Correspondence: Jan S. Suchodolski, Gastrointestinal Laboratory, Department of Small Animal Clinical Sciences, 4474 TAMU, College Station, TX 77843-4474, USA. Tel.: +1979 458 0933; fax: +1979 458 4015; e-mail: firstname.lastname@example.org
The study aim was to describe the diversity of the intraluminal intestinal microbial community in dogs by direct sequence analysis of the 16S rRNA gene. Intestinal content was collected from the duodenum, jejunum, ileum, and colon from six healthy dogs. Bacterial 16S rRNA gene was amplified with universal bacterial primers. Amplicons were ligated into cloning vectors and near-full-length 16S rRNA gene inserts were analyzed. From a total of 864 clones analyzed, 106 nonredundant 16S rRNA gene sequences were identified. Forty-two (40%) sequences showed<98% sequence similarity to 16S rRNA gene sequences reported previously. Operation taxonomic units were classified into four phyla: Firmicutes, Fusobacteria, Bacteroidetes, and Proteobacteria. Clostridiales predominated in the duodenum (40% of clones) and jejunum (39%), and were highly abundant in the ileum (25%) and colon (26%). Sequences affiliated with Clostridium cluster XI and Clostridium cluster XIVa dominated in the proximal small intestine and colon, respectively. Fusobacteriales and Bacteroidales were the most abundant bacterial order in the ileum (33%) and colon (30%). Enterobacteriales were more commonly observed in the small intestine than in the colon. Lactobacillales occurred commonly in all parts of the intestine.
Alterations in the intestinal microbial community have been associated with intestinal disease such as inflammatory bowel disease (IBD) in humans. This observation is believed to be due to an abnormal interaction of intraluminal commensal bacteria and the immune system (Rioux et al., 2005). Gastrointestinal disease because of alterations in the microbial community is also believed to occur very commonly in dogs. Idiopathic IBD is the most common cause of chronic diarrhea in dogs, and is believed to share similarities with human IBD (Jergens et al., 1992; Qin, 2007). In a recent study evaluating histiocytic ulcerative colitis, a specific form of canine IBD that predominantly affects Boxer dogs, an association with the presence of adherent and invasive Escherichia coli and intestinal inflammation, was observed (Simpson et al., 2006). This would suggest that the commensal bacterial flora can be associated with the development of idiopathic intestinal inflammation in dogs (Hostutler et al., 2004). Antibiotic-responsive diarrhea is also a very common gastrointestinal disorder in dogs and is suspected to be caused due to an alteration or imbalance in the commensal intestinal microbial community (Westermarck et al., 2005). Therefore, the knowledge about the bacterial diversity encountered in healthy dogs is important as a baseline for future studies evaluating alterations in bacterial diversity in dogs with gastrointestinal disease.
Previous studies aiming to characterize the intestinal bacterial flora in dogs have focused almost exclusively on the identification and enumeration of bacterial species through cultivation of intestinal content. Enterobacteriaceae, Bacteroides, Clostridium, Lactobacillus, and Bifidobacterium spp. were found to be the major bacterial groups cultivable from the canine intestine (Davis et al., 1977; Benno et al., 1992; Buddington, 2003). Bacterial culture, however, has limitations for assessing the complex bacterial diversity found in the gastrointestinal tract of mammals. Samples of intestinal fluid must be processed immediately in order to preserve both aerobic and anaerobic species. Thus, an on-site microbiology laboratory is necessary in order to obtain a reliable assessment of bacterial species present and their abundance when culturing intestinal content. The intestinal tract harbors many anaerobic bacteria, which are prone to damage during sample handling. Identification of bacteria is based on phenotypic and biochemical identification systems and may lead to limitations in the characterization of all microorganisms in a given sample. It is also increasingly recognized that the majority of microbial species present in biological samples escape identification when standard culture techniques are used alone (Suau et