OUP user menu

Massive parallel 16S rRNA gene pyrosequencing reveals highly diverse fecal bacterial and fungal communities in healthy dogs and cats

Stefanie Handl , Scot E. Dowd , Jose F. Garcia-Mazcorro , Jörg M. Steiner , Jan S. Suchodolski
DOI: http://dx.doi.org/10.1111/j.1574-6941.2011.01058.x 301-310 First published online: 1 May 2011

Abstract

This study evaluated the fecal microbiota of 12 healthy pet dogs and 12 pet cats using bacterial and fungal tag-encoded FLX-Titanium amplicon pyrosequencing. A total of 120 406 pyrosequencing reads for bacteria (mean 5017) and 5359 sequences (one pool each for dogs and cats) for fungi were analyzed. Additionally, group-specific 16S rRNA gene clone libraries for Bifidobacterium spp. and lactic acid-producing bacteria (LAB) were constructed. The most abundant bacterial phylum was Firmicutes, followed by Bacteroidetes in dogs and Actinobacteria in cats. The most prevalent bacterial class in dogs and cats was Clostridia, dominated by the genera Clostridium (clusters XIVa and XI) and Ruminococcus. At the genus level, 85 operational taxonomic units (OTUs) were identified in dogs and 113 OTUs in cats. Seventeen LAB and eight Bifidobacterium spp. were detected in canine feces. Ascomycota was the only fungal phylum detected in cats, while Ascomycota, Basidiomycota, Glomeromycota, and Zygomycota were identified in dogs. Nacaseomyces was the most abundant fungal genus in dogs; Saccharomyces and Aspergillus were predominant in cats. At the genus level, 33 different fungal OTUs were observed in dogs and 17 OTUs in cats. In conclusion, this study revealed a highly diverse bacterial and fungal microbiota in canine and feline feces.

Keywords
  • microbiota
  • pyrosequencing
  • fecal
  • Lactobacillus
  • Bifidobacterium
  • feline
  • canine

Introduction

The intestinal microbiota plays a crucial role in host health. It aids in digestion and energy harvest from the diet, provides nutrition for the enterocytes, plays a role in the development of the immune system, and acts as a barrier against pathogen invasion (Neish, 2009). The intestinal ecology has been mostly investigated using culture-dependent techniques. The discovery of molecular methods, such as comparative 16S rRNA gene analysis, has revealed a much more diverse intestinal ecosystem than recognized previously (Suau et al., 1999). Alterations of this complex ecosystem have been associated with various diseases in humans, including inflammatory bowel disease (IBD) (Sokol et al., 2006; Frank et al., 2007), irritable bowel syndrome (Kassinen et al., 2007), atopic disorders (Penders et al., 2007), and obesity (Ley et al., 2006; Zhang et al., 2009). Also, in dogs and cats, changes in the intestinal microbiota have been related to chronic diarrhea (Bell et al., 2008; Jia et al., 2010) and small intestinal IBD (Janeczko et al., 2008; Xenoulis et al., 2008; Suchodolski et al., 2010).

Despite these findings, in-depth molecular investigations of the gastrointestinal communities in dogs and cats are still scarce. Studies using cultivation methods revealed that Bacteroides, Clostridium, Lactobacillus, Bifidobacterium spp., and Enterobacteriaceae are the predominant bacterial groups in the canine and feline intestine (Benno et al., 1992; Johnston et al., 2001; Mentula et al., 2005). Ritchie (2008) and Suchodolski (2008a) constructed 16S rRNA gene clone libraries to characterize the microbiota in chyme samples from various segments of the gastrointestinal tract of healthy dogs and cats. Firmicutes, Bacteroidetes, and Proteobacteria were the most abundant phyla, with Clostridiales being the most prevalent bacterial order. Although these studies yielded a valuable insight into the microbial ecosystem along the gastrointestinal tract, chyme samples are usually not available for clinical use. Fecal material is more easily accessible for diagnosing and monitoring gastrointestinal diseases. Recent characterizations of microbial communities in fecal samples from cats using chaperonin 60 gene (Desai et al., 2009) and 16S rRNA gene (Ritchie et al., 2010) based clone libraries indicate that Firmicutes, Proteobacteria, Actinobacteria, and Bacteroidetes were the most abundant bacterial groups in feces. Newer high-throughput sequencing-by-synthesis techniques, such as pyrosequencing, are able to provide a much deeper coverage of the intestinal microbiome (Hamady & Knight, 2009) and have only been used to analyze fecal microbial communities of research dogs (Middelbos et al., 2010), but not in pet dogs and pet cats. Such profound studies are urgently needed to serve as a reference for further evaluation of the role of the intestinal microbiome in canine and feline gastrointestinal diseases.

