Characterizing the bacterial community across the gastrointestinal tract of goats: Composition and potential function

Abstract The composition and function of the microbial community in the gastrointestinal tract (GIT) have increasingly captured the attention of nutritionists because these traits affect the nutrient utilization efficiency and health of host animals. Little information has been reported on these aspects of the goat GIT. This study used 12 female goats (weighing 20.70 ± 1.60 kg and 10 months of age) to examine the composition and function of the microbiota in the rumen, abomasum, jejunum, cecum, and colon. Total genomic DNA was extracted from chyme samples from different sections of the GIT, and the hypervariable region of the 16S rRNA gene was amplified by PCR using bacterial universal primers. The amplicons were sequenced on an Illumina MiSeq platform, and the biological information was analyzed using QIIME software. A total of 857 genera that belonged to 39 phyla were observed across the goat GIT, with Bacteroidetes and Firmicutes dominating. Our results revealed significant differences in the composition, diversity, and species abundance of the bacterial communities in the different sections of the GIT. However, the compositions of the bacterial communities in adjacent GIT segments showed similarities in addition to differences. The study indicated that there were significant differences in microbial function among the GIT regions. In particular, the relative abundances of genes involved in energy metabolism, amino acid metabolism, nucleotide metabolism, and glycan metabolism were overrepresented in samples from the forestomach, and genes related to energy metabolism, amino acid metabolism, and glycan metabolism were mainly enriched in samples from the small intestine. Additionally, the relative abundances of bacteria at the phylum and genus levels were significantly correlated with these metabolic functions. In general, there were significant differences in composition and potential function among the bacterial communities in the goat GIT.


| INTRODUC TI ON
The gastrointestinal tract (GIT) of ruminants harbors a dense and diverse microbiota that has long been recognized as an essential factor in converting plant materials into digestible substances. The existing physiological and biochemical knowledge has revealed that the bacteria in distinct regions of the GIT have different functions. For example, the microbes in the rumen mainly help the host degrade dietary components such as fiber, but the microbes in the small intestine play a significant role in maintaining the health of the host as well as in digesting nutrients (Bauer et al., 2018;Cervantesbarragan et al., 2017;Dodd et al., 2017;Kadoki et al., 2017;Koppel, Maini, & Balskus, 2017). The function of the microbiota is based on its composition and phylogenetic distribution, and the differences in composition and structure inevitably lead to the differences in function between different microbiota. Nevertheless, because of their convenience, microbiological samples derived from the rumen or feces are often used when assessing the health and digestive function of the whole GIT (Abderzak et al., 2012;Ramírez-Restrepo et al., 2016;Riyanti, Suryahadi, & Evvyernie, 2015). Little research has been conducted to analyze the microbial composition in other GIT compartments of goats (such as in the small and large intestine) (De Oliveira et al., 2013). However, experiments in chicken (Zhao et al., 2013), donkeys (Liu et al., 2014), horses (Dougal et al., 2012), and mice (Gu et al., 2013) have shown high variation among the microbial communities of different regions of the GIT. Ruminal or fecal microbiota cannot reflect the microbial communities in other segments of the GIT (Mao, Zhang, Liu, & Zhu, 2015). The use of samples from the rumen or feces to speculate on the structure and composition of bacterial communities in other GIT compartments would not allow researchers to fully understand the microbial function of the different communities. To gain a comprehensive understanding of functional localization, the microbiota in different parts of the GIT should be analyzed.
Previous studies have found that the microbiota varied greatly with the animal species (Ley et al., 2008). Thus, although the microbial compositions in the GIT of steers (De Oliveira et al., 2013) and dairy cattle (Mao et al., 2015) have been revealed, information on the compositions, functions, and metabolic activities of the bacterial communities in the GIT of goats remains unknown (Ramírez-Restrepo et al., 2016;Riyanti et al., 2015). In the present study, we hypothesized that the diversity and function of the microbial community in different regions along the GIT of goats varied significantly, and an experiment was conducted to characterize the compositions and distributions of the gastrointestinal microbiota in goats using highthroughput 16S rRNA gene amplicon sequencing and to analyze their potential functional differences using PICRUSt (phylogenetic investigation of communities by reconstruction of unobserved states).

