Introduction

The presence of a core microbiome in the gut (i.e. key microbes that are present in all or the majority of individuals within a population) has been much debated [1-3]. It has been suggested that if a core community could be identified across all healthy members of a species this may provide a basis for disease diagnosis, prevention and possibly provide therapeutic targets [4,5]. Since the concept was initially posed there has been speculation as to whether a core may be gene or functionally based [1,6]; and while an apparent core microbiome in the human gut has been demonstrated in several studies no clear definition of what constitutes a core and how large the core is in any community exists [3].

The human gut microbiota appears to be relatively stable in composition over time, with individuals maintaining a similar microbial profile whilst on a uniform diet with no gastrointestinal upset [7,8]. In horses it is apparent that the microbiome of the hind gut changes rapidly in early life from birth up until 56 days of age [9] and even in adult animals may not be stable. Indeed recent work using terminal restriction fragment length polymorphism demonstrated a statistically significant (P<0.001) bacterial community in the gut after a ten week period in six ponies fed a constant high fibre based diet [10]. This would suggest that the equine gut microbial community may not be as stable as that found in the human gut. The horse receives much of its dietary energy through microbial hydrolysis and fermentation of fibre predominantly in the large intestine/hindgut [11] as the microbiota ferment fibre to yield volatile fatty acids which can be absorbed and utilised as energy sources [12-14].

Horses are however extremely susceptible to gastrointestinal disturbance when subjected to sudden changes in diet which can result in alteration of fermentation patterns and ultimately metabolic disorder. Colic and laminitis commonly occur with colic being the biggest cause of mortality in the horse [15,16]. Colic occurs in almost 6% of thoroughbred training facilities in the UK [17] and is the cause of almost 15% of equid deaths in the USA [18]. The prevalence of laminitis in the UK is poorly documented but between 0.5 and 3% of the equid population has been reported as suffering from the condition [19,20]. When horses are challenged by the sudden incorporation of high levels of soluble carbohydrate in the diet (e.g. starch from grain or starch/simple sugars/fructan from pasture) normal enzymatic breakdown in the small intestine is overloaded and large amounts of readily fermentable starch/sugar pass to the large intestine where microbial fermentation occurs [21-23]. Rapid fermentation occurs with a rapid accumulation of lactic acid producing bacteria such as Lactobacilli and Streptococci occurring [21,22,24,25]. High amounts of lactic acid and gas are produced causing a rapid drop in hindgut pH and death/ suppression of the normal fiber degrading bacteria. These events are thought to contribute in some instances to impaction colic due to excess gas accumulation combined with dehydration of digesta [21,26] and may also help to explain some instances of laminitis [25,27].

Recently it has been suggested that microbial factors might influence the development and progression of chronic laminitis and colic [28,29]. In particular Steelman and colleagues [28] found a higher bacterial diversity in horses with chronic laminitis but it is not clear if this represented a less stable microbiota in these animals. Very little attention however, has been given to the question of whether horses have a core gut microbiome, although the recent publication by Costa and colleagues [29] found common operational taxonomic units (OTUs) across several horses. Here we have investigated the possibility that either 1) the horse has an intrinsically unstable gut microbial fermentation because it has an unusual core bacterial community that is large and dependent on only a few key bacterial phyla or 2) alternatively there is no core or a core made up of many low abundance species leading to an intrinsically unstable bacterial community in the equine hindgut.

Materials and Methods

Results

One million, four hundred and sixty thousand, one hundred and twenty four sequences of average length 358bp were obtained from the 454 FLX Titanium sequencing. Quality filtering resulted in 559 623 high quality sequences providing 32 250 sequences per gut region after normalisation. Sequences were clustered into 4818 unique OTUs across the complete data set. Rarefaction curves (which were calculated from non-normalised data, Figure S1) showed that for each gut region curves had not plateaued indicating that complete sampling of these environments had not yet been achieved. Good’s coverage estimates, however, indicate that a large part of the diversity in all regions had been captured with the average coverage by gut region being 95.7% (s.d. 0.9) and by animal also 95.7% (s.d. 1.0) (Figures S2 & S3).

The bacterial community within the different regions of the horse gut was found to be highly diverse and even (as indicated by the Simpson and Shannon-Weiner diversity indices, see Table 1) with significant differences between gut regions (P<0.001). The microbiota of the ileum was significantly less diverse than that of any region of the large intestine, with little variation between the different large intestine regions. Similarly the species richness (number of OTUs) and bacterial load (determined by q-PCR) was lowest in the ileum. Although not significantly different across the large intestine regions, species richness was highest in the RVC, with the quantity of DNA increasing progressively through the intestinal tract, being highest in the faeces.

