Multi-Omic Analysis in a Metabolic Syndrome Porcine Model Implicates Arachidonic Acid Metabolism Disorder as a Risk Factor for Atherosclerosis

Background The diet-induced gut microbiota dysbiosis has been suggested as a major risk factor for atherothrombosis, however, the detailed mechanism linking these conditions is yet to be fully understood. Methods We established a long-term excessive-energy diet-induced metabolic syndrome (MetS) inbred Wuzhishan minipig model, which is characterized by its genetic stability, small size, and human-like physiology. The metabolic parameters, atherosclerotic lesions, gut microbiome, and host transcriptome were analyzed. Metabolomics profiling revealed a linkage between gut microbiota and atherothrombosis. Results We showed that white atheromatous plaque was clearly visible on abdominal aorta in the MetS model. Furthermore, using metagenome and metatranscriptome sequencing, we discovered that the long-term excessive energy intake altered the local intestinal microbiota composition and transcriptional profile, which was most dramatically illustrated by the reduced abundance of SCFAs-producing bacteria including Bacteroides, Lachnospiraceae, and Ruminococcaceae in the MetS model. Liver and abdominal aorta transcriptomes in the MetS model indicate that the diet-induced gut microbiota dysbiosis activated host chronic inflammatory responses and significantly upregulated the expression of genes related to arachidonic acid-dependent signaling pathways. Notably, metabolomics profiling further revealed an intimate linkage between arachidonic acid metabolism and atherothrombosis in the host-gut microbial metabolism axis. Conclusions These findings provide new insights into the relationship between atherothrombosis and regulation of gut microbiota via host metabolomes and will be of potential value for the treatment of cardiovascular diseases in MetS.


Metagenome and Metatranscriptome Sequencing
We respectively sequenced DNA and RNA obtained from the feces and indicated intestinal segment (i.e., ileum, cecum, and colon) contents. A total of ~2.2 Tb and ~700 Gb sequences were generated from gut microbial metagenome and metatranscriptome sequencing (Tables S3-4), respectively. Following quality control and de novo assembly, a total of 4,612,189 and 992,787 non-redundant genes were included in the gut microbial metagenomic and metatranscriptomic datasets (Figure 1), respectively.

Human and Mouse
Here, we compared our inbred Wuzhishan minipig fecal microbial catalog with two previously published fecal microbial (from human and mouse) gene catalogs (1, 2), both of which were generated using similar (Illumina sequencing) data with similar computational procedures. A much larger number of the minipig fecal bacterial genes mapped to the human catalog as compared to mouse: 35.02% (520247/1485440) of minipig sample bacterial genes were included in the human gut bacterial gene catalogs, compared to only 29% (745908/2572074) of mouse sample bacterial genes.
Moreover, only 15.49% (230141/1485440) of minipig bacterial genes were found in the mouse gut bacterial gene catalog.
Moreover, we also compared the microbiome using KO (KEGG orthology) and genus level relative abundances, respectively, to quantify the overlap of the minipig and mouse fecal microbiomes with that of human. Of note, the similarity of annotated KOs among the minipig, human and mouse gut microbiota was very high (minipig and human: Spearman's r = 0.801, P < 0.001; mouse and human: Spearman's r = 0.865, P < 0.001; minipig and mouse: Spearman's r = 0.779, P < 0.001; Figure 2A).
We further identified 872 KOs (relative abundance > 0.01%) involved in metabolic functions (e.g., carbohydrate, amino acid, nucleotide, cofactors and vitamins, energy metabolism and glycan biosynthesis), genetic information processing (e.g., translation and replication) and cellular processes (e.g., cell motility) that are shared among the minipig, human and mouse gut microbiomes. However, correlations in abundance for genera showed that the minipig gut microbiome was closer to the human microbiome than the mouse microbiome (minipig and human: Spearman's r = 0.767, P < 0.001; mouse and human: Spearman's r = 0.682, P < 0.001; minipig and mouse: Spearman's r = 0.708, P < 0.001; Figure 2A). We also found bacterial genera that occurred in all samples from minipig, human and mouse. Among the 35 most abundant genera in each species, 14 genera shared were commonly detected as gut microbiota ( Figure   2B), including previously reported genera like Prevotella, Bacteroides, Clostridium, Eubacterium, Parabacteroides and Ruminococcus and so on (3). Taken together, the minipig gut microbiome had a higher taxonomic and functional overlapped with the human gut microbiome.

RNA Sequencing and Identification of LncRNA and mRNA
RNA-seq-based transcriptome profilings revealed a total of 1,727 million raw reads and 1,638 million clean reads after quality control (Table S10). The percentage of clean reads compared to raw reads for each library ranged from 90.03% to 96.96%.
Among the clean reads, the percentage of reads with Q30 ranged from 87.  (Table S10). Thus, these results suggested that the RNA-seq data was stable and reliable. We totally identified 25,491 mRNAs from both the HED and ND groups.