Corynebacteriaceae0.5
Propionibacteriaceae0.2
Spirochaetes10.3
Spirochaetes10.3
Spirochaetales10.3
Spirochaetaceae10.3
Fusobacteria3.63.2
Fusobacteria3.63.2
Fusobacteriales3.63.2
Fusobacteriaceae3.63.2
Unclassified0.92
  • The proportion (percentage) of clones in each category (Phylum, Class, Order, Family) in relation to the total number of clones selected in each group (418 in healthy; 406 in IBD dogs) is reported.

Table 1 also shows the phylogenetic distribution of the sequences identified in dogs of the IBD group. The vast majority of these sequences (78.8%) belonged to three orders: Clostridiales (40.4%), Enterobacteriales (20.9%) and Lactobacillales (17.5%). The majority of the sequences in these orders belonged to the families Clostridiaceae (84.8%), Enterobacteriaceae (100%) and Streptococaceae (40.8%), respectively.

Principal components analysis based on the UniFrac distance metric were performed in order to investigate any possible association between specific OTUs (at 98% phylogenetic depth) and health or disease status. The results of these analyses suggest that the small intestinal microbial communities of IBD dogs and control dogs are composed of distinct microbial communities (Fig. 2). In addition, the dogs of the IBD group were found to have a significantly lower species richness (9.8±1.6) than the control group dogs (15.8±2.2; P=0.038). There was no significant difference in the Shannon–Weaver diversity index between the dogs of the control group and those of the IBD group (P=0.216).

2

PCA based on the UniFrac distance metric. Each ball represents a duodenal sample obtained from one dog and it is shaded according to disease status. Grey balls represent dogs with IBD while black balls represent healthy control dogs. In general, samples clustered according to disease status of the dog, suggesting that IBD and control dog samples are composed of distinct bacterial communities.

Using the RDP Classifier, we identified several families that, based on total numbers of clones, were significantly associated with health or diseased status (outlined inTable 2). In addition, using Fisher's exact tests, we identified bacterial families that were associated with disease status based on the number of dogs that they had been identified in (Table 2). Compared with the dogs of the control group, samples from dogs of the IBD group were significantly enriched in sequences of the Enterobacteriaceae family. There was a significant difference in the proportion of clones belonging to the Enterobacteriaceae family between dogs of the control (0.7%) and the IBD (20.9%) group (P<0.001; odds ratio, 36.6; 95% CI, 11.5–116.9). In addition, there was a significant difference in the proportion of dogs that harbored organisms of the Enterobacteriaceae family between the control (two out of nine dogs) and the IBD (nine out of 10 dogs) group, and dogs of the IBD group were 31.5 times more likely (odds ratio, 31.5; P=0.006; 95% CI, 2.3–422.6) to harbor organisms of the Enterobacteriaceae family compared with the control group. Also, using Mann–Whitney tests, sequences belonging to the Enterobacteriaceae family were the only bacterial group that was significantly more commonly identified in IBD dogs than in controls (P=0.0033). After performing additional sequencing of one representative of each OTU that was classified in the family Enterobacteriaceae, the majority of the sequences obtained from dogs of the IBD group could be further classified as E. coli-type sequences.

View this table:
2

Comparison of the phylogenetic analysis in duodenal samples from healthy dogs (controls) and dogs with IBD (IBD)

FamilyNumber of clonesSignificanceNumber of dogsSignificance
ControlsIBDControlsIBD
Enterobacteriaceae385<0.00129<0.01
Clostridiaceae58139<0.00177NS
Enterococcaceae020<0.00101NS
Spirochaetaceae430<0.00140<0.05
Turicibacteraceae370<0.00130NS
Carnobacteriaceae312<0.00132NS
Helicobacteraceae5515<0.00131NS
Porphyromonadaceae303<0.00152NS
Streptococcaceae1129<0.0155NS
Moraxellaceae80<0.0130NS
Prevotellaceae90<0.0130NS
Lachnospiraceae723<0.0132NS
Peptostreptococcaceae81<0.0551NS
Peptococcaceae50<0.0530NS
Pasteurellaceae2712<0.0552NS
Lactobacillaceae82NS11NS
Campylobacteraceae36NS21NS
Fusobacteriaceae1513NS44NS
Flavobacteriaceae42NS11NS
Staphylococcaceae138NS34NS
Aerococcaceae918NS24NS
Bacteroidaceae46NS12NS
Mycoplasmataceae33NS22NS
Thermodesulfodiaceae30NS20NS
Neisseriaceae42NS22NS
Cardiobacteriaceae12NS12NS
Burkholderiaceae30NS10NS
Eubacteriaceae10NS10NS
Hyphomicrobiaceae10NS10NS
Microbacteriaceae31NS11NS
Micrococcaceae10NS10NS
Succinivibrionaceae40NS10NS
Alcaligenaceae20NS10NS
Actinomycetaceae01NS01NS
Acidaminococcaceae01NS01NS
Corynebacteriaceae02NS01NS
Propionibacteriaceae01NS01NS
Comamonadaceae01NS01NS
  • The number of clones (number of clones) in each family was compared between groups using the RDP Library Compare tool. The number of dogs in which a bacterial family was identified is shown, and the proportion of these dogs in relation to the total number of dogs in each group was compared between groups using Fisher's exact tests. Several families were significantly associated with disease status based on the total numbers of clones, but only two showed statistical significance based on the number of dogs in which they were identified in. Significance was set at P<0.05. NS, not significant (P>0.05).