Lactic acid-producing bacteria (LAB), including Lactobacillus spp. and Bifidobacterium spp., are commonly used as probiotics because they may inhibit pathogen invasion and enhance the immune response (Ogué-Bon et al., 2010). Although probiotics are frequently administered in small animal practice, only limited data are currently available on the prevalence and diversity of lactic acid-producing microorganisms in the gastrointestinal tract of pets. Ritchie (2010) identified 23 Lactobacillus-like and 11 Bifidobacterium-like sequences in the feces of cats using group-specific primers and 16S rRNA gene clone libraries, but no similar study has been performed on canine feces.

Information about the prevalence and the classification of fungal organisms in the gastrointestinal tract of mammals is currently limited. Neocallimastigales are commonly found in the large intestine of herbivores, where they are supposed to contribute to fiber digestion (Nicholson et al., 2010). A highly diverse fungal microbiota was observed in the intestinal contents of mice using oligonucleotide fingerprinting of rRNA genes (Scupham et al., 2006). Using various culture-independent techniques, Ott (2008) discovered differences in the diversity and composition of the fungal communities in colonic biopsies from healthy humans vs. IBD patients. Similarly, Suchodolski (2008b) identified the fungal phyla Ascomycota and Basidiomycota in healthy dogs and dogs with chronic enteropathies. However, the role of the intestinal fungal population is poorly understood, and more detailed characterization of the fungal communities in gastrointestinal health and disease is needed.

Therefore, the aims of this study were to describe the bacterial and fungal communities present in fecal samples from healthy pet dogs and pet cats using massive parallel 16S rRNA gene pyrosequencing, and to detect Bifidobacterium spp. and LAB in canine fecal samples using group-specific 16S rRNA gene primers. Since a similar analysis in cats was reported recently by our group (Ritchie et al., 2010), this procedure was not performed for the feline samples in the present study.

Materials and methods

Sample material

Fecal samples were collected from 12 privately owned dogs and 12 privately owned cats that were unrelated (Table 1). All animals were free of any clinical signs, received no medications that are expected to alter the gut microbiota (i.e. antibiotics), and were regularly vaccinated and dewormed. Owners were asked to estimate the body condition score of their animal based on a nine-point scale (Laflamme, 1997a, b). A complete blood count and serum biochemistry profile were analyzed at the Texas Veterinary Medical Diagnostic Laboratory (College Station, TX). Serum concentrations of cobalamin, folate, pancreatic lipase immunoreactivity (PLI), and trypsin-like immunoreactivity (TLI) were measured at the Gastrointestinal Laboratory (Texas A&M University) to rule out gastrointestinal or pancreatic disease. All animals were fed commercial diets, and none of the animals had undergone a diet change for at least 3 weeks before sample collection. One fecal sample per animal was collected immediately after spontaneous defecation, transported immediately to the laboratory, and frozen at −80 °C without any additives or pretreatment.

View this table:
1

Animals participating in the study

NumberGenderBreedAge (years)BCS
Dogs participating in the study
1fsLabrador Retriever-mix10.26
2mMaltese-Poodle-mix0.84
3fsBlue Heeler3.78
4fLabrador Retriever1.95
5mcMiniature Dachshund3.55
6mcRhodesian Ridgeback-mix2.84
7fsJack Russel Terrier-mix3.76
8fsPembroke Welsh Corgi0.95
9mcMongrel0.74
10mcBoxer5.07
11fsChesapeake Bay Retriever7.95
12mcLabrador Retriever6.95
Cats participating in the study
1mcDomestic Short Hair5.07
2mcMixed breed0.75
3mcDomestic Short Hair2.06
4mcDomestic Short Hair6.7nn
5mcDomestic Short Hair1.35
6fsDomestic Long Hair1.74
7mcDomestic Short Hair1.45
8fsPersian1.24
9mcDomestic Short Hair4.85
10mcDomestic Medium Hair2.85
11mcDomestic Short Hair4.26
12fsDomestic Short Hair3.06
  • Body condition score (BCS) was measured on a nine-point scale.

  • f, female; fs, female spayed; m, male; mc, male castrated; nn, not known.