| Animals and sample collection
Twelve female Nubian black goats, which were 10 months old and weighed 20.70 ± 1.60 kg, were used in this study. Throughout the experimental period, the goats were fed a total mixed ration (TMR) to avoid the selection of feed components. The TMR contained 38.47% corn, 20.00% alfalfa meal, 35.00% Leymus chinensis, 4.50% soybean meal, 0.45% NaCl, 0.45% baking soda, 0.08% CaCo 3 , 0.60% CaHPO 4 , and 0.45% premix and had a nutritive content of 9.71% CP, 24.07% ADF, 36.11% NDF, 2.95% EE, and 9.33 MJ/kg ME on a dry matter basis. All goats were fed twice daily with equal amounts of feed at 8:00 a.m. and 5:00 p.m. and were kept in individual cages under controlled environmental conditions with free access to food and water.
The experiment lasted for 60 days, including 15 days for adaptation.
On day 60, the goats were slaughtered, and the luminal contents were collected from the rumen, abomasum, jejunum, cecum, and colon (50 ml). The sampling procedure was as follows: the goats were transferred to a biopsy table postmortem. Subsequently, the rumen and abomasum were cut with sterilized scissors, and the contents of these compartments were collected. During the intestinal sampling, the jejunum, cecum, and colon were isolated by tying off each anatomical section at both ends with thread to prevent the movement of the luminal contents from one region to another. All samples were kept at −80°C until DNA extraction.

| DNA extraction, PCR amplification, and Illumina MiSeq sequencing
Total microbial DNA was extracted from the luminal contents and purified using a method described previously (Guo et al., 2015). The quality of the DNA was determined using agarose electrophoresis and a Nanodrop 8000 spectrophotometer (Thermo Scientific, Australia). The high quality DNA was amplified using the 515F/806R primer set (forward primer 515F with a sequence of 5′-GTGCCAGCMGCCGCGGTAA-3′ and reverse primer 806R with a sequence of 5′-GGACTACVSGGGTATCTAAT-3′) (Caporaso et al., 2011) that targets the V4 hypervariable region of the bacterial 16S rRNA gene, with a unique 5-to 8-base error-correcting barcode for multiplexed DNA sequencing.
The amplification was initiated with denaturation at 94°C for 3 min, followed by 30 cycles at 94°C for 30 s, 58°C for 30 s, and 72°C for 90 s, and a last extension at 72°C for 5 min. The 50 μl reaction mixture contained 200 nM of each primer, 5 μl of 2.50 mmol/L dNTP mixture, 5 μl of 10× Ex Taq buffer (20 mmol/L Mg2 + ; Takara Inc., Dalian, China), 0.35 μg of template DNA, 2 mM of MgCl 2 , 4 units of Taq DNA polymerase (Takara Inc.), and approximately 37 μl Milli-Q water. The amplicons were purified using a PCR Clean-Up system (Promega, Madison) with a purification kit (QIAGEN, Australia) and were quantified using a QuantiFluor™-ST fluorometer (Promega, China). Finally, the samples were sequenced on the MiSeq Illumina sequencing platform (Novogene Technology Co., Ltd, Beijing, China), according to the protocols described in previous article (Caporaso et al., 2012).
Uclust (version 1.2.22q) (Edgar, 2010) was then used to cluster the obtained clean and high-quality sequences into operational taxonomic units (OTUs) for an eventual taxonomy assignment based on 97% sequence similarity (http://www.mothur.org/wiki/ Greengenes-formatted_databases, gg_otu_13_8). The most abundant sequence was selected as the representative for each OTU and was assigned to a taxonomic group using RDP Classifier (version 2.12) (Cole et al., 2009).
The chimeric OTUs were removed from the analysis against the sequence from the SILVA database (Quast et al., 2013) (http:// www.mothur.org/wiki/Silva-reference-files). Good's coverage and rarefaction curves were determined to estimate the coverage and sampling effort using the analysis of alpha diversity. Mothur was also used to calculate the population diversity (Simpson index), evenness (Shannon index), richness (Chao1) and phylogenetic diversity (PD).
Beta diversity was measured by calculating the weighted and unweighted UniFrac distances between each pair of samples, and the unweighted UniFrac distance matrix was measured and visualized using a principal coordinate analysis (PCoA) (Lozupone, Lladser, Knights, Stombaugh, & Knight, 2011). A PCoA was applied to the resulting distance matrices to generate two-dimensional plots using R (version x64 3.4.2) (http://cran.rstudio.com).
According to the results of the species classification, OriginPro (version 9.0) software was used to draw a relative abundance histogram of the dominant bacterial phyla. In addition, the genera that were shared by all samples were selected to create a heatmap using R (version x64 3.4.2).
Finally, the putative bacterial metabolic pathways and functions were assessed via PICRUSt (Langille et al., 2013). PICRUSt is a bioinformatics tool designed to predict the gene functions of a microbial community. The inferred genes and their functions were aligned with the Kyoto Encyclopedia of Genes and Genomes (KEGG, http://www.genome.jp/kegg/), which is a database resource for understanding the high-level functions and utilities of biological systems. A similarity search with an E-value <10 −5 was performed for the prediction and functional annotation (Fu et al., 2016).

