Comparative genomics of the Bifidobacterium breve taxon

Abstract

Background: Bifidobacteria are commonly found as part of the microbiota of the gastrointestinal tract (GIT) of a broad range of hosts, where their presence is positively correlated with the host's health status. In this study, we assessed the genomes of thirteen representatives of Bifidobacterium breve, which is not only a frequently encountered component of the (adult and infant) human gut microbiota, but can also be isolated from human milk and vagina.

Results: In silico analysis of genome sequences from thirteen B. breve strains isolated from different environments (infant and adult faeces, human milk, human vagina) shows that the genetic variability of this species principally consists of hypothetical genes and mobile elements, but, interestingly, also genes correlated with the adaptation to host environment and gut colonization. These latter genes specify the biosynthetic machinery for sortase-dependent pili and exopolysaccharide production, as well as genes that provide protection against invasion of foreign DNA (i.e. CRISPR loci and restriction/modification systems), and genes that encode enzymes responsible for carbohydrate fermentation. Gene-trait matching analysis showed clear correlations between known metabolic capabilities and characterized genes, and it also allowed the identification of a gene cluster involved in the utilization of the alcohol-sugar sorbitol.

Conclusions: Genome analysis of thirteen representatives of the B. breve species revealed that the deduced pan-genome exhibits an essentially close trend. For this reason our analyses suggest that this number of B. breve representatives is sufficient to fully describe the pan-genome of this species. Comparative genomics also facilitated the genetic explanation for differential carbon source utilization phenotypes previously observed in different strains of B. breve.

Figure 1

Comparative genomics of fully sequenced B. breve genomes and phylogenetic supertree. a) Venn diagram representing the orthologous and unique gene families as based on BLASTP comparison (E-value cut-off of 0.0001) and MCL clustering algorithm analyses. b) Cluster of Orthologues (COG) classification of the 1141 families of orthologues. For each COG entry the average percentage of hits among B. breve has been indicated. The most abundant families have also been indicated and they are assigned to housekeeping functions. From outer to inner circle: B. breve UCC2003, B. breve JCM 7017, B. breve 689b, B. breve NCFB 2258, B. breve S27, B. breve JCM 7019, B. breve 12L, B. breve ACS-071-V-Sch8b. c) Phylogenetic supertree showing the relationship between thirteen B. breve strains (B. breve UCC2003, B. breve 689b, B. breve 12L, B. breve NCFB 2258, B. breve S27, B. breve JCM 7017, B. breve JCM 7019, B. breve ACS-071-V-Sch8b, B. breve 2L, B. breve 31L, B. breve CECT 7263, B. breve DPC 6330, B. breve DSM 20213), B. longum subsp. longum NCC2705, B. longum subsp. infantis ATCC 15697, B. bifidum PRL2010, B. adolescentis ATCC 15703, B. dentium Bd1, B. animalis subsp. animalis ATCC 25527, B. animalis subsp. lactis DSM 10140 and B. asteroides PRL2011), three actinobacteria (G. vaginalis ATCC 14019, L. xyli subsp. xyli CTCB07 and T. whipplei TW08/27) and a single member of the Firmicutes as an outlier (Lb. plantarum WCFS1).

Figure 2

Regions of variability among the B. breve genomes. a) BLAST-based genome atlas showing the presence of each ORF from B. breve UCC2003 and the other B. breve complete representatives. From outer to inner circle: B. breve UCC2003, B. breve JCM 7017, B. breve 689b, B. breve NCFB 2258, B. breve S27, B. breve JCM 7019, B. breve 12L, B. breve ACS-071-V-Sch8b. b) Alignment showing in red the ORFs with significant G+C mol% in B. breve (B. breve UCC2003, B. breve 689b, B. breve 12L, B. breve NCFB 2258, B. breve S27, B. breve JCM 7017, B. breve JCM 7019, B. breve ACS-071-V-Sch8b); highlighted in grey are regions of variability (REG1-8) that were identified based on gene presence/absence and G+C mol% deviation in each strain. c) Hierarchical clustering heatmap representing the variability of B. breve in terms of presence/absence of genes for eight complete genomes of B. breve with associated percentage of core and variable gene families represented as pie chart.

Figure 3

Pan-genome and core-genome of B. breve. a) Accumulated number of new genes in the B. breve pan-genome plotted against the number of genomes added. The deduced mathematical function is also indicated. b) Accumulated number of genes attributed to the core-genome plotted against the number of added genomes. The deduced mathematical function is also reported.

Figure 4

B. breve carbohydrate profiling. a) Hierarchical clustering analysis performed on the phenotype observed in B. breve UCC2003, B. breve NCFB 2258, B. breve 12L, B. breve 2L, B. breve 31L, B. breve 689b, B. breve JCM 7017 and B. breve JCM 7019, B. breve S27 tested for growing on 24 sugars. b) The in silico prediction and numerical presence of all identified GH family members according to the Cazy classification in the eight fully sequenced B. breve strains. c) Heatmap showing the in silico gene-trait matching exercise performed on the 51 clusters that were derived from the hierarchical clustering analysis. The relative matching distance of each cluster is indicated with colour gradient and the main results are highlighted.

Figure 5

Sorbitol gene cluster in B. breve and insertion mutant growth curve. a) Locus map showing the comparative analysis of the gene cluster putatively involved in the utilization of the sugar alcohol sorbitol in certain B. breve chromosomes. All genes are coloured coded based on their function. The percentage of similarity based on BLASTP alignment and the alcohol dehydrogenase-encoding gene targeted for the genetic insertion experiment are indicated. b) Diagram showing the growth curves of B. breve JCM 7017 wild-type and B. breve JCM 7017–1848 insertion mutant on Rogosa modified MRS (mMRS) with the addition of lactose and sorbitol 0.5% over 24 hours.