The study protocol was approved by the Clinical Research Review Committee of Texas A&M University (CRRC#07-38), and informed consent was obtained from the owner of each animal before participating in the study.

DNA extraction

Genomic DNA was extracted from each fecal sample individually using a bead-beating method followed by phenol–chloroform–isoamyl alcohol extraction as described previously (Suchodolski et al., 2004).

Construction of group-specific 16S rRNA gene clone libraries in canine feces

The 16S rRNA genes were amplified using the group-specific primers Lac1-f 5′-AGCAGTAGGGAATCTTCCA-3′ and Lac2-r 5′-ATTYCACCGCTACACATG-3′ for producing LAB (Walter et al., 2001) and the genus-specific primers Bif164-f 5′-GGGTGGTAATGCCGGATG-3′ and Bif664-r 5′-CACCGTTACACCGGGAA-3′ for Bifidobacterium (Satokari et al., 2001). Primers were purchased from the Gene Technologies Laboratory, Texas A&M University. Amplification of DNA, cloning, and sequencing were performed as described elsewhere (Ritchie et al., 2010).

Massive parallel 16S rRNA gene pyrosequencing for bacterial and fungal organisms

Samples for bacterial sequencing were processed individually. For fungal sequencing, extracted DNA was pooled for dogs and cats, respectively, for economical reasons. Bacterial and fungal tag-encoded FLX-Titanium amplicon pyrosequencing (bTEFAP and fTEFAP) and data processing were performed at the Research and Testing Laboratory (Lubbock, TX) as described previously (Dowd et al., 2008). For this study, the fTEFAP and bTEFAP were processes based on the Titanium sequencing platform rather than FLX (Roche Applied Science, Indianapolis, IN). Titanium differs in that it generates average read lengths of 400 base-pairs (bp) rather than 250 bp as with the previous FLX. Also, the bacterial primers used differed from previous methods and extended from 27F numbered in relation to Escherichia coli 16S ribosome gene (forward28F: GAGTTTGATCNTGGCTCAG; reverse519R: GTNTTACNGCGGCKGCTG). Finally, rather than a double PCR utilized in the previous methods, only a single step reaction (35 cycles) was used and 1 U of HotStar Highfidelity Polymerase was added to each reaction (Qiagen, Valencia, CA). The fungal primers were based on the 18S subunit and sequencing was performed forward from 458F in relation to Candida albicans.

bTEFAP and fTEFAP sequence processing pipeline

Raw data from bTEFAP and fTEFAP were screened and trimmed based on quality scores (nominal phred20) and were binned into individual sample collections. After sequencing, individual collections of sequences were depleted of chimeras using B2C2 (http://www.researchandtesting.com/B2C2.html). The resulting files of short reads (<350 bp) were depleted, thereby ensuring that all sequences evaluated contained sufficient discriminating data from the V1–V3 hypervariable regions. Tentative bacterial species were identified using blastn in comparison with a curated high-quality 16S rRNA gene database derived from the National Center for Biotechnology Information (NCBI) as described previously (Dowd et al., 2008). The compiled data were used to determine the relative percentages of bacteria for each individual sample. Data were also compiled at each individual taxonomic level according to the NCBI taxonomy criteria as described previously (Dowd et al., 2008). Sequences with identity scores to known or well-characterized 16S sequences >97% identity (<3% divergence) were resolved at the species level, between 95% and 97% at the genus level, between 90% and 95% at the family level, and between 80% and 90% at the order level. Collection and sequence information is available through GenBank within a short read archive (SRA) under accession SRA012231.1.

Clostridium-cluster identifications were made using nearest neighbor techniques. Sequences were aligned against a set of known Clostridium-clusters (Krogius-Kurikka et al., 2009) to identify the cluster into which each sequence would most likely fall. The alignments were then further verified and analyzed to assess the total composition within each of the Clostridium-clusters for the sequences in question.

Statistical analyses

Percentages of bacterial sequences are given as mean ± SD.

To assess the diversity of the fecal microbiota, the Shannon–Weaver diversity index, the Simpson's reciprocal index, and evenness were calculated from the number of operational taxonomic units (OTUs) assigned at the genus and the species level as described by Steele (2005).

Rarefaction curves were calculated using the software program dotur (Schloss & Handelsmann, 2005). They estimate diversity and can therefore serve as an indicator for the completeness of the sampling method. A Richard's equation (Seber & Wild, 1989) was fit to the curves to predict the maximum number of OTUs that might be present in canine and feline feces as described by Suchodolski (2009).