| Statistical analysis
Nonparametric tests were performed using SPSS (version 20.0) for Windows (SPSS Inc., Chicago, IL) to analyze the effects of GIT region on bacterial prevalence and the relative abundance values of the KEGG pathways. The results are shown as the means ± SD.
Correlations were determined using Spearman correlation analysis.
Differences between means were considered significant at p < 0.05 and extremely significantly different at p < 0.01. Note. Values are expressed as the Ms ± SD. Values within the same column with same superscripts were not significantly different from one another (p > 0.05); however, Values with different lowercase letter superscripts were significantly different (p < 0.05), and values with different capital letter superscripts were extremely significantly different (p < 0.01). OTU: operational taxonomic unit; PD: phylogenetic diversity.
The alpha diversity in the large intestine samples was greater (p < 0.01) than that in the jejunum and forestomach samples ( Table 1).
The samples from the large intestine had the highest diversity, while those from small intestine had the lowest Chao 1, Shannon and Simpson values. The PD, calculated as the sum of all the branch lengths in a 16S rRNA tree, was found to be variable across the goat GIT, reaching a maximum value (p < 0.01) in the large intestine sample (Table 1).
Good's coverage across the GIT was >0.99, implying that the sampling depth was sufficient to estimate the microbial diversity (Table 1). This result was confirmed by rarefaction curves (Figure 1).
All the curves asymptotically approached a plateau, suggesting that the curves accurately reflected the microbial community.

| Beta diversity analysis
A PCoA of overall diversity based on unweighted UniFrac values was also performed to compare the microbial diversity of all samples. The analysis showed that microbial communities from the same/adjacent GIT regions (forestomach, jejunum, and large intestine) were more similar to each other than to those from other regions ( Figure 2). Furthermore, the microbiota in the large intestine was clearly different from that from other regions, as shown by PC1, which accounted for 40.74% of the total variation, and the microbiota in forestomach was different from that in the jejunum, as shown by PC2, which represented 2.52% of the total variation.

| Phylum-and genus-level microbial composition
A total of 39 bacterial phyla were identified in all samples, 15 were common among the samples (Figure 3a), and Bacteroidetes and Firmicutes were the most abundant phyla in all samples (Figure 3b).
The relative abundance of Bacteroidetes was the highest in the forestomach (63.62 ± 1.81% in the rumen and 45.23 ± 2.45% in the abomasum) and was significantly (p < 0.01) higher than that in the jejunum (10.14 ± 4.02%) and large intestine (20.46 ± 1.62% in the colon and 19.48 ± 1.56% in the cecum). The most abundant phylum in the forestomach was Bacteroidetes, while that in jejunum and large intestine was Firmicutes. The relative abundance of Firmicutes in the jejunum, colon, and cecum was 61.19 ± 5.23%, 66.05 ± 2.93%, and 64.77 ± 1.67%, respectively and was significantly higher (p < 0.01) than that in the forestomach (28.52 ± 1.79% in the rumen; 28.75 ± 1.71% in the abomasum, Figure 3c).
At the genus level, a total of 857 bacterial genera were detected, and the average relative abundances of the top 10 abundant genera were compared among the GIT segments ( Table 2).

