Comparative genomic analysis of the gut bacterium Bifidobacterium longum reveals loci susceptible to deletion during pure culture growth

Abstract

Background: Bifidobacteria are frequently proposed to be associated with good intestinal health primarily because of their overriding dominance in the feces of breast fed infants. However, clinical feeding studies with exogenous bifidobacteria show they don't remain in the intestine, suggesting they may lose competitive fitness when grown outside the gut.

Results: To further the understanding of genetic attenuation that may be occurring in bifidobacteria cultures, we obtained the complete genome sequence of an intestinal isolate, Bifidobacterium longum DJO10A that was minimally cultured in the laboratory, and compared it to that of a culture collection strain, B. longum NCC2705. This comparison revealed colinear genomes that exhibited high sequence identity, except for the presence of 17 unique DNA regions in strain DJO10A and six in strain NCC2705. While the majority of these unique regions encoded proteins of diverse function, eight from the DJO10A genome and one from NCC2705, encoded gene clusters predicted to be involved in diverse traits pertinent to the human intestinal environment, specifically oligosaccharide and polyol utilization, arsenic resistance and lantibiotic production. Seven of these unique regions were suggested by a base deviation index analysis to have been precisely deleted from strain NCC2705 and this is substantiated by a DNA remnant from within one of the regions still remaining in the genome of NCC2705 at the same locus. This targeted loss of genomic regions was experimentally validated when growth of the intestinal B. longum in the laboratory for 1,000 generations resulted in two large deletions, one in a lantibiotic encoding region, analogous to a predicted deletion event for NCC2705. A simulated fecal growth study showed a significant reduced competitive ability of this deletion strain against Clostridium difficile and E. coli. The deleted region was between two IS30 elements which were experimentally demonstrated to be hyperactive within the genome. The other deleted region bordered a novel class of mobile elements, termed mobile integrase cassettes (MIC) substantiating the likely role of these elements in genome deletion events.

Conclusion: Deletion of genomic regions, often facilitated by mobile elements, allows bifidobacteria to adapt to fermentation environments in a very rapid manner (2 genome deletions per 1,000 generations) and the concomitant loss of possible competitive abilities in the gut.

Figure 1

(A) B. longum strain DJO10A (B) NCC2705. Circle (a) indicates the coding regions by strand. The color of each gene refers to the COG functional categories. Circle (b) lists the number of base pairs (bp). Circle (c) contains the unique genes organized according to the coding strand (first two blocks) with the third block indicating the larger unique regions as defined in the text (blue), a prophage region (red), and rRNA operons (green). Circle (d) illustrates the G+C content. Circle (e) shows tRNA genes with block arrows indicating their coding strand. Circle (f) indicates insertion sequences (IS).

Figure 2

Genome unique regions. (A) Base deviation index (BDI) analysis of the B. longum DJO10A and NCC2705 genomes. Unique regions of each genome as defined in the text are numbered. The locations of oriC and terC are indicated by green arrows. Letters refer to predicted gene phenotypes from regions with definitive BDI peaks that are present in both genomes, a, GTPase, b, cation transport ATPase, c, DNA partitioning protein, d, choloylglycine hydrolase, e, glutamine synthase beta chain, f, alanyl-tRNA synthetase, g, pyruvate kinase, h, cation transport ATPase, I, fibronectin type III, j, aminopeptidase C, k, subtilisin-like serine protease, l, sortase, m, fatty acid synthase. (B) Organization of the unique region 1 showing the location of a 361 bp DNA remnant, indicated by the green bar, from the ushA gene remaining at the predicted deletion location in NCC2705. Sky blue colored ORFs indicate common genes between both genomes. ^a, mobile integrase cassette.

Figure 3

Comparsion of oligosaccharide utilization gene cluster 7 between two B. longum genomes. DJO10A-unique genes in unique region 10 are colored dark grey, ISL3-type IS element is colored black and other matched genes are colored white. galA, α-galactosidase; lacI, LacI-type repressor; malEFG, ABC-type transport system; ISL3, ISL3-type IS element; agl1, glycosidase; ilvA, threonine dehydratase; SIR2, NAD-dependent protein deacetylase; glyH, glycosyl hydrolase; hyp, hypothetical protein.

Figure 4

Organization of genes involved in polyol metabolism in the unique region 13 in strain DJO10A and comparison with an analogous region in B. adolescentis ATCC 15703. Amino acid identities are indicated between homologous genes. ORFs shaded black are from unique region 13 and corresponding homologs in B. adolescentis ATCC 15703.

Figure 5

Arsenic resistance of selected bacteria. (A) Genetic organization of arsenic resistance gene clusters compiled from the completed genome sequences of Bifidobacterium longun DJO10A, Bacillus subtilis 168 [34], Bacteroides thetaiotamicron VPI-5482 [35], Lactobacillus brevis ATCC 367 [36], L. plantarum WCFS1 [37], L. johnsonii NCC 533 [38] and E. coli K-12 [39]. ^a, 48 kb element that is excised by the site-specific recombinase SpoIVCA during sporulation, ^b, indicates a plasmid sequence, arsR, repressor, arsA, arsenite stimulated ATPase, arsB, arsenite efflux pump, arsC, arsenate reductase, arsD, arsenic chaperone, hyp, hypothetical protein. (B) Comparison of arsenic resistance activity in B. longum DJO10A with fermentation adapted B. animalis subsp. lactis strains, E. coli and Lactobacillus plantarum. ^c, calculated from data presented in van Kranenburg et al., [40].

Figure 6

Lantibiotic prodiction by B. longum DJO10A. (A) Organization of the lantibiotic encoding unique region 12 of B. longum DJO10A and the corresponding genome locations in strains NCC2705 and DJO10A-JH1. The A or B designator following IS30 refer to unique classes of IS30 elements that are only found at this location in the genome. The ' designator indicates a fragmented IS30 element. (B) Pulsed Field Gel Electrophoresis (PFGE) analysis of XbaI-digested total DNA from B. longum DJO10A and its fermentation adapted isolate, DJO10A-JH1. White arrows indicate bands missing from strain DJO10A-JH1. (C) Bioassay for lantibiotic production by B. longum DJO10A with strains DJO10A and DJO10A-JH1 as indicator bacteria.

Figure 7

IS30 'jumping' in the genome of B. longum DJO10A. (A) Genome positioning of the IS30 elements in the genome of B. longum DJO10A and the laboratory adapted strain DJO10A-JH1. The gray arrows indicate the five elements identified by direct sequencing of DJO10A genomic DNA. The white arrows indicate the location of elements that were detected in some sequencing clones prepared from DJO10A genomic DNA. The asterisk under A6 indicates this element was missing from some sequencing clones of DJO10A DNA. (B)NruI digested genomic DNA from DJO10A shown in the left gel and its Southern hybridization (right gel) using probes specific for four different IS element families. (1) refers to DJO10A and (2) refers to DJO10A-JH1. Arrows indicate bands in DJO10A corresponding to specific IS30 elements as illustrated in (A).

Figure 8

Simulated fecal competitive analysis of B. longum DJO10A and its in vitro adapted derivative, strain DJO10A-JH1, against Clostridium difficile and E. coli. (A) Viable cell counts of E. coli DJOec1 at the beginning of the competitive study (black), following competition with B. longum DJO10A-JH1 (horizontal lines) and B. longum DJO10A (hatched). (B) Viable cell counts of C. difficile DJOcd1 at the beginning of the competitive study (black), following competition with B. longum DJO10A-JH1 (horizontal lines) and B. longum DJO10A (hatched). N = 3.