Characterization and comparison of the bacterial microbiota in different gastrointestinal tract compartments of Mongolian horses

Abstract The intestinal microbiota plays an important role in the health and metabolism of the host. Next‐generation sequencing technology has enabled the characterization of the gut microbiota of several animal species. We analyzed the intestinal microbiota in six different parts of the gastrointestinal tracts (GITs) of five Mongolian horses by sequencing the 16S rRNA gene V3‐V4 hypervariable region. All horses were kept in the natural habitat of the Inner Mongolia grassland. Significant differences were observed among the microbiota compositions of the distinct GIT regions. In addition, while the microbial community structures of the small and large intestine were significantly different, those of the cecum and colon were similar. In the foregut, Firmicutes (65%) and Proteobacteria (23%) were the most abundant, while Firmicutes (45%) and Bacteroidetes (42%) were the most common in the hindgut. At the level of family, Ruminococcaceae (p = .203), Lachnospiraceae (p = .157), Rikenellaceae (p = .122), and Prevotellaceae (p = .068) were predominant in the hindgut, while the relative abundance of the Akkermansia genus (5.7%, p = .039) was higher in the ventral colon. In terms of the putative functions, the ratio of microbial abundance in the different parts of the GIT was similar, the result can help characterize the gut microbial structure of different animals.


| INTRODUC TI ON
The horse is a herbivorous nonruminant animal with highly compartmentalized gastrointestinal tract (GIT), which can utilize a variety of plant fibers (Harris et al., 2017;Santos, Rodrigues, Bessa, Ferreira, & Martin-Rosset, 2011). Each segment of the GIT has an independent ecosystem with unique biotic and abiotic (temperature, water, pH, oxygen, etc.) characteristics. The composition (diversity and structure) and function (metabolic mechanism and end products) of the GIT microbiome are highly significant to animal health and metabolism. In normal circumstances, the gut microbes and host are in the symbiotic and highly dynamic relationship. In horses, for example, 60%-70% energy comes from volatile fatty acids (VFAs) (Argenzio, 1975;Vermorel & MartinRosset, 1997) produced by the cecum and colon microorganisms, 30% of which is produced by the cecum microbiota alone (Glinsky, Smith, Spires, & Davis, 1976). Therefore, the balance and stability of the intestinal microbiota are essential for the health and function of GIT. Several diseases of the GIT are related to change in the composition or function of its microbiota. In addition, metabolic diseases, such as laminitis that can affect the musculoskeletal system, are also related to the intestinal microbiota (Milinovich et al., 2007;Steelman, Chowdhary, Dowd, Suchodolski, & Janecka, 2012).
The Mongolian horse is one of the most ancient grassland horse bred in the world and found in Inner Mongolia, China. At present, studies of the intestinal microorganisms of Mongolian horses have been limited in feces (Zhao et al., 2016). Horse feces can only represent the microbial changes in the distal regions of the posterior intestine (Costa, Silva, et al., 2015;Dougal et al., 2012) rather than the whole gastrointestinal microflora, and this had been demonstrated by studies of human intestinal microflora (Durban et al., 2011;Eckburg et al., 2005). In this study, we analyzed the characterization of the microbial composition of different parts of the Mongolian horse GIT by using the next-generation sequencing (NGS) firstly.

| Horses and sample collection
Five healthy Mongolian horses (three males and two females with an average age of 4.4 years ranged from 3 to 6 years and weight of 292.8 ± 11.9 Kg) grazed in the Xilin Gol League prairie in Inner Mongolia Autonomous Region, and horses were euthanized in October and November 2017. All horses came from the same pasture fence, maintained in same grazing condition, and were fed by same pasture. The dry matter intake (DMI) horse is 16.51 kg day −1 per Mongolian (Table A1) (Wei et al., 2015). The animals were examined by a veterinarian to confirm there were no obvious metabolic and gastrointestinal disorders. After euthanasia and dissection, all organs of the gastrointestinal tract were collected by tying up the narrow interface between each segment with ropes, the middle of each segment was collected when the organs were placed horizontally.
To ensure the consistency, samples were collected at the same position of each segment. The sampling was as follows: stomach (the pylorus), jejunum (the site 10 cm after the duodenojejunal junction), ileum (the site 10 cm before the ileum-cecum orifice), cecum (the tip of the cecum), ventral colon (the middle of the ventral colon), and dorsal colon (the middle of the dorsal colon; Liu et al., 2019). The contents were stored in a 50-ml sterile and enzyme-free centrifuge tube, mixed, and immediately placed in liquid nitrogen, and then cryopreserved at −80°C. The animal experiments were approved by the Animal Welfare Committee of Inner Mongolia Agricultural University, and all procedures were conducted in accordance with the guidelines of the China Animal Protection Association. The characteristics of the individual horses, including age, sex, weight, height, length, bust, hair, and color, are summarized in Table 1.

