Genomic Features and Niche-Adaptation of Enterococcus faecium Strains from Korean Soybean-Fermented Foods

Abstract

Certain strains of Enterococcus faecium contribute beneficially to human health and food fermentation. However, other E. faecium strains are opportunistic pathogens due to the acquisition of virulence factors and antibiotic resistance determinants. To characterize E. faecium from soybean fermentation, we sequenced the genomes of 10 E. faecium strains from Korean soybean-fermented foods and analyzed their genomes by comparing them with 51 clinical and 52 non-clinical strains of different origins. Hierarchical clustering based on 13,820 orthologous genes from all E. faecium genomes showed that the 10 strains are distinguished from most of the clinical strains. Like non-clinical strains, their genomes are significantly smaller than clinical strains due to fewer accessory genes associated with antibiotic resistance, virulence, and mobile genetic elements. Moreover, we identified niche-associated gene gain and loss from the soybean strains. Thus, we conclude that soybean E. faecium strains might have evolved to have distinctive genomic features that may contribute to its ability to thrive during soybean fermentation.

Fig 1. Evolutionary relationships of 10 soybean strains based on 8,850 SNPs from 990 core genes.

The optimal tree with the sum of branch length = 2.15 is shown. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) are shown next to the branches. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances are in the units of the number of base substitutions per site.

Fig 2. Genome size, number of accessory CDS, and GC content (%) of soybean strains.

Genome sizes (A), accessory genes (B), and GC content (C) were compared among soybean (SB), clinical (CL) and non-clinical (NC) groups. Each gray spot indicate a single E. faecium strain.

Fig 3. Hierarchical clustering of the 113 E. faecium strains used in this study.

Based on ortholog presence/absence, clusters were obtained. Asterisks indicate the 10 soybean strains (Prefix: SB). Clinical and non-clinical strains were indicated by different prefixes, CL and NC, respectively. Approximately unbiased probabilities for the bootstrapping are shown in the center of the dendrogram.

Fig 4. Evolutionary relationships of 113 strains based on 59,739 SNPs from 990 core genes.

The optimal tree with the sum of branch length = 3.21 is shown. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances are in the units of the number of base substitutions per site.

Fig 5. Frequencies of AR, VF, ME, and BA genes.

Frequencies of four types of genes, AR (A), VF(B), ME (C) and BA (D), were compared among soybean (SB), clinical (CL) and non-clinical (NC) groups. Statistical significance was examined using Tukey's HSD (Honestly Significant Difference). Non-significant comparisons were omitted (P>0.05). Each gray spot indicate a single E. faecium strain.

Fig 6. Niche-specific gene gain and loss in soybean E. faecium strains.

Enriched or missing genes in soybean strains were identified and sorted according to P values (S2 Table). If P values of a gene both between soybean and clinical strains and between soybean and non-clinical strains were less than 0.005 (Fisher’s exact test), the gene was regarded as enriched or missing in the soybean strains.