Double dendrograms were generated using comparative functions and multivariate hierarchical clustering methods in NCSS 2007 (NCSS, Kaysville, Utah). The most abundant bacterial families and species are included in the final dendrograms with clustering based on Ward's minimum variance and utilizing Manhattan distance calculations with no scaling. It should be noted that the dendrogram linkages of the bacterial families and species are not phylogenetic, but clustered based on the relative abundance among animals. Thus, those samples with more similarity (less distance) are more closely related in overall bacterial diversity. Similarly, those bacteria that have similar percentages across all samples are more closely clustered.

Results

Animals

The results of the complete blood count, the serum biochemistry profile, and serum concentrations of cobalamin, folate, PLI, and TLI provided no indication of disease (results not shown).

Pyrosequencing for bacterial organisms

A total of 120 406 bTEFAP reads were analyzed across all animals with a mean of 5017 reads per fecal sample (range: 4050–5907). At 5% dissimilarity, a cutoff commonly used to describe the genus level (McKenna et al., 2008), a total of 85 OTUs were identified in dogs, with a mean of 30 ± 8.4 OTUs (range: 18–47) per dog. Of these, only 5 OTUs (6%) were detected in all 12 dogs, 22 OTUs (26.2%) were obtained from >50% of dogs, and 25 OTUs (29.8%) occurred infrequently and were found only in individual dogs. At the species level (3% dissimilarity; Stackebrandt & Goebel, 1994; McKenna et al., 2008), a total of 136 OTUs were found in the canine samples, with a mean of 39 ± 11.5 OTUs (range: 26–64) per dog. Of these, only 4 OTUs (2.9%) were identified in all dogs, 25 (18.4%) were detected in more than six dogs, and 52 OTUs (38.2%) were observed in individual dogs. In cats, a total of 113 OTUs were observed at the genus level, with a mean of 45 ± 8.4 OTUs (range: 35–67) per cat. Fourteen OTUs (12.4%) were found in all cats, 37 OTUs (32.7%) were obtained from more than six cats, and 34 OTUs (30.1%) were detected in individual cats. At total of 185 OTUs were obtained from the feline feces at the species level, with a mean of 61 ± 13.3 OTUs (range: 49–98) per cat. While 13 OTUs (7%) were detected in all 12 cats, 43 OTUs (23.2%) were observed in >50% of cats, and 70 OTUs (37.8%) were found in individual cats. Table 2 shows the numbers of observed OTUs, maximum predicted OTUs using the Richard's equation, the Shannon diversity index, the Simpson's reciprocal index, and the evenness calculated on the genus and on the species level.

View this table:
2

Diversity indices, number of observed OTUs per animal (mean ± SD), number of total observed OTUs, and maximum predicted OTUs at the genus level (5% dissimilarity) and the species level (3% dissimilarity), calculated using the Richard equation, identified in dogs and cats using massive parallel pyrosequencing

DogsCats
Genus level (5% dissimilarity)
Shannon diversity index1.76 ± 0.272.18 ± 0.28
Simpson's reciprocal index4.17 ± 1.306.46 ± 2.23
Evenness0.34 ± 0.050.42 ± 0.05
Mean observed OTUs per animal30 ± 8.445 ± 8.4
Total observed OTUs85113
Maximum predicted OTUs91422
Species level (3% dissimilarity)
Shannon diversity index1.82 ± 0.302.30 ± 0.30
Simpson's reciprocal index4.32 ± 1.446.99 ± 2.66
Evenness0.50 ± 0.060.56 ± 0.08
Mean observed OTUs per animal39 ± 11.560 ± 13.7
Total observed OTUs136195
Maximum predicted OTUs1812438

The rarefaction curves (Fig. 1) illustrate that the number of sequences collected in this study was adequate only at the genus level in dogs, because only the curve for 5% dissimilarity approached a plateau. The species and strain richness in dogs and the genus, species, and strain richness in cats were underestimated with the number of sequences obtained in the present study. Based on the Richard's equation, the fecal microbiota comprises approximately 180 OTUs in dogs and 2400 OTUs in cats at 3% dissimilarity.

1

Rarefaction curves for bacterial 16S rRNA gene sequences obtained by massive parallel pyrosequencing of canine (a) and feline fecal samples (b). The continuous line represents the genus level (dissimilarity of 5%), the dashed line represents the species level (dissimilarity of 3%), and the dotted line represents the strain level (dissimilarity of 1%). The number of sequences collected was adequate only at the genus level in dogs, because only the curve for 5% dissimilarity approaches a plateau. The species and strain richness in dogs as well as the genus, species, and strain richness in cats were underestimated by the number of sequences obtained in the present study.