| DNA extraction, 16S rRNA gene PCR, and sequencing
Total genomic DNA was extracted from the GIT samples using the CTAB/SDS method, and the concentration and purity were evaluated by electrophoresing in 1% agarose gels. The distinct regions of the 16S rRNA (V3-V4 hypervariable regions) were amplified using barcode-tagged specific primers (16SRNA V3-V4: 341F-806R). Each PCR mixture consisted of 15 μl Phusion® High-Fidelity PCR Master Mix (New England Biolabs), 0.2 μM forward and reverse primers, and ~10 ng template DNA (1 ng/µl) for a final volume of 30 µl. The PCR mixture was denatured at 98°C for 1 min firstly, then followed by 30 cycles of denaturation at 98°C for 10 s, annealing at 50°C for 30 s, and elongation at 72°C for 30 s, and the final elongation was performed at 72°C for 5 min. The PCR products were electrophoresed on a 2% agarose gel and purified by Gene JETTM Gel Extraction Kit (Thermo Scientific).

| Library preparation and sequencing
Library construction and sequencing were performed by the Novogene Company. Sequencing libraries were generated using Ion Plus Fragment Library Kit (48 reactions, Thermo Scientific) according to the manufacturer's instructions. The library quality was assessed on the Qubit ® 2.0 Fluorometer (Thermo Scientific) and sequenced on an Ion S5 TM XL platform. 400-bp/600-bp single-end reads were generated by sequencing finally.

| Data analysis
Single-end reads were assigned to samples based on their unique barcode and truncated by excising the barcode and primer sequences. The raw reads were first filtered according to the Cutadapt (V1.9.1, http://cutad apt.readt hedocs.io/en/stabl e/) quality control process to obtain high-quality reads. The latter were compared with the reference database using the UCHIME algorithm (http://www.drive5.com/usear ch/manua l/uchim e/algo. html) (Edgar, Haas, Clemente, Quince, & Knight, 2011) to detect chimaera sequences, which were then removed .
Then, the clean reads were obtained (Table A2). Sequence analyses were performed with Uparse software (v7.0.1001, http:// drive5.com/upars e/) (Edgar, 2013), and sequences with ≥97% similarity were assigned to the same operational taxonomic units (OTUs). Representative sequences of each OTU were subjected to species annotation (threshold set at 0.8 to 1) and abundance analysis using the Mothur software and SSU rRNA SILVA128 (http:// www.arb-silva.de/) (Accessed Date: November 2017) database (Wang, Garrity, Tiedje, & Cole, 2007) (Quast et al., 2013). With the minimum amount of data in the sample as the standard, the data of each sample were homogenized for subsequent alpha and beta diversity analyses.
To calculate alpha diversity, the OTU table was rarefied and two metrics were calculated, observed species and Shannon index, the observed species is to estimate the amount of unique OTUs found in each sample. Rarefaction curves were generated based on these two metrics. For beta diversity analysis, UniFrac distance was calculated, and unweighted pair group method with arithmetic (UPGMA) mean sample clustering trees were constructed using QIIME software (version 1.9.1). The unweighted UniFrac was used for principal coordinate analysis (PCoA). PCoA can be used for determining principal coordinates and visualizing complex, multidimensional data. Differences in community structure among groups were tested by analysis of molecular variance (AMOVA), and species differences among groups were analyzed with LDA effect size (LEfSe, LDA score of 4). The functional composition of the microorganisms was predicted by the PICRUSt (version 1.1.2) programs. Default parameters were used for all analyses except those specific parameters.
All data analyses were performed using SPSS software, version 22.0. The different parameters of horse GIT were expressed as mean ± standard deviation. Statistical significance was analyzed with ANOVA, and multiple groups were compared using the LSD test.
TA B L E 1 Details of the horses used for the characterization of the microbiota present in different compartments of the GIT  and dorsal colon were 11.69%, 11.79%, and 24.95%, respectively.
The proportion of common OTUs was 31.98%. The alpha diversity index analysis showed significantly higher microbial diversity in the individual LG segments than different UG segments (p < .001; Figure 1c,d), whereas no significant differences were observed among the individual segments of the LG or those of the UG.
While the thick-walled Firmicutes was the most abundant phylum in the UG, the relative abundance of Firmicutes and Bacteroides was similar in the LG (Table A4). The results of analysis by individual segments showed Firmicutes were significantly more abundant in the mid-ileum than stomach (p = .039), cecum (p < .001), ventral colon (p = .015), and dorsal colon (p = .005). Firmicutes were also more abundant in the jejunum than cecum (p = .004) and DC (p = .022).
At the genus level, significant differences were also seen between the microbial compositions of the small and large intestines, whereas those of the cecum and colon were more consistent (Figure 2c). The abundance of all genus did not exceed 35% in the UG, and only slight differences were seen between the abundance of different genera in the LG. However, the relative abundance of microorganisms across the different GIT segments was significantly different (Table   A5). The results of AMOVA showed that the microbial community structures were significantly different across the distinct GIT regions (p < .05; F = 12.26), while those of the jejunum, ileum, cecum, and VC were similar (Table 2)