al., 1999; Paster et al., 2001; Leser et al., 2002; Brooks et al., 2003). Studies using a molecular biological approach, i.e. based on identification of 16S ribosomal RNA (rRNA) or 16S ribosomal DNA (DNA encoding 16S rRNA), have been described and many new bacterial phylotypes have been identified in the gastrointestinal tract of various mammalian species using this method. This indicates that only a small proportion of bacterial species are cultivable using standard bacterial culture techniques (Wang et al., 1996; Suau et al., 1999; Paster et al., 2001; Greetham et al., 2002; Hayashi et al., 2002; Lan et al., 2002; Leser et al., 2002; Brooks et al., 2003). Reasons for this inability to culture many bacterial species include nonviable or stressed microorganisms, obligate requirements for a coexisting flora or host-derived products, bias due to selectivity of culture media, and our lack of knowledge regarding essential nutrients for some bacterial species (Collins & O'Mahony, 2002). Based on these studies, it is very likely that previous culture-dependent approaches have underestimated the bacterial diversity found in the intestine of dogs, and an approach based on molecular methods may identify greater bacterial diversity than previously reported in the intestinal tract of the domestic dog (Davis et al., 1977; Benno et al., 1992). While some authors have used a 16S rRNA gene PCR-denaturing gradient gel electrophoresis (DGGE) approach to describe the microbial diversity in the intestine of dogs, to the best of our knowledge, no reports are available that characterize the canine intestinal microbial community by direct sequencing of the 16S rRNA gene (Simpson et al., 2002; Suchodolski et al., 2005).
While several studies have used a comparative 16S rRNA gene approach to analyze the intestinal microbial community in humans and various animal species, most of these studies have focused almost exclusively on the microbial community present in the colon or in fecal samples. Only few studies, performed in humans, pigs, and chickens, have attempted to characterize the microbial community in more proximal parts of the intestine (i.e. jejunum and ileum) (Leser et al., 2002; Wang et al., 2003; Gong et al., 2007). However, to our knowledge no molecular study has assessed the composition of the microbial community in the duodenum of dogs. Because the duodenum is endoscopically the only accessible segment of the small intestine, it is important to determine whether the duodenal bacterial diversity is representative for the ecosystem in more distal segments of the intestine. Therefore, an accurate and comparative characterization of the microbial community in all intestinal segments is warranted. While a previous study using molecular fingerprinting techniques in healthy dogs has revealed marked qualitative differences in molecular fingerprints between individual intestinal segments, suggesting that each area of the intestine harbors an unique ecosystem, phylogentic information about the microbial community found in these segments has not been reported previously (Suchodolski et al., 2005). Therefore, the aim of this study was to characterize the bacterial microbial community in the different segments of the intestinal tract in a group of healthy dogs by comparative 16S rRNA gene analysis.
Materials and methods
The protocol for sample collection was approved by the University Laboratory Animal Care Committee at Texas A&M University. Intestinal content was collected from the duodenum, jejunum, ileum, colon, and rectum from six healthy unrelated Hound dogs (three male and three female; mean age 6.2 years, range 3.6–7 years). These dogs were euthanatized for an unrelated study. No dog received any treatment (e.g. antibiotic therapy) that would be expected to have an impact on the composition of the intestinal microbial community at least 1 month before beginning of the study. In all dogs food was withheld for 24 h before euthanasia. Immediately after euthanasia the abdominal cavity was opened, and the intestines were isolated. Approximately 0.5 mL of intestinal fluid or, in distal parts of the intestine, solid intestinal content was collected from each collection site using a sterile 16G needle attached to a 3-mL syringe or a fecal collection tube, respectively.