Firmicutes was the most abundant bacterial phylum in both dogs and cats (dogs 95.36 ± 5.19% of all bacterial sequences, cats 92.10 ± 7.46%), comprising the classes Clostridia, Bacilli, and Erysipelotrichi. Over 65% of all bacterial sequences (69.35 ± 18.45% in dogs and 65.14 ± 23.11% in cats) belonged to the class Clostridia. This class was dominated by the genera Clostridium (22.73 ± 15.46% of all bacterial sequences in dogs and 20.42 ± 11.52% in cats) and Ruminococcus (17.37 ± 11.18% in dogs, 13.58 ± 9.02% in cats). Further prevalent genera within the class Clostridia were Dorea and Roseburia. The most prevalent Clostridium-clusters (Collins et al., 1994) were cluster XIVa (59.59 ± 23.2% of all Clostridiales sequences in the dogs, and 63.24 ± 16.18% of all Clostridiales sequences in the cats) and cluster XI (dogs: 33.64 ± 17.13%, cats: 24.25 ± 11.93%); see Supporting Information, Fig. S1. All bacterial species affiliated with the various Clostridium-clusters in canine and feline feces are listed in Table S2. Besides Clostridia, additional prevalent bacterial classes were Bacilli (17.14 ± 26.12% of all sequences in dogs and 9.39 ± 12.42% in cats) and Erysipelotrichi (13.01 ± 15.13% in dogs and 13.34 ± 11.24% in cats). The class Bacilli consisted almost exclusively of the order Lactobacillales in both dogs and cats. This order was dominated by the genera Streptococcus and Lactobacillus in dogs, and Enterococcus and Lactobacillus in cats. The class Erysipelotrichi consisted only of the order Erysipelotrichales. This order was mainly comprised of the genera Turicibacter, Catenibacterium, and Coprobacillus.

Bacteroidetes was the second most abundant phylum in dogs, comprising the genera Prevotella, Bacteroides, and Megamonas. Bacteroidetes were found at a prevalence of 2.25 ± 5.37% in dogs, but only 0.45 ± 0.69% in cats.

Actinobacteria was the second most abundant phylum in cats (7.31 ± 7.71%). In dogs, only 1.81 ± 1.32% of OTUs belonged to Actinobacteria. Most OTUs of this phylum belonged to the order Coriobacteriales. Within this order, all sequences were within the family Coriobacteriaceae, which consisted mainly of the genera Collinsella and Slackia in dogs. Eggerthella and especially Olsenella were more prevalent in cats. In addition to Coriobacteriales, the order Actinomycetales was seen in three dogs and two cats. One sequence identified as Bifidobacterium sp. was found in two cats each, while the genus Bifidobacterium was not detected in dogs.

The phylum Fusobacteria was found at an abundance of 0.3 ± 0.43% in dogs and 0.04 ± 0.05% in cats. In both animal species, all OTUs within this phylum were identified as Fusobacterium sp.

The most common bacterial orders in the canine and feline fecal samples are summarized in Fig. 2. The ten most abundant bacterial families identified are depicted as a double dendrogram in Fig. 3. Table S1 lists all bacterial genera identified in canine and feline feces, their abundance, and the number of animals harboring each genus. The most abundant species found in dogs and cats are shown in a double dendrogram in Fig. S2.

2

Composition of the fecal bacterial communities in dogs and cats at the order level. The shade of the bars represents the relative abundance of the most abundant bacterial orders (>3% of all bacterial sequences) in canine and feline fecal samples investigated using 16S rRNA gene pyrosequencing.

3

Double dendrogram of major bacterial families. This double dendrogram, based on the Ward minimum variance clustering method, shows the 10 most abundant bacterial families identified in feces from healthy dogs and cats investigated using 16S rRNA gene pyrosequencing. The heat map indicates the relative percentage of each family within each fecal sample. On top of the figure, the distance of the samples based on weighted pair linkage and Manhattan distance methods with no scaling is shown, along with a distance score. The 10 most abundant bacterial families are provided on the y-axis with their associated distance scores.