| D ISCUSS I ON
Compared with traditional isolation methods, the next-generation sequencing appears more efficient to analyzing microbiome structures, especially for the species that are hard to cultivate in vitro . Therefore, this technique had been used extensively to characterize the intestinal microbiota of several species (Kim, Gu, Lee, Joh, & Kim, 2012;Orpin, 1981;Wu et al., 2016;Yang et al., 2017;Zhang et al., 2016;Zhou et al., 2016). Present studies on the gut microbiota were focused on fecal samples, which only represent the microbial structures of the right dorsal colon but not F I G U R E 2 The relative abundance of luminal Mongolian horse GIT microbiota. UPGMA clustering analysis with weighted UniFrac distance matrix on the left and relative abundance of bacteria on the right in each group at the phylum (a), family (b), and genus levels (c) the entire gut microbiota. Therefore, direct sampling of the different parts of the GIT can reflect the function of the coevolving bacterial communities in complex mammalian ecosystems (Isaacson & Kim, 2012;Willing et al., 2009) more accurately. At the same time, the study shows that the fecal microbial diversity of wild horses is higher than that of captive horses (Metcalf et al., 2017). Therefore, this paper adopts grazing to simulate the natural state as much as possible.

| Composition of the GIT microbiota of the Mongolian horse
The composition of the intestinal microbiota is the result of longterm evolutionary adaptation of the host to its diet; therefore, there are great differences among herbivores, carnivores, and omnivores.
Herbivores have a higher proportion of Firmicutes and Bacteroides, reflecting the high cellulose content from ingested plants (Isaacson & Kim, 2012

Mongolian horses
As mentioned above, Firmicutes and Bacteroidetes were the dominant bacteria at a ratio of 1:1 in the LG of the horses, the result contradicted the observations using fecal samples (Costa et al., 2012;Costa, Stampfli, et al., 2015;Schoster, Mosing, Jalali, Staempfli, & Weese, 2016;Zhao et al., 2016). However, this result is consistent with studies on microbial communities in different parts of the intestine (Costa, Silva, et al., 2015;Ericsson, Johnson, Lopes, Perry, & Lanter, 2016). Therefore, the feces do not fully represent the entire gut microbiota. In addition, previous studies indicated that the proportion of dominant intestinal microbiota is dependent on the geographical location or seasonal feed (Ericsson et al., 2016), but the availability in Mongolian horses needs further investigation. We observed distinct microbial communities in the different parts of the GIT, but the compositions of adjacent parts were usually similar (except for the ileum and cecum). The greater microbial diversity in the distal gut indicated a more complex microenvironment in that region. This is in agreement with studies that the ecology of the GIT is not static but with significant regional changes . Based on the gut microbiota, the equine GIT could be This bacterium is an appealing candidate to become a human probiotic because of negative correlation with the incidence of obesity, diabetes, inflammation, and metabolic disorders (Everard et al., 2013;Hansen et al., 2012;Png et al., 2010;Wang, Bose, Kim, Han, & Kim, 2015). Only four previous studies (Costa, Stampfli, Allen-Vercoe, & Weese, 2016;Costa, Stampfli, et al., 2015;Rodriguez et al., 2015;Zhao et al., 2016)

| Functional prediction of the Mongolian horse intestinal microbiota
In previous studies, there was no prediction of the function of gastrointestinal flora in different parts of the gastrointestinal tract of horses (Costa, Silva, et al., 2015;Ericsson et al., 2016). This study

| CON CLUS IONS
The microbial communities of the different parts of the Mongolian horse GIT were significantly different, and there was greater diversity between the LG and UG. Direct sampling of the different segments of GIT provided a more complete diagram of the gut microbiota compared with fecal analysis. The vegetarian diets and adaptability of Mongolian horses were likely related not only to their stable and complicated gastrointestinal microbiota but also to their special herbivorous digestive physiology.

ACK N OWLED G EM ENTS
The authors would like to thank Dr Sachula Wu, Dr Xiaoqing Zhao, Dr Yumei Shan, and Dr A Naer for their technical help.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

E TH I C S S TATEM ENT
The animal experiments were approved by the Animal Welfare Committee of Inner Mongolia Agricultural University, and all procedures were conducted in accordance with the guidelines of the China Animal Protection Association. Note: The cage technique was used as follows: ten 1.5 m × 1.5 m grazing cages were placed within 35.23 ha pasture, and after 34-day grazing of 26 horses, the forage inside the cages and outside the cages in ten random areas was clipped. The weight of fresh forage was measured, and after drying, the daily dry matter intake of each horses was calculated according to the formula.

PD
= (A1−A2)×H D×N PD: average daily dry matter intake per horse (kg/day); A1: the weight of dry forage inside the cages (g/m 2 ); A2: the weight of dry forage outside the cages (g/m 2 ); H: grazing area (ha); D: grazing days (d); N: number of horses grazing. Note: Raw reads: filter out the sequences of low-quality bases; clean reads: After filtering the chimera, the final sequence is used for subsequent analysis; AvgLen: average length of clean reads; Q20: the percentage of bases whose mass value is greater than 20 in clean reads; GC (%): GC base content in clean reads; effective (%): the number of clean reads versus the number of raw reads.