| Similarity analysis of the bacteria at the genus level
Statistical dissimilarities were observed across the GIT regions with   Note. Relative abundances of the most abundant genera (genera whose relative abundance indicated that they were among the top 10 genera). Values are expressed as the Ms ± SD. Values within the same column with same superscripts were not significantly different from one another (p > 0.05); however, values with different lowercase letter superscripts were significantly different (p < 0.05), and values with different capital letter superscripts were extremely significantly different (p < 0.01). Taxa that could not be assigned to a genus but were present in all samples were displayed using the highest taxonomic level that they could be assigned to.
F I G U R E 4 Cluster heatmap of the shared genera. Note. The heatmap was constructed to determine the relationship between the operational taxonomic units and experimental treatments based on log transformed relative abundances. The phylogenetic tree was constructed with maximum likelihood using FastTree 2.1.3 (Y-axis clustering). Hierarchical clustering based on the distances of the five samples along the X-axis and the bacterial genera along the Y-axis are indicated in the upper part and on the left side of the figure, respectively. The closer to blue, the higher is the relative abundance, while the closer to green, the lower is the relative abundance.

| Correlation between the bacterial community and metabolic function
The main contributors to the abundant functional pathways were analyzed at the phylum and genus level (Figure 7). At the phylum level Ruminococcus (belonging to Firmicutes) was positively correlated with carbohydrate metabolism (r = 0.721, p < 0.01) and glycolysis/ gluconeogenesis (r = 0.752, p < 0.01).

| D ISCUSS I ON
This study aimed to describe the compositions and the potential functions of the microbial communities across the GIT of goats using nextgeneration sequencing technology. The results showed significant differences in the structures of the microbial communities among the GIT sections. For example, the most abundant phylum in the samples of the forestomach was Bacteroidetes, whereas that of the small and large intestine was Firmicutes. Additionally, the genus Prevotella, which was the main genus under the phylum Bacteroidetes, reached up to 20.44% and 19.48% of the total abundance in the rumen and abomasal samples, respectively ( Table 2). The predominant genera in the small and large intestine microbiota were unclassified Clostridiales and unclassified Ruminococcaceae, respectively (Table 2), which belong to phylum Firmicutes. This finding agreed with those of previous studies (Frey et al., 2010;Stevenson & Weimer, 2007), in which the relative abundance of Prevotella was thought to be related to the genetic variability in the different compartments of the GIT. The reason that unclassified Clostridiales and unclassified Ruminococcaceae were enriched in the intestine is not clear yet, but the dominance of the genus Prevotella in the forestomach of goats was not unexpected (Abderzak et al., 2012;Huo, Zhu, & Mao, 2014;Riyanti et al., 2015). Compared to the other regions of the GIT in ruminants, the rumen is the place where nutrient digestion and metabolism mostly occur. Previous results showed that Bacteroidetes possess a strong ability to degrade protein and polysaccharides (Huo et al., 2014;Pitta et al., 2016), and these results were confirmed by the present study (Figure 7a). The genus Prevotella was found not only to degrade nonstructural carbohydrates and protein (Belanche et al., 2012;Purushe et al., 2010;Thompson, Monteagudomera, Cadenas, Lampl, & Azcarateperil, 2015) but also to be involved in amino acid metabolism, nucleotide metabolism, energy metabolism, and glycan biosynthesis, as revealed in this study (Figure 7b) and in a previous study (Hook et al., 2011) as well. In the present study, the family Lachnospiraceae was found in all the five compartments of the goat GIT. Previous studies showed that all species of the family Lachnospiraceae are anaerobic and can only be found in human and mammalian gut microbiota (Huynh et al., 2008). Our results also showed that Lachnospiraceae possessed a significantly higher relative abundance in the large intestine samples than in the forestomach and jejunum samples (Table 2). A higher abundance F I G U R E 5 Similarity of the bacteria at the genus level. Pearson correlation analysis of the relative abundance of the bacterial community in the goat gastrointestinal tract. Only the taxa whose relative abundance was >0.1% of community are presented. Note. A correlation coefficient >0.5 indicates the existence of a correlation (p < 0.05), and that >0.7 indicates described a strong correlation ( health and some members can protect against colon cancer by producing butyric acid (Meehan & Beiko, 2014;Surana & Kasper, 2017).
In addition, significant differences in the diversity and richness of bacteria among the GIT regions (Table 1 and Figure 6) (Gong et al., 2017;Jiao et al., 2016;Mao et al., 2015;Wang et al., 2016). In the present study, all of these factors were consistent for the 12 experimental goats; however, the relative abundances of the dominant phyla and genera varied considerably among the GIT compartments, which emphasized that the sampling site was the major determinant of the microbial composition and community structure along the GIT. This phenomenon has been noticed by researchers in past scientific research reports (De Oliveira et al., 2013).
The composition of the bacterial community in different GIT sections also showed similarities in addition to differences. Samples from adjacent GIT compartments had more similar microbial communities than those from other segments ( Figure 5). The microbial flora in the cecum and colon had the highest degree of similarity (0.964 ± 0.041), followed by that in the rumen and abomasum (0.884 ± 0.053). This result was in agreement with the result obtained from cattle studies (De Oliveira et al., 2013;Mao et al., 2015). These results may suggest that the similarities in the living environments (pH values, the gut motility, and secretion) of bacterial communities in adjacent compartments of the GIT explain the similarities in the microbiota in these regions (Turnbaugh et al., 2009) The results showed that the most prevalent function could be categorized as metabolism (Figure 6b), agreeing with the results from previous studies (Lu et al., 2014;Ridaura et al., 2013). This finding can be explained by the fact that carbohydrates, proteins, and amino acids are essential ingredients for microbial growth (Erickson et al., 2012;Lamendella, Domingo, Ghosh, Martinson, & Oerther, 2011).
The present study showed that the metabolic functions of the bacteria in the goat GIT, such as carbohydrate metabolism, amino acid metabolism, and energy metabolism, were highly represented, which was consistent with the results from previous studies (Wetzels et al., 2015). The findings of the present study revealed significant differences (p < 0.002) in bacterial function among the GIT regions of goats ( Figure 6c). For example, genes related to amino acid metabolism were more abundant in the rumen than in the small and large intestine. The rumen bacteria may possibly derive energy from amino acid fermentation (Malmuthuge et al., 2012), which implies that the bacteria in the rumen may be more necessary for amino acid degradation than that in other sections.
Moreover, the results from the PCA (Figure 6d in the small and large intestine have not been studied as frequently as that in the rumen. The present study showed that the relative abundances of Ruminococcus and Butyrivibrio in the jejunum samples were significantly higher (p < 0.001) than those in the rumen samples (Table 2). Previous studies have shown that Ruminococcus and Butyrivibrio are important in carbohydrate metabolism in the GIT (Stevenson & Weimer, 2007), and those results were verified in the present study (Figure 7). These results suggested that the jejunum may also participate in carbohydrate metabolism, and previous studies also have shown that the intestines of ruminants can compensate for the carbohydrate metabolism that mainly occurs in the forestomach (Wang, Elekwachi, et al., 2017;Wang, Liu, Yin, Zhu, & Mao, 2017;Zoetendal et al., 2012). Therefore, enhanced Ruminococcus and Butyrivibrio in the small intestine may increase the bioavailability of carbohydrate for the host.