Similarly, compared with the dogs of the control group, samples of the group of IBD dogs were significantly enriched in sequences of the Clostridiaceae family (Table 2). There was a significant difference in the proportion of clones belonging to the Clostridiaceae family between dogs of the control (13.9%) and the IBD (34.2%) group (P<0.001; odds ratio, 3.2; 95% CI, 2.3–4.6). However, there was no difference in the proportion of dogs that harbored organisms of the Clostridiaceae family between the control and the IBD group, as members of this family were commonly found in dogs of both groups (seven dogs in each group).

Several other significant differences were identified in the number of clones of sequences belonging to certain families based on the RDP Classifier (outlined inTable 2). Of those, only the Spirochaetaceae family also showed a significant difference in the number of healthy and IBD dogs that harbored these sequences.

There was no statistically significant correlation between histologic scores and CIBDAI (P=0.44). However, there was a significant positive correlation between the number of clones belonging to the Clostridiaceae family and CIBDAI in IBD dogs (Spearman r=0.75; P=0.026). The correlation between the number of clones belonging to the Enterobacteriaceae family and histologic severity in IBD dogs did not reach statistical significance (Spearman r=0.66; P=0.059).

Discussion

It is generally accepted that the pathogenesis of IBD is based on a dysfunctional interaction between the intestinal microbial communities and the mucosal immune system of the intestine, although most aspects of this interaction have not been characterized yet. Regarding the role of the intestinal microbial communities in the pathogenesis of IBD, two major theories exist: either select bacterial components of the intestine play a crucial role in the pathogenesis of IBD or, alternatively, the overall composition and balance of the intestinal microbial communities are relevant for the pathogenesis of IBD. Currently, there is evidence to support both theories (Darfeuille-Michaudet al., 2004; Ottet al., 2004; Mylonakiet al., 2005; Gophnaet al., 2006; Franket al., 2007). In the present study, we used molecular–phylogenetic analyses to characterize and compare the small intestinal microbial communities of healthy control dogs and dogs with IBD.

To our knowledge, this is the first study documenting that canine IBD is associated with imbalances of the small intestinal microbial communities. A reduced species richness of the small intestinal microbial communities was evident in the IBD group dogs in the present study. Reduced diversity also has been documented in human patients with IBD (Ottet al., 2004). Although the IBD group dogs were similar to the control group dogs in that they were both dominated by bacteria of the phyla Firmicutes and Proteobacteria, the two groups differed significantly in the relative proportions in which subgroups of these phyla were present. Also, organisms of the phylum Bacteroidetes, which are commonly found in the distal small intestine of humans and are considered to be beneficial for the health of the gastrointestinal tract, were found to be depleted in samples from the intestinal microbial communities of dogs in the IBD group in this study. Similarly to our findings, reduced numbers of Bacteroidetes sequences have been found in the small intestine of humans with IBD, and it has been hypothesized that depletion of specific beneficial bacteria may exacerbate some forms of IBD (Franket al., 2007).

Sequences of the Enterobacteriaceae family were substantially more abundant in dogs of the IBD group than in control dogs, and were present in nine of the 10 IBD dogs but only in two of nine of the control dogs. The majority of these sequences were further classified as E. coli-type sequences. The finding that E. coli-type organisms were more commonly identified in dogs with IBD is in agreement with findings of several studies in humans and animal models with IBD (Sellonet al., 1998; Darfeuille-Michaudet al., 2004; Schuppleret al., 2004; Mylonakiet al., 2005; Kotlowskiet al., 2007). In one study, invasive and adherent-invasive E. coli strains were identified in the ileum of 21.7–36.4% of patients with Crohn's disease compared with 6.2% of controls, and it was concluded that adherent-invasive E. coli is associated with ileal Crohn's disease (Darfeuille-Michaudet al., 2004). In a more recent study, a high prevalence of E. coli strains belonging to the B2+D phylogenetic group was found in humans with IBD (Kotlowskiet al., 2007). These and similar studies provide evidence that increased numbers of E. coli strains are present in the intestinal mucosa of human patients with IBD.