Pyrosequencing for fungal organisms

For fungal organisms, a total of 5359 sequencing reads (2917 in the canine pool and 2442 in the feline pool) were analyzed. In the cats, all OTUs were affiliated with the phylum Ascomycota. In dogs, 99.62% of the sequences were classified as Ascomycota. Ten sequences were identified as Basidiomycota, class Exobasidiomycetes, family Microstromataceae, three sequences were classified as Glomeromycota (Glomus sp.), and one single sequence was found to be Mucorales.

The most abundant fungal class in both dogs and cats was Saccharomycetes (dogs: 85.46%; cats: 58.56%). In dogs, this class was mainly comprised of the genus Nakaseomyces (76.72%). Candida castellii was the most abundant species in dogs. In cats, the family Saccharomycetaceae consisted solely of the genus Saccharomyces (58.31% of all fungal sequences). Based on our similarity cutoff of 97%, this OTU could not be classified on the species level (Saccharomyces cariocanus or Saccharomyces cerevisiae). The second most abundant fungal class in dogs (6.96%) was an unidentified class closely related to Pleosporales, Taphrinales, and Trichosphaeriales. This class was not found in cats. In cats, the second most abundant class was Eurotiomycetes (23.87%), which was found at a much lower abundance in dogs (3.94%). In both dogs and cats, the class Eurotiomycetes consisted of the family Trichocomaceae, comprised of the genera Aspergillus and Penicillium.

A total of 33 different OTUs were observed in the canine feces at the genus level, of which 21, however, could not be classified and may represent previously uncharacterized fungi. By far the most abundant genus in dogs (76.7%) was Nakaseomyces. In the feline feces, 17 OTUs were found at the genus level, of which five could not be identified. The most prevalent genera in cats were Saccharomyces (58.3%) and Aspergillus (11%). Table 3 shows the percentages of all fungal sequences affiliated with the identified genera in the canine and feline feces. At the species level, 44 OTUs were identified in the pooled canine sample and 29 OTUs in the feline sample.

View this table:
3

Fungal genera found in pooled canine and feline feces using massive parallel pyrosequencing

Fungal genus% Canine sequences% Feline sequences
Aspergillus1.2010.97
Aspergillus/Eupenicillium0.030.12
Aspergillus/Penicillium00.04
Bullera0.410
Circinomucor/Mucor0.030
Cladosporium0.030
Cryptococcus0.380
Dipodascus/Geotrichum0.030
Geosmithia/Saccharomyces0.147.37
Glomus0.100
Lulworthia0.030
Microstroma0.340
Nakaseomyces76.720.25
Nakaseomyces/Saccharomyces0.170
Neosartorya00.04
Penicillium06.59
Phialocephala04.91
Pichia0.140
Plectosphaerella01.56
Pyrenochaeta0.100
Rhizidium/Schizosaccharomyces0.030
Saccharomyces2.5458.31
Schizosaccharomyces1.510
Thermomyces03.69
Thysanophora0.070
Trematosphaeria0.480.04
Zygosaccharomyces0.310
Not identified16.196.14
  • The percentages of pyrosequencing tags that were affiliated with fungal genera (5% dissimilarity) in pooled canine and feline fecal samples are listed. Where two names are given, this OTU could not be affiliated with one known genus, and the two next known neighbors are listed.

Group-specific 16S rRNA gene clone libraries for LAB and Bifidobacterium spp.

To identify the sequences belonging to LAB and the genus Bifidobacterium in canine fecal samples, group-specific 16S rRNA gene clone libraries were constructed (Table 4). Sequences belonging to LAB were found in eight dogs, with 17 different OTUs being identified. The most abundant species represented a previously uncultured Lactobacillus, which was found in four dogs. The second most prevalent species were Lactobacillus aviarius and Weissella cibaria, which were identified in three dogs each. Bifidobacterium spp. were recovered from 10 dogs, and eight different species were identified. The most prevalent species were Bifidobacterium bifidum and Bifidobacterium subtile, which were detected in three dogs each.

View this table:
4

LAB and Bifidobacterium spp. identified in feces from dogs using genus-specific 16S rRNA gene clone libraries, their similarity to the closest neighbor in the NCBI database, the number of dogs harboring this species, and the percentage of clones identified

SpeciesSimilarityNumber of dogs% of clones
Lactic acid-producing bacteria
Uncultured Lactobacillus100416
Lactobacillus aviarius99314
Weissella cibaria10037
Lactobacillus paracollinoides9425
Lactobacillus secaliphilus9925
Uncultured10025
Lactobacillus ruminis100116
Lactobacillus algidus10015
Uncultured9915
Lactobacillus sanfranciscensis10015
Uncultured10015
Lactobacillus composti9612
Lactobacillus jensenii9612
Lactobacillus reuteri10012
Lactobacillus salivarius10012
Uncultured10012
Bifidobacterium spp.
Bifidobacterium subtile100349
Bifidobacterium bifidum10039
Bifidobacterium animalis100240
Bifidobacterium longum100230
Bifidobacterium catenulatum100123
Bifidobacterium pseudolongum99114