Extraction of DNA and 16S rRNA gene amplification
Genomic DNA was extracted from all samples individually using a bead beating method followed by phenol : chloroform : iso-amylalcohol extraction as described previously (Suchodolski et al., 2004). The PCR amplification was carried out individually on all samples. The 16S rRNA gene was amplified using primers Bact-27F (5′-AGAGTTTGATCMTGGCTCAG-3′) and Univ-1492R (5′-GGTTACCTTGTTACGACTT-3′) (Leser et al., 2002). Both primers were purchased from the Gene Technologies Laboratory, College Station, TX. DNA was amplified using the following reaction conditions: 20 mM Tris-HCl (pH 8.8), 2 mM MgSO4, 10 mM KCl, 10 mM (NH4)2SO4, 0.1% Triton® X-100, 0.1 mgmL−1 bovine serum albumin, 150 μM deoxynucleoside triphosphate, 1 mM MgCl2, 0.25 μM of each primer, 2.5 U Pfu DNA polymerase (proofreading capacity) with exonuclease activity (Strategene, La Jolla, CA), and 2 μL DNA template (c. 100 ng of DNA) in a 50 μL reaction volume. The samples were amplified in a thermocycler (Mastercycler Gradient, Eppendorf AG, Hamburg, Germany) using the following PCR protocol: an initial denaturation step at 94 °C for 3 min 15 s; 15 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. For samples that were obtained from the colon and ileum, five independent PCR reactions were performed; for samples that were obtained from the duodenum and jejunum, 10 independent PCR reactions were performed. PCR products belonging to the same individual 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 resulting PCR amplicons (c. 1450 bp) were assessed on 1.2% agarose electrophoresis gels, stained with ethidium bromide (staining for 15 min and destaining in H2O for 60 min), and visualized under UV light.
Cloning of 16S rRNA gene amplicons
Approximately equal amounts of PCR products from the corresponding intestinal locations from always two dogs were pooled and subjected to cloning and sequencing. This resulted in a total of 12 libraries: three per intestinal segment. Blunt end PCR products were ligated into linearized pCR-Blunt vectors (pCR®4Blunt-TOPO, Invitrogen, Carlsbad, CA) as specified by the manufacturer. Competent One Shot TOP10 E. coli organisms (Invitrogen) were transformed with ligation products by heat shock following the manufacturer's instructions. Recombinant organisms were grown on Luria–Bertani medium with ampicillin (50 μg mL−1) at 37 °C overnight. The pCR®4Blunt vector allows direct selection of recombinant cells via disruption of the lethal E. coli gene ccdB. Colonies were picked randomly and transferred to 1.5 mL Luria–Bertani broth and grown at 37 °C for 24 h in 2-mL well 96-well blocks (Perfectprep® BAC 96, Eppendorf) sealed with AirPore film (Eppendorf).
Plasmid extraction and sequencing of 16S rRNA gene
Plasmid extraction was performed in a 96-well format using the Perfectprep® BAC 96 plasmid purification kit (Eppendorf) and a single vacuum manifold (Eppendorf) following the manufacturer's instructions. Plasmid DNA was eluted with 50 μL of deionized water and the products were stored at −30 °C until further use. The 16S rRNA gene inserts were analyzed by cycle sequencing using the ABI PRISM BigDye Terminator Cycle Sequencing Kit (Applied Biosystems, Perkin-Elmer Corporation, Foster City, CA) and the products were analyzed with an automated sequence analyzer (ABI PRISM 377 DNA Sequencer, Applied Biosystems).
For provisional grouping of clones, all clones were reamplified from the 5′-terminal of the 16S rRNA gene using a single primer (Bact-27F; 5′-AGAGTTTGATCMTGGCTCAG-3′) (Suau et al., 1999). Sequences were aligned with the clustal_w program. A phylip distance matrix was generated and used as the input file for the dotur software to determine operational taxonomical units (OTUs) (Schloss & Handelsman, 2005). An OTU was defined as a group of sequences with <2% sequence divergence (98% similarity) to each other. One representative clone of each group was subjected to near-full-length bidirectional sequencing of both strands from positions 27 to 1492 of the 16S rRNA gene (E. coli numbering) using the following primers for sequencing: Bact-683R (5′-GCATTTCACCGCTACAC-3′), Bact-968F (5′-GAACGCGAAGAACCTTAC-3′), Bact-1054R (5′-ACGAGCTGACGACAGCCATG-3′), and Univ-1492R (5′-GGTTACCTTGTTACGACTT-3′).
All near-full-length sequences were edited to exclude the PCR primer-binding sites and tested for possible chimeric artifacts using the check_chimera program and the bellerophon software (Huber et al., 2004), both available through the Ribosomal Database Project (RDP). Putative chimeras were excluded from further analysis.
