Comparative fecal metagenomics unveils unique functional capacity of the swine gut

Abstract

Background: Uncovering the taxonomic composition and functional capacity within the swine gut microbial consortia is of great importance to animal physiology and health as well as to food and water safety due to the presence of human pathogens in pig feces. Nonetheless, limited information on the functional diversity of the swine gut microbiome is available.

Results: Analysis of 637, 722 pyrosequencing reads (130 megabases) generated from Yorkshire pig fecal DNA extracts was performed to help better understand the microbial diversity and largely unknown functional capacity of the swine gut microbiome. Swine fecal metagenomic sequences were annotated using both MG-RAST and JGI IMG/M-ER pipelines. Taxonomic analysis of metagenomic reads indicated that swine fecal microbiomes were dominated by Firmicutes and Bacteroidetes phyla. At a finer phylogenetic resolution, Prevotella spp. dominated the swine fecal metagenome, while some genes associated with Treponema and Anareovibrio species were found to be exclusively within the pig fecal metagenomic sequences analyzed. Functional analysis revealed that carbohydrate metabolism was the most abundant SEED subsystem, representing 13% of the swine metagenome. Genes associated with stress, virulence, cell wall and cell capsule were also abundant. Virulence factors associated with antibiotic resistance genes with highest sequence homology to genes in Bacteroidetes, Clostridia, and Methanosarcina were numerous within the gene families unique to the swine fecal metagenomes. Other abundant proteins unique to the distal swine gut shared high sequence homology to putative carbohydrate membrane transporters.

Conclusions: The results from this metagenomic survey demonstrated the presence of genes associated with resistance to antibiotics and carbohydrate metabolism suggesting that the swine gut microbiome may be shaped by husbandry practices.

Figure 1

Taxonomic composition of bacterial phyla using 16S rRNA gene sequences retrieved from GS20 and FLX swine fecal metagenomes. Using the "Phylogenetic Analysis" tool within MG-RAST, the GS20 and FLX sequencing runs were searched against the RDP and greengenes databases using the BLASTn algorithm. The percent of sequences assigned to each of the bacterial phyla from the pig fecal GS20 (A and B) and FLX (C and D) metagenomes is shown. The e-value cutoff for 16S rRNA gene hits to RDP and greengenes databases was 1×10^-5 with a minimum alignment length of 50 bp.

Figure 2

Taxonomic distribution of bacterial phyla from swine and other currently available gut microbiomes within MG-RAST. The percent of sequences assigned to each bacterial order from swine and other gut metagenomes is shown. Using the "Phylogenetic Analysis" tool within MG-RAST, each gut metagenome was searched against the RDP and greengenes databases using the BLASTn algorithm. The percentage of each bacterial phlya from swine, human infant, and human adult metagenomes were each averaged since there was more than one metagenome for each of these hosts within the MG-RAST database. The e-value cutoff for 16S rRNA gene hits to the RDP and greengenes databases was 1×10^-5 with a minimum alignment length of 50 bp.

Figure 3

Taxonomic distribution of bacterial genera from swine and other currently available gut microbiomes within MG-RAST. The percent of sequences assigned to each bacterial order from swine and other gut metagenomes is shown. Using the "Phylogenetic Analysis" tool within MG-RAST, each gut metagenome was searched against the RDP and greengenes databases using the BLASTn algorithm. The percentage of each bacterial phlya from swine, human infant, and human adult metagenomes were each averaged since there was more than one metagenome for each of these hosts within the MG-RAST database. The e-value cutoff for 16S rRNA gene hits to the RDP and greengenes databases was 1×10^-5 with a minimum alignment length of 50 bp.

Figure 4

Hierarchical clustering of gut metagenomes available within MG-RAST based on the taxonomic (A) and functional (B) composition. A matrix consisting of the number of reads assigned to the RDP database was generated using the "Phylogenetic Analysis" tool within MG-RAST, using the BLASTn algorithm. The e-value cutoff for 16S rRNA gene hits to the RDP database was 1×10^-5 with a minimum alignment length of 50 bp. A matrix consisting of the number of reads assigned to SEED Subsytems from each gut metagenome was generated using the "Metabolic Analysis" tool within MG-RAST. The e-value cutoff for metagenomic sequence matches to this SEED Subsystem was 1×10^-5 with a minimum alignment length of 30 bp. Resemblance matrices were calculated using Bray-Curtis dissimilarities within PRIMER v6 software [38]. Clustering was performed using the complete linkage algorithm. Dotted branches denote that no statistical difference in similarity profiles could be identified for these respective nodes, using the SIMPROF test within PRMERv6 software.

Figure 5

Two-way hierarchical clustering of functional gene groups from swine and other currently available gut metagenomes within JGI's IMG/M database. Hierarchical clustering was performed using a matrix of the number of reads assigned to COGs from each gut metagenome, which was generated using the "Compare Genomes" tool in IMG/M ER. COGs less abundant in a given metagenome are shown in black/darkgreen, while more abundant COGs are shown in red.

Figure 6

Pair-wise comparisons of functional gene groups from swine versus other gut metagenomes. Pair-wise comparisons were calculated for the pig fecal metagenome versus (A) lean mouse cecum (B) cow rumen (C) human adult (D) termite gut (E) human infant (F) fish gut (G) and chicken cecal metagenomes is shown. Each point on this exploratory plot represents a different SEED Subsystem and it's relative abundance within the pig fecal metagenome compared to other available gut metagenomes within the MG-RAST database. Points closer to y-axis represent functions more abundant in the swine gut metagenome, while points closer to the x-axis are more abundant in other gut metagenomes. Points laying on or near the dotted midline have equal or very similar abundances within both metagenomes. A matrix of the abundance of sequences assigned to each SEED Subsystem from each gut metagenome was generated using the "Metabolic Analysis" tool in MG-RAST. The number of reads from each individual pig, human infant, and human adult metagenomes were each combined since there was more than one metagenome for each of these hosts within the MG-RAST database. The e-value cutoff for metagenomic sequence matches to SEED Subsystems was 1×10^-5 with a minimum alignment length of 30 bp. Fisher exact tests were used with the Benjamin-Hochberg FDR multiple test correction to generate a list of significantly different SEED Subsystems using STAMP v1.0.2 software [39]. The Newcombe-Wilson method was used to calculate the 95% confidence intervals.