Discussion

Previous studies using cultivation-based methods (Benno et al., 1992; Johnston et al., 2001; Mentula et al., 2005) and 16S rRNA gene or chaperonin 60 gene-based studies have described the intestinal microbiota of dogs and cats (Suchodolski et al., 2008a; Desai et al., 2009; Middelbos et al., 2010; Ritchie et al., 2010). Using massive parallel 16S rRNA gene pyrosequencing, the results of the current study have added additional information about the composition of the bacterial and fungal microbiota of individual healthy pet dogs and pet cats. Our results demonstrate that dogs and cats harbor a highly diverse intestinal microbiota. The feline fecal bacterial microbiota appears to be more diverse, as shown by the higher number of OTUs found at the genus and species levels and by the bacterial diversity indices evaluated. At the same time, less interindividual differences in the abundance of most bacterial groups were seen in cats, and more cats than dogs shared the same bacterial genera. In the fungal microbiota, the genus Nakaseomyces was predominant in dogs, while Saccharomyces, Aspergillus, and Penicillium were prevalent in cats. At this stage, we can only speculate that these differences in the bacterial and fungal microbiota might be due to the adaption of the intestinal microbiota to the diet and gut morphology of the host. The microbiome of the omnivorous canids might be more diverse to cope with a varied diet than the microbiota of the strictly carnivorous felids.

The predominant bacterial phylum found in the fecal samples from both dogs and cats was Firmicutes, with a mean abundance of >92%. Bacteroidetes was the second most abundant phylum in dogs, with a mean abundance of 2.3%. The only other investigation using pyrosequencing on fecal samples obtained from research dogs (Middelbos et al., 2010) reported a lower abundance of Firmicutes (25%) and a higher prevalence of Bacteroidetes and Fusobacteria of 35% and 40%, respectively. Further investigations have to clarify to what extent these discrepancies are due to variances in the methodology. In the present study, bead-beating was used for DNA extraction, while Middelbos (2010) avoided a bead-beating step. A recently published comparison of various methods for DNA extraction from human feces (Salonen et al., 2010) revealed that mechanical cell disruption by repeated bead-beating resulted in an improved extraction efficiency and a higher bacterial diversity compared with chemical or enzymatic cell lysis.

Actinobacteria was the second most abundant phylum in cats (7.3% of all bacterial sequences), but was at a low abundance in dogs (1.8%). A similar low abundance of this phylum was reported previously in canine feces (Middelbos et al., 2010). However, a chaperonin 60 gene based clone library revealed a high abundance of Actinobacteria in feline feces (Desai et al., 2009). It is possible that the prevalence of Actinobacteria in the mammalian gut is currently underestimated (Krogius-Kurikka et al., 2009; Ritchie et al., 2010). It has been suggested that some 16S rRNA gene primers may underestimate the prevalence of Bifidobacterium spp. (Dethlefsen et al., 2008; Ritchie et al., 2010). Furthermore, 16S rRNA gene-based approaches using universal bacterial primers might underestimate the abundance of Actinobacteria, because the high G and C contents of their DNA could hamper the dissociation of the DNA strands during PCR. It is also possible that bacterial groups with a high number of 16S rRNA gene operons may be overestimated with the use of universal primers (Rastogi et al., 2009). This might also explain why in the present study, Bifidobacterium spp. were detected in dogs only with the genus-specific clone library, not by pyrosequencing. The possibility of adding group-specific primers for these otherwise under-represented groups (Dethlefsen et al., 2008; Ritchie et al., 2010) should be considered in future studies.

The most abundant bacterial class found in our study in both dogs and cats was Clostridia, with Clostridium-cluster XVIa being predominant. This is in agreement with previous findings in the feces (Middelbos et al., 2010) and the colon of dogs (Suchodolski et al., 2008a). Clostridium-cluster XIVa further showed the highest species variety, followed by Clostridium-cluster IV. Members of these two clusters produce butyrate and other short-chain fatty acids, which are important sources of energy for colonic epithelial cells. A consistent depletion of Clostridium-clusters XIVa and IV in humans with chronic intestinal inflammation (Sokol et al., 2006; Frank et al., 2007) indicates the important role of these bacterial groups in gastrointestinal health in humans. It has been shown that these groups are also depleted in dogs with IBD (Suchodolski et al., 2009).