All newly obtained near-full-length sequences were compared with the existing sequences in RDP and the closest neighbor for each sequence was identified. Sequences from both, the intestinal clone library and public databases, were aligned with the clustal_w program. The resulting alignment was inspected and manually adjusted using the alignment editor in the bioedit software package. Phylogenetic trees were inferred and drawn based on the neighbor-joining algorithm using the treecon software package (version 1.3b) and the Jukes-Cantor model for inferring evolutionary distances (Van de & De Wachter, 1993). Aquifex pyrophilus was used as an outgroup. The stability of branches was assessed by the bootstrap method (100 replicates) by using the algorithms available in the treecon software package.
The coverage of the clone library (i.e. the probability that any additional analyzed clone is different from any previously analyzed single clone) was calculated according to Good (1953) using the formula [1−(n/N)] × 100, where n is the number of molecular species represented by one clone and N is the total number of sequences. The data were used to calculate bacterial diversity indices, which yield information about species diversity in a bacterial community: these calculated indices included the Simpson reciprocal diversity index and the Shannon–Weaver diversity index (Hurlbert, 1971; Atlas & Bartha, 1998). The Simpson reciprocal diversity index was defined as 1/Σ(n/N)2, where n is the number of organisms of a particular species and N is the number of organisms of all species. The Shannon–Weaver index (Hs) was defined as −Σpiln(pi), where pi is the proportion of individual bacteria found in a certain species (Atlas & Bartha, 1998). High values for the two diversity indices indicate high bacterial diversity in the sample. Rarefaction curves were produced using the software program dotur. Rarefaction analysis is used to estimate diversity and can serve as an indicator for the completeness of sampling of a specific clone library (Hurlbert, 1971).
Nucleotide sequence accession numbers
Obtained near-full-length 16S rRNA gene sequences were deposited into the GenBank database with accession numbers DQ113666–DQ113771.
A total of 988 clones were randomly selected from all samples. A total of 864 clones contained an insert that yielded a sequence of adequate quality. The partial sequence at the 5′ end of the 16S rRNA gene comprising the variable region V1–V3 of the 16S rRNA gene was used for provisional grouping of sequences based on a 98% similarity criterion. One representative from each of the provisional groups was subjected to near-full-length sequencing yielding a total of 124 unique 16S rRNA gene sequences. Eighteen (14.5%) of these near-full-length sequences were identified as possible chimeras and, together with the clones of the group they represented, were excluded from further analysis.
A total of 106 nonredundant near-full-length 16S rRNA gene sequences, representing a total of 711 clones, were used for subsequent phylogenetic analysis. Table 1 summarizes the coverage, the number of nonredundant OTUs identified, and the results for the Shannon–Weaver and the reciprocal Simpson's diversity indices for each intestinal segment. Figure 1 displays the calculated rarefaction curves for each intestinal segment.
Table 1. Clone library coverage and bacterial diversity indices for the 16S rRNA gene clone library constructed from samples obtained from various segments of the canine intestinal tract
Forty-two (40%) of the obtained near-full-length sequences showed <98% sequence similarity to existing 16S rRNA gene sequences in the GenBank and RDP databases, and may represent as of yet uncharacterized bacterial species. The results of the phylogenetic positioning of the clones are shown in Figs 2–6. Four major phylogenetic lineages were identified: the Firmicutes (47.7%), Proteobacteria (23.3%), Fusobacteria (16.6%), and Bacteroidetes (12.4%).
A total of 88 clones were affiliated with the class Bacteroides representing 13 individual phylotypes. Of these, 38 clones representing eight phylotypes belonged to the Bacteroides fragilis subgroup. Fifty clones representing five phylotypes fell into the Prevotella subgroup.
Fusobacterium and relatives
A total of 118 clones representing nine phylotypes belonging to the class Fusobacteria were identified. The genus Fusobacterium was the most predominant group within this class with 97 clones comprising eight individual OTUs. One OTU observed in the jejunum, ileum, and colon, which was represented by 36 clones, showed 96% similarity to Fusobacterium variumX55413. One OTU from the jejunum represented by nine clones showed 99% similarity with Clostridium rectum. Two jejunal clones showed 98% similarity with Fusobacterium necrogenes. The Cetobacterium subgroup and Fusobacterium perfoetens were represented by 1 and 2 OTUs, respectively.