Based on classification of recovered OTUs, differences between the ileum and the large intestine can be explained by the most abundant phyla (Figure 1). In all regions of the large intestine the most dominant phyla were the Firmicutes (46%) and Bacteroidetes (43%). In the ileum the Firmicutes (70%) were the most dominant followed by Proteobacteria (14%) then Bacteroidetes (10%). The pattern of phyla seen in Figure 1 shows a high degree of similarity throughout the large intestine. Figure 2 shows a 2D heatmap of OTUs present at 0.1% or more of the total number of sequences across all samples, clustered by sample as well as by OTU. It can be clearly seen that a specific group of OTUs cluster in all ileal samples and a distinct and different signature exists for the proximal regions of the large intestine (caecum, RVC & LVC) and the distal regions (RDC, SC & faeces) with the LDC appearing in both groups. Although the regions of the gut cluster as detailed there also appears to be a large amount of inter-animal variation within these groupings.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Phyla identified and relative proportion of each in different compartments of the horse’s large intestine.

Data shown from the Ileum, caecum, right ventral colon (RVC), left ventral colon (LVC), left dorsal colon (LDC), right dorsal colon (RDC), small colon and faeces. The category others includes; TM7, Tenericutes, synergistetes, Cyanobacteria/Chloroplast, SR1, Elusimicrobia, Fusobacteria, Chloroflexi.

https://doi.org/10.1371/journal.pone.0077660.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Heatmap showing OTUs in different compartments of the horse’s large intestine.

Data shown from the ileum (A) caecum (B), right ventral colon (C), left ventral colon (D), left dorsal colon (E), right dorsal colon (F), small colon (G) and faeces (H) found at 0.1% or greater abundance as a percentage of the total number of sequences (accounts for 49.28% of the total sequences). The cladogram on the vertical axis shows the phylogenetic relationship between the individual OTUs included in this analysis. The cladogram on the horizontal axis shows the phylogentic relationship between all the OTUs identified in each sample.

https://doi.org/10.1371/journal.pone.0077660.g002

In every gut region a number of OTUs have been identified which make up the core community (found in all animals). To account for ambiguity surrounding the definition of bacterial classification OTUs were clustered at 90%, 95%, 97% and 99% similarity (Table 2) allowing the size of the core community to be shown in terms of a percentage of the total number of OTUs identified at each level of similarity. As the clustering threshold increases the total number of OTUs identified increases and as such the size of the core decreases. At all levels of clustering similarity the core community is smallest in the ileum and largest in the caecum. When taking account not only of the number of OTUs making up the core but also the number of occurrences within each OTU a different impression of the core microbiota can be seen (Figure 3, OTUs clustered at 97% similarity and present at 0.1% of community or greater). As with all previous data the community within the ileum is different from that of the large intestine having a core made up of approximately 32% of all sequences but comprised of only seven OTUs of varying abundance, including one OTU responsible for 11% of all sequences. The core in the large intestine, however, is much smaller (5-15% of the total number of sequences) and shows a different pattern to that of the ileum with a much larger number of OTUs being present but in low abundance (the largest single OTU in any region of the large intestine accounts for only 2% of all sequences for that region).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. The core bacterial community in different compartments of the horse’s large intestine.

The core community is defined as those OTUs (clustered at 97% similarity) present in all ten animals and which abundances are 0.1% (or greater) of the total number of sequences in that gut region. The lower pale blue section of the bar indicates the proportion that is not part of the core. The remaining individual coloured sections represent each OTU of which the core is comprised; Ileum (7), caecum (31), right ventral colon (RVC) (33), left ventral colon (LVC) (28), left dorsal colon (LDC) (12), right dorsal colon (RDC) (18), small colon (16) and faeces (25). For further details see table S3.

https://doi.org/10.1371/journal.pone.0077660.g003

When classified to family level (See Table 3) it can be seen that the core community in the ileum is dominated by Lactobacillaceae a member of the Firmicutes phyla (16.4%) followed by Pasteurellaceae belonging to the phyla Proteobacteria (8.57%) with only four families making up the core. In the regions of the large intestine the core is comprised of more families (from 6 in the SC to 11 in the RVC & RDC) and the dominant families change in a similar way to the pattern described previously in Figure 2. The core in the regions of the proximal large intestine (Caecum, RVC & LVC) is dominated (in order of largest to smallest abundance) by an unclassified family belonging to the order Bacteroidales, the family Lachnospiraceae (Firmicutes phyla), Prevotellaceae (Bacteroidetes phyla), Erysipelotrichaceae (Firmicutes phyla), Ruminococcaceae (Firmicutes phyla) and Fibrobacteraceae (Fibrobacteres phyla). The core in the regions of the distal large intestine (RDC , SC & faeces) is then dominated by (in order of largest to smallest abundance) Prevotellaceae, Fibrobacteraceae, Lachnospiraceae, unclassified family only classified to phyla level as Bacteroidetes and Clostridiaceae 1 (Firmicutes phyla). The LDC is dominated by the Lachnospiraceae, Clostridiaceae 1, unclassified family belonging to the order Bacteroidales and Erysipelotrichaceae. The way in which the bacterial families representing the core bacterial community in the large intestine change in terms of relative abundance through the gut can be seen in Figure 4. This clearly illustrates the distinct change in the bacterial community between the ileum and the hind gut and at the pelvic flexure separating the proximal and distal large intestine.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Relative abundance of the most prevalent bacterial families identified as the core bacterial community.