Several potential pathogens were detected in our analyzed samples. Clostridium perfringens was found in four dogs and nine cats, Enterococcus spp. were identified in five cats and 11 dogs, E. coli was detected in four dogs and three cats, and Helicobacter spp. were found in two dogs and four cats. All animals participating in our study were clinically healthy and showed no signs of gastrointestinal disease. These findings demonstrate that bacterial groups that are considered potential pathogens are part of the microbiota of healthy animals, and their exact role and pathomechanism in gastrointestinal disease requires further study.

Information on fungal organisms colonizing the gastrointestinal tract of pets is currently limited. The presence of yeasts and molds has been anecdotally described in dogs using cultural methods (Benno et al., 1992; Mentula et al., 2005). PCR revealed fungal DNA affiliated with two phyla (Ascomycota and Basidiomycota) in the duodenum of dogs (Suchodolski et al., 2008b). In the present study, we identified 66 fungal genera in canine feces belonging to the phyla Ascomycota, Basidiomycota, Glomeromycota, and Zygomycota, and 16 genera in the feline feces, all from the phylum Ascomycota. However, information in interindividual variances in fungal populations cannot be provided by the present study, because feces from dogs and cats were pooled because of economic reasons. Differences in the diversity and composition of the fungal ecosystem between patients with chronic intestinal inflammation and healthy individuals were found in humans and dogs (Ott et al., 2008; Suchodolski et al., 2008b), but the role of fungal communities in intestinal physiology is still poorly understood. Therefore, further research on fungal communities in the intestinal mucosa, chyme, and feces of healthy and diseased individuals is warranted. Future studies will need to evaluate differences in fungal populations in individual animals and also evaluate the fungal microbiome in diseased individuals.

The investigation of the presence of LAB, such as Lactobacillus spp. and Bifidobacterium spp., in the digestive tract of healthy animals is of particular interest as they are most commonly used in probiotic preparations. A recent study (Stecher et al., 2010) suggested that the chances for a new incoming species to colonize the gut are higher if closely related species are already present. A previous study has shown that members of the order Lactobacillales seem to be highly prevalent in the duodenum, jejunum, and colon of dogs (Suchodolski et al., 2008a). In the present investigation, we detected 17 different species of LAB in canine feces. LAB appear to be prevalent in the gut microbiota of most dogs, because they were detected in eight out of 12 dogs in the current study. However, we observed high animalindividual differences in the LAB sequences identified. To the authors' knowledge, this is the first study to evaluate the prevalence of Bifidobacterium spp. in canine fecal samples. Ten out of 12 dogs included in our study were positive for Bifidobacterium spp. A total of eight different species were identified, with B. subtile and B. bifidum being the most prevalent species. Similar to LAB, we also observed a high variability in the observed Bifidobacterium spp. between individual dogs, suggesting that each dog harbors a unique LAB and Bifidobacterium population. It has to be mentioned that there are several other bacterial species that are currently applied as probiotics and that cannot be detected with the primers used for the clone libraries in the present study, such as Enterococcus faecium. However, members of the genus Enterococcus were detected using universal pyrosequencing primers in four dogs and 11 cats (Table S1).

In conclusion, this study is the first to extensively describe the bacterial and fungal communities in feces from healthy cats and dogs using massive parallel 16S rRNA gene pyrosequencing. We observed a high diversity of bacterial and fungal organisms in fecal samples. We further demonstrated that the majority of dogs harbor LAB and Bifidobacterium spp., but individual dogs harbor specific organisms.

Supporting Information

Fig. S1. Most prevalent Clostridium-clusters found in canine and feline fecal samples, using 16S rRNA gene pyrosequencing, with a prevalence of >1% of all Clostridiales sequences.

Fig. S2. Double dendrogram of major bacterial species in canine and feline feces.

Table S1. Bacterial genera in canine and feline fecal samples identified using 16S rRNA gene pyrosequencing.

Table S2. Species affiliated with Clostridium-clusters in canine and feline fecal samples using 16S rRNA gene pyrosequencing.

Please note: Wiley-Blackwell is not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

Acknowledgement

Financial support for this study was provided in part by Nutramax Laboratories Inc. (Edgewood, MD).

References

View Abstract