Eighty-one clones representing 15 individual phylotypes were affiliated with the order Lactobacillales. The genus Lactobacillus was the largest subgroup with 41 clones representing six individual phylotypes. Several clones from the duodenum, jejunum, and colon showed >98% similarity with Lactobacillus reuteri, Lactobacillus murinus, and Lactobacillus johnsonii, respectively. One OTU observed in the jejunum showed 96% similarity with Lactobacillus aviarius.
Twenty-five clones representing five phylotypes were affiliated with the genus Streptococcus. One OTU observed in the jejunum, ileum, and colon showed 99% similarity with Streptococcus lutetiensis. One OTU in the duodenum showed 99% similarity with Streptococcus alactolyticus, and one OTU from the jejunum showed 99% similarity with Streptococcus suis. Two OTUs in the duodenum and jejunum showed 95% similarity with Streptococcus agalactiae.
One OTU showed 99% with Enterococcus cecorumY18355. Two OTUs were affiliated with the Abiotrophia group.
Clostridium and relatives
A total of 203 clones were affiliated with the class Clostridia representing 34 different phylotypes. Twenty-five clones representing eight individual OTUs were affiliated with the Clostridium coccoides subgroup (Clostridium cluster XIVa). A total of 135 clones representing 15 OTUs were affiliated with the Clostridium lituseburense subgroup (Clostridium cluster XI). Two OTUs were affiliated with the Clostridium leptum subgroup (Clostridium cluster IV) and five OTUs were affiliated with the Clostridium barati subgroup (Clostridium cluster I). One OTU in the jejunum showed 99% similarity to Clostridium hiranonis, a bacterial species that displays bile acid 7-α-dehydroxylating activity (Kitahara et al., 2001). One OTU in the jejunum showed 99% similarity with Clostridium glycolicum. Finally, one OTU in the jejunum showed 93% similarity with Clostridium propionicum. One OTU in the duodenum displayed 93% similarity to Candidatus arthromitus sp. (AY007720), which represents a subline within the clostridium subphylum (Snel et al., 1995).
A total of 164 clones representing 25 OTUs were affiliated with the phylum Proteobacteria. The class Gammaproteobacteria was represented by 138 clones representing 22 OTUs. Within this class the family Enterobacteriaceae was the predominant subgroup with 126 clones representing 18 individual OTUs. The genus Escherichia was the most common representative with 85 clones, followed by the genus Klebsiella, which was represented by 20 clones.
Spatial differences within the canine intestinal tract
Bacterial diversity indices increased gradually along the intestinal tract from the duodenum to the colon (Table 1). Figure 7 summarizes the percentage of OTUs belonging to the predominant bacterial orders. Clostridiales was the most abundant bacterial order in the duodenum and jejunum. Fusobacteriales and Bacteroidales were the most abundant bacterial order in the ileum and the colon, respectively. Figure 8 shows the distribution of the two predominant Clostridium clusters XI and XIVa along the canine intestinal tract. Sequences affiliated with Clostridium cluster XI were dominant in the small intestine, while sequences affiliated with Clostridium cluster XIVa were more abundant in the colon.
The molecular approach, as described in this study, has revealed the presence of a complex intestinal microbial community in the canine intestine. In the present study, Firmicutes was the most diverse (58 OTUs) and also the most abundant phylum (339 clones) in the canine intestinal tract. Clostridiales was overall the most diverse bacterial order in the intestinal tract with a total of 34 OTUs identified, forming several Clostridium clusters (Fig. 2). Clostridiales was the most abundant bacterial order in the duodenum and jejunum and was also a major constituent of the microbial community in the ileum and the colon. The two predominant Clostridium clusters XI and XIVa differed in their relative abundance within the canine intestinal tract. While sequences affiliated with Clostridium cluster XI were dominant in the small intestine, sequences affiliated with Clostridium cluster XIVa were more abundant in the colon. These findings are consistent with previous reports in humans (Wang et al., 2003) and horses (Daly et al., 2001): Clostridium cluster XIVa was, similar to our study, the predominant contributor to Clostridiales sequences in the human and equine colon (Wang et al., 2003). In contrast, sequences affiliated with Clostridium cluster XI were one of the major constituents of the order Clostridiales observed in the jejunum. This spatial distribution of Clostridia is most likely an indicator of functional and metabolic differences of this phylogenetic heterogeneous class of bacteria within the various ecosystems of the canine intestinal tract.