Regions of the large intestine included are the caecum, right ventral colon (RVC), left ventral colon (LVC), left dorsal colon (LDC), right dorsal colon (RDC), small colon and faeces. Values presented do not account all for members of stated families found in each gut region only those identified as being part of the core; error bars represent the standard deviation. A-Lactobacillaceae B-Clostridiales_Insertae Sedis XIII C- Coriobacteriaceae D-Acidaminococcacaea E-Porphyromondaceae F-Spirochaetaceae G-Clostridiaceae 1 H-Erysipelotrichacea I-Fibrobacteraceae J-Ruminococcaceae K-Lachnospiraceae L-Prevotellaceae.

https://doi.org/10.1371/journal.pone.0077660.g004

Discussion

The herbivore gastrointestinal tract harbors an abundant and diverse microbial community which provides the host with essential nutrients through degradation and fermentation of complex carbohydrates from the diet that would otherwise be unavailable [38,39]. In addition, to this primary function, the microbial community serves the host by modulating immune function and maintaining health [40]. The horse receives much of its dietary energy through hydrolysis and fermentation of fibre in the large intestine which comprises of large chambers including the caecum and the colon. In these regions the microbiota ferments fibre to yield volatile fatty acids and lactic acid which can be absorbed and utilised as energy sources [12-14]. Despite the importance to the horse of fermentation in the large intestine this system is easily perturbed. The largest factor influencing the microbiota is the diet of the animal; work by Ley and colleagues [39] demonstrates phylogenetic separation of carnivores, omnivores and herbivores. In the horse differences in bacterial number, composition and activity are known to occur when feeding a fibre based compared to a starch rich diet [22,41,42].

In previous work we have demonstrated that there are differences in the microbial community found in the caecum compared to the RDC and faeces of horses and ponies [37]; this has been reinforced by the current work. This change coincides physiologically with the pelvic flexure between the LVC and LDC where the intestine significantly narrows and folds back on itself. This anatomical feature has a marked effect on digesta passage and is one of the most common locations for impaction colic in the horse [43,44]. Here we show, similar to the human gut [45], that the bacterial diversity in the ileum of the equid is lower than that found in the large intestine. The diversity and species richness in the large intestine increases from the caecum to be highest in the RVC before declining through the remainder of the gut to the faeces, confirming our previous findings [37].

Firmicutes and Bacteroidetes tend to be the predominant phyla in most mammalian gut environments studied to date [38,46,47]. A higher ratio of Bacteroidetes to Firmicutes has been linked with conditions such as obesity [48] and IBD [49]. In the human colon and faeces the microbiota is approximately 50-95% Firmicutes, 3-50% Bacteroidetes and 1-3% Actinobacteria [4,38,50,51]. In the horse published work utilising next generation sequencing is only starting to appear and is limited to a few studies, using faeces, that are of variable quality in terms of sample replication and sequencing depth [28,29,52]. From these early studies the dominant phylum in the horse appears to be Firmicutes (43-69%), with inconsistency over the next most abundant (Bacteroidetes 14.2%, Proteobacteria 10.2% [29], Proteobacteria and Verrucomicrobia 4.1% each, Bacteroidetes 3.65% [52] or Verrucomicrobia 18.1%, Bacteroidetes 5.7% [28], with Spirochaetes and Actinobacteria also identified. Older culture independent work that utilised cloning or probe based methods describe a similar profile [53-55] but with much higher levels of Bacteroidetes (45-49%) identified in some studies [55,56]. Our work identifies a phyla profile (from animals on a grass based diet) that is reasonably consistent across all regions of the large intestine with the most dominant phylum being Firmicutes (average 46%), Bacteroidetes (average 43%) followed by Fibrobacteres, Spirochaetes and Proteobacteria (all <4%) with a notable lack of Verrucomicrobia (similar to Costa and colleagues) [29]. We also found that the ileum, although still dominated by the Firmicutes (70%), had a high abundance of Proteobacteria (14%) and a lower amount of Bacteroidetes (10%); a pattern shown in the human ileum [45,57] and in the gastric mucosa of the horse [58].

