A quantitative account of genomic island acquisitions in prokaryotes

Abstract

Background: Microbial genomes do not merely evolve through the slow accumulation of mutations, but also, and often more dramatically, by taking up new DNA in a process called horizontal gene transfer. These innovation leaps in the acquisition of new traits can take place via the introgression of single genes, but also through the acquisition of large gene clusters, which are termed Genomic Islands. Since only a small proportion of all the DNA diversity has been sequenced, it can be hard to find the appropriate donors for acquired genes via sequence alignments from databases. In contrast, relative oligonucleotide frequencies represent a remarkably stable genomic signature in prokaryotes, which facilitates compositional comparisons as an alignment-free alternative for phylogenetic relatedness. In this project, we test whether Genomic Islands identified in individual bacterial genomes have a similar genomic signature, in terms of relative dinucleotide frequencies, and can therefore be expected to originate from a common donor species.

Results: When multiple Genomic Islands are present within a single genome, we find that up to 28% of these are compositionally very similar to each other, indicative of frequent recurring acquisitions from the same donor to the same acceptor.

Conclusions: This represents the first quantitative assessment of common directional transfer events in prokaryotic evolutionary history. We suggest that many of the resident Genomic Islands per prokaryotic genome originated from the same source, which may have implications with respect to their regulatory interactions, and for the elucidation of the common origins of these acquired gene clusters.

Figure 1

Size distribution of 1787 Genomic Islands > 10 kb in 246 genome sequences (note the logarithmic scale on the vertical axis). The GIs are binned per 2 kb in size.

Figure 2

Number of Genomic Islands per genome for the 246 genomes tested (Genome size > 800 kb, with GIs size > 10 kb and no conflicts).

Figure 3

Distribution of the relative compositional similarity and GC similarity of all GIs (1787) with their respective genomes, with 1395 (78%, in red) of the GIs having a relative dissimilarity of 90%.

Figure 4

Number of clustered GIs per genome.

Figure 5

Clustering of the 24 Genomic Islands > 10 kb in Escherichia coli O157H7 strain Sakai in seven clusters and nine singletons. Below the cut-off value (red line; dissimilarity < 1.44, see Additional File 2), seven clusters are identified (six clusters with two GIs, and one with three GIs), with a total of 15 Genomic Islands (indicated with seven colored bars). The GIs and their numbers are identified in Additional File 5.