However, other studies have questioned the above findings. In a recent study, the presence of adherent and invasive E. coli was not found to be disease specific, as it was found in similar percentages in patients with Crohn's disease and patients with colon cancer (Martinet al., 2004). In addition, studies using molecular techniques failed to identify specific and sufficiently reproducible phylogenetic profiles that were associated with IBD (Kleessenet al., 2002; Swidsinskiet al., 2002; Franket al., 2007). Thus, a cause and effect relationship between E. coli strains or other specific phylogenetic groups and human IBD has not been established yet. In dogs, an association between E. coli and certain intestinal inflammatory diseases has recently been described. It has been shown that intramucosal colonization by E. coli is associated with histiocytic ulcerative colitis (i.e. granulomatous colitis of Boxer dogs), which is a rare idiopathic form of intestinal inflammation that typically affects Boxer dogs (Van Kruiningenet al., 2005; Simpsonet al., 2006). In a study by Simpsonet al. (2006), E. coli strains isolated from affected Boxers were found to display virulence behavior in vitro, as they were found to adhere to, invade, persist in, and replicate in cultured epithelial cells. However, in this particular study, no E. coli strains were found to invade the colonic mucosa of dogs with forms of IBD other than histiocytic ulcerative colitis.

Sequences of the Clostridiaceae family were found to be enriched in dogs with IBD in this study. Similar findings have been reported in human patients with IBD involving the colon (Mylonakiet al., 2005), but other studies have shown a decrease in abundance of Firmicutes, especially those belonging to the class of Clostridia, in human IBD patients (Gophnaet al., 2006). Specifically, decreased abundance of the Clostridium leptum phylogenetic group has been found in fecal samples from humans with IBD (Manichanhet al., 2006; Sokolet al., 2006). In the present study, C. leptum was not identified in any of the dogs. This was to be expected, however, because C. leptum has been shown to be a constituent of the canine large intestine and has not been identified in the small intestine of healthy dogs (Suchodolskiet al., 2008).

Several other families were found to be significantly associated with IBD in the present study based on number of clones identified (Table 2), but they were not reproducibly identified in dogs of the IBD group. Thus, an association between the identification of these families and canine IBD is more difficult to establish. Similarly, several families were more commonly or exclusively found in healthy control dogs than in dogs with IBD in the present study. Whether organisms belonging to these families play a protective role against IBD, or whether depletion of organisms belonging to these families can predispose to IBD remains to be determined. Nevertheless, organisms belonging to these families can be considered constituents of the normal canine small intestinal microbial communities.

It is not known whether the altered intestinal microbial communities observed in the present study preceded the development of IBD or whether it was secondary to intestinal inflammation. Collection and analysis of biopsy specimens taken from both inflamed and noninflamed areas of the intestine of dogs with IBD could answer this question. Studies in humans with IBD show that the dominant microbial communities are similar in inflamed and noninflamed areas of the intestine of the same individual. These findings suggest that changes in the microbial communities are more likely primary rather than secondary to inflammation, and that inflammation is unlikely to be caused directly by a mucosa-associated pathogen (Swidsinskiet al., 2002; Seksiket al., 2005; Gophnaet al., 2006; Vasquezet al., 2007).

In this study, a significant correlation was found between the number of clones belonging to the Clostridiaceae family and CIBDAI in IBD dogs. In addition, a trend towards a significant positive correlation was found between the number of clones belonging to the Enterobacteriaceae family and histologic severity in IBD dogs. Although these findings are interesting and might prove to have some clinical usefulness, they should be evaluated with caution because of the semi-quantitative character of the analysis used in the present study.

Although culture-independent approaches provide several advantages compared with culture-based analyses, they are not free of limitations and molecular studies can lead to a certain bias (Suzuki & Giovannoni, 1996). In addition, in 16S rRNA gene analysis-based studies, clone abundance is only partly quantitative, and functional changes in the intestinal microbial communities (e.g. mucosal adherence, enhanced virulence) cannot be identified. However, comparisons between clone libraries that have been constructed using the exact same methods and procedures are acceptable and can lead to useful information.

In conclusion, results of the present study suggest that, similarly to humans, imbalances of the intestinal microbial communities occur in dogs with IBD. The most profound finding in this study was that samples from dogs with IBD were significantly enriched in sequences of the Enterobacteriaceae family (E. coli-type sequences). Also, several other alterations were identified in the intestinal microbial communities of dogs with IBD, suggesting that imbalances of the overall intestinal microbial communities are present in these dogs. However, most of these alterations were not observed in a significant proportion of dogs, and, therefore, need to be interpreted with caution. Further studies are needed to determine the reproducibility and significance of these findings, and to identify and quantify specific E. coli strains that might be associated with canine IBD. In addition, comparison of the intestinal microbial communities between dogs with IBD and dogs with other intestinal diseases (e.g. neoplasia, infectious enteritis) is necessary in order to identify which changes of the microbial communities (if any) are specific for IBD.

Acknowledgements

This work was supported, in part, by a grant by the Comparative Gastroenterology Society and Waltham.

Footnotes

  • Editor: Julian Marchesi

References

View Abstract