The presence of a core microbial community in the human gut has been identified but it differs in size dependent on the study design. A shared core was only identified in 50% of individuals [51] whereas Hamady and Knight [59] proposed that a core may exist within sub-communities rather than in the entire community. Another suggestion was that a core may not exist as a phylogenetic one but at a functional level [1]. More recently core species were identified but at massive sequencing depth or by using different methods of OTU clustering and classification [2,3,60]. What is clear is that what constitutes a core microbiome has not been well defined. The sequencing depth used may restrict identification of the presence of a core and thus the true size of that core, similarly OTU clustering methods may influence detection of core members [3,4,60]. Although varying numbers of phylotypes/ species/ OTUs have been identified as potentially belonging to the core community in the human there is a common thread identifying Clostridia as the most prevalent class of bacteria in the core [2-4,45,51]. Members of this class identified as being part of the core are Ruminococcaceae, Lachnospiraceae, Clostridiaceae, Streptococcaceae [2-4,51]. Bacteroidetes have also been shown to be core but at low levels [4,51]. Lachnospiraceae in particular have been shown to exist in most samples [2-4]. When Sekelja and colleagues [3] analysed sequence data from Ley and others [8] they found this family across many mammals including 71% of the order Perissodactyla included in the study suggesting this may appear in the core of most mammals. Interestingly low levels of this family in the human gut have been implicated in IBD [49]. This role is perhaps not surprising given that Lachnospiraceae are prolific producers of butyrate [4,61] and butyrate is known to have a protective function on colonocytes in the gut wall [62,63].

To the authors’ knowledge only one study has commented on the possibility of a core bacterial community in the horse gut [29]. They identified 123 shared OTUs from the faeces of 4 horses (out of 1620 identified in total) but of those only 6 had an abundance of greater than 25 occurrences per animal. They classified 5 of these OTUs as Lachnospiraceae with the remaining one being unclassified. This is larger than the core that we have identified in faeces (25/2566 in this study) however we have selected only those occurring at least once in every animal and at 0.1% or greater of the total sequences, so if treated in a similar way to Costa and others [29] it is likely that a higher number of OTUs would have been ‘core’. This again highlights the lack of a universal definition as to what constitutes a core. A study in cattle of the core community identified from faecal samples was dominated by Prevotella spp followed by the Lachnospiraceae family, Faecalibacterium spp then Ruminococcus spp [64]. The most abundant family we identified as belonging to the core in equid faeces was Prevotellaceae followed by Ruminoccocaceae, Fibrobacteraceae then Lachnospiraceae which is similar to what was reported in cattle [64,65] and we also noted a high amount of Clostridia similar to the human literature described earlier. The fact that both our work and that of Costa and others [29] identified high levels of Lachnospiraceae would seem sensible given the work of Sekelja and others [3], discussed earlier. Unfortunately, as Costa and colleagues [29] do not state what all of the shared OTUs were classified as, it is not possible to compare this with the prevalence of the other families we identified as part of the core. As the most dominant core bacteria change through the large intestine it would be of interest to investigate their functional significance as to whether the role these bacteria play alter in a similar way .

We have shown for the first time that a phylogenetic core bacterial community exists in all regions of the large intestine of healthy horses on a fibre based, grass diet. This core community is smaller than found in the rumen of the cow [65] but unlike most other core communities that have been identified from other environments is not dominated by any particular OTUs. In the oral microbiota of both the dog and human a small core is seen but with several highly abundant OTUs [66,67] and in the porcine tonsils a larger core was observed but again with a few highly dominant OTUs [68]. These other environments are notably less susceptible to disturbance than the gut of the horse and as such the fact they appear to have ‘key’ members of their core which are highly prevalent (compared to the horse hindgut where the core is made up of many low abundance OTUs) may suggest an explanation as to why the equine hind gut is so vulnerable to change when challenged leading to an alteration in fermentation patterns and consequent metabolic disorders.

Further consideration must now be given to the functional role of these bacteria in order to establish how much functional redundancy exists and which bacteria are essential to maintain normal digestive function and health. By knowing what makes up the core community comparisons can now be made to different diets and when metabolic disorders (such as Laminitis and colic) are encountered allowing suggestion to be made regarding prophylaxis, diagnosis and treatment.

Supporting Information

Acknowledgments

We would like to thank Gary Easton and Dr. Justin Pachebat for assistance with sample collection (both IBERS, Aberystwyth University).