Anaerobic Fusobacteriales and Bacteroidales were only sporadically found in the proximal small intestine (i.e. duodenum and jejunum) but increased in their relative clone abundance along the intestinal tract and were the most abundant bacterial order in the ileum and the colon, respectively.
Clones belonging to the class Fusobacteriales were the most abundant bacterial group in the ileum (32.6% of clones) and also a major component of the colonic microbial community (28.9% of clones). These results are in accordance with previous studies using bacterial culture techniques. Up to 104 and 108 CFU of Fusobacterium spp. per mL intestinal content (CFU mL−1) were found in the jejunum and in the colon of healthy dogs, respectively, indicating that this bacterial group is a major constituent of the canine intestinal microbial community (Davis et al., 1977; Mentula et al., 2005). Interestingly, Fusobacteria appear to be a minor part of the intestinal ecosystem in other species including humans, pigs, horses, and chickens. In biopsy samples from humans, of all clones evaluated, <3% and <1% were classified as Fusobacteria in the jejunum and in the large intestine, respectively (Wang et al., 2003; Mangin et al., 2004). No Fusobacteria sequences were reported in studies evaluating the intestinal microbial community in pigs, chicken, or the colonic microbial community in horses (Daly et al., 2001; Leser et al., 2002; Gong et al., 2007). The reason for this difference in the abundance of Fusobacterium spp. between animal species remains unclear at this point but warrants future investigations into the role of these organisms in gastrointestinal health.
Proteobacteria (including E. coli-like organisms) were a substantial constituent of the duodenal microbial community (32%), but there was a low abundance of this phylum in the colon (1.4%). This low abundance in the colon is consistent with studies in humans, where it has been shown that facultative anaerobic species represent only c. 0.1% of bacteria in the strict anaerobic environment of the human colon (Suau et al., 1999). Bacterial culture analysis of dog intestinal content also revealed significant differences in the ratio of aerobic to anaerobic bacteria between the jejunum and feces (Mentula et al., 2005). While the jejunum harbored a relative similar ratio of aerobes to anaerobes, anaerobic bacteria dominated in feces (Mentula et al., 2005).
Members of the order Lactobacillales were present in high abundance in the duodenum, jejunum, and colon in all dogs. While Lactobacillus spp. were also observed in the ileum, they were only present as a minor fraction (1.4%) of all identified clones in the ileum. Similar to a study performed using bacterial culture on jejunal fluid in dogs, several OTUs with >98% similarity to S. alactolyticus, L. murinus, and L. reuteri were observed in the proximal small intestine of dogs (Rinkinen et al., 2004). Other prominent members of the Bacillus–Lactobacillus–Streptococcus subdivision were L. johnsonii and L. aviarius.