| CON CLUS ION
In general, this research revealed the composition and diversity, and partially revealed the potential functions of the microbial communities across the goat GIT. The microbes differed greatly by GIT region and that there were similarities between the adjacent GIT segments. These findings can be potentially used to modulate gastrointestinal microbiota and therefore improve the health and nutrient utilization of goats.

ACK N OWLED G EM ENTS
We would like to thank the National Key R&D Program of China

(Grant No. 2017YFD0502005) and Sichuan Beef Cattle Innovation
Group (Grant No. 035Z389) for their financial support.

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

AUTH O R S CO NTR I B UTI O N
L.Z., L.J., and B.X. designed the experiments; L.Z., L.J., Z.W., and Q.P. performed the experiments; B.X., Z.W., and Q.P. contributed reagents/materials/analysis tools, all authors analyzed the data; L.Z. and L.J. wrote the manuscript. All authors read the final manuscript.

E TH I C S S TATEM ENT
The experimental protocol used in the present study was approved by the Animal Policy and Welfare Committee of the Agricultural Research Organization of Sichuan Province, China and was in accordance with the guidelines of the Animal Care and Ethical Committee of the Sichuan Agricultural University.

DATA ACCE SS I B I LIT Y
All sequence data in the present study were deposited in the sequence read archive (SRA) of the NCBI database under the number SRP185613.