In the present molecular study no Bifidobacterium spp. were observed. Bifidobacterium spp. are considered beneficial microorganisms and part of the normal human intestinal microbial community. In dogs, isolation of Bifidobacterium spp. has not been reported consistently. Based on bacterial culture, up to 1010 CFU mL−1 of Bifidobacterium spp. have been reported in the large intestine of Beagle dogs (Davis et al., 1977; Benno et al., 1992). A lower abundance of Bifidobacterium spp. has also been reported in the proximal small intestine (Davis et al., 1977; Benno et al., 1992). Up to 5% of total bacterial counts have been identified in the jejunum and feces by bacterial culture (Mentula et al., 2005). However, Bifidobacterium spp. were identified in the jejunum only in 41% and in the feces only in 64% of these dogs (Mentula et al., 2005). Other authors have not reported isolation of Bifidobacterium spp. from the canine small intestine (Delles et al., 1994; Willard et al., 1994). Also, a study characterizing the fecal microbial community from a Labrador Retriever demonstrated that despite using Beerens agar, a medium specifically designed for the isolation of Bifidobacteria, a mixture of various organisms other than Bifidobacterium spp. was isolated (Greetham et al., 2002). The use of certain universal bacterial primers can introduce a bias in the detection of some bacterial groups, and this might have affected our results. However, the universal bacterial primers used in the present study have been shown to be able to detect Bifidobacterium spp. in previous studies assessing the intestinal bacterial microbial community in chickens (Lu et al., 2003). Also, the here used PCR protocol could have introduced bias for the detection of Bifidobacteria. Because Bifidobacterium spp. have a high G+C content in their DNA, the dissociation of the two DNA strands during the PCR assay may be hampered compared with gram-negative bacteria and low G+C gram-positive bacteria. Further studies, using specific primers and PCR assays for Bifidobacterium spp. are needed to determine the true prevalence and the diversity of Bifidobacterium spp. in the canine intestinal tract.
Previous studies using molecular fingerprinting and bacterial culture have observed a significant difference in the composition of bacterial microbial community between different segments of the canine intestinal tract (Mentula et al., 2005; Suchodolski et al., 2005). Interestingly, in this study rarefaction curve analysis would indicate that the number of phylotypes is expected to be higher in the jejunum than in the ileum (Fig. 1). In a previous study using PCR-DGGE, bacterial diversity indices in the ileum were higher but not significantly different from the jejunum (Suchodolski et al., 2005). It is possible that some of the here evaluated dogs had a less diverse microbial community, leading to this observed lower diversity in the ileum compared with the jejunum. However, the pooling of PCR amplicons in this study may have led to an overrepresentation of sequences from individual dogs, leading to confounding results.
Another limitation of this study is that PCR products from always two individual dogs were pooled. This procedure is more advantageous than the initial pooling of intestinal samples, because it potentially reduces the bias due to the preferential amplification of certain bacterial groups that are present in higher abundance in individual dogs. However, our approach by pooling PCR products still may have led to confounding results, and the best method would have been to create clone libraries for each dog and intestinal location. However, this has not been performed because of economic reasons. Therefore, bacterial sequences with low abundance may have remained undiscovered. It is also possible that clones from any of the six dogs are overrepresented in the different libraries, leading to confounding results. Also, total bacterial numbers may differ between individual sample locations, and the evaluation of the total percentages for the different bacterial groups should be interpreted with caution.
Intestinal samples were collected c. 24 h after feeding. Withholding food for 24 h before endoscopy is a routine recommendation and standard in veterinary clinical practice, and, therefore, it was also performed in this study. In adult middle-sized dogs, the mean oro-cecal transit time lasts for c. 14 h, and the mean total intestinal transit time is c. 20–30 h (Weber et al., 2002; Hernot et al., 2006). However, total gastric emptying time (i.e. the time until all consumed food has emptied from the stomach) may last up to 16 h (Weber et al., 2002). It is reasonable to assume that at the time of sample collection (24 h after food intake) remaining food residues were present in the intestinal tract of the evaluated dogs. Therefore, we speculate that the identified 16S rRNA gene clones represent mostly the normal luminal microbial community. However, because of a decrease in the concentration of nutrients that pass through the intestine after such a time period, it is possible that the makeup of the intestinal microbial community changes gradually in relation to feeding time. Therefore, in dogs that may have a faster intestinal transit time, it is possible that the identified microbiota are more representative of starved conditions. Longitudinal studies with repeated sampling of intestinal content after food intake would be useful to answer the question how fast the composition of the small intestinal microbiota change after food withholding.
Despite these limitations, this study represents the first comprehensive attempt to characterize the tremendous bacterial diversity found in the canine intestinal tract by describing the luminal microbial community of different intestinal segments from the duodenum to colon. In conclusion, the molecular approach as described in this study facilitated the identification of several previously uncharacterized bacterial 16S rRNA gene sequences in the intestinal tract of healthy dogs. A molecular approach may further aid in the identification of uncharacterized bacteria in dogs with intestinal disease, and further studies in this area are warranted.