Analysis of a comprehensive dataset of diversity generating retroelements generated by the program DiGReF

Abstract

Background: Diversity Generating Retroelements (DGRs) are genetic cassettes that can introduce tremendous diversity into a short, defined region of the genome. They achieve hypermutation through replacement of the variable region with a strongly mutated cDNA copy generated by the element-encoded reverse transcriptase. In contrast to "selfish" retroelements such as group II introns and retrotransposons, DGRs impart an advantage to their host by increasing its adaptive potential. DGRs were discovered in a bacteriophage, but since then additional examples have been identified in some bacterial genomes.

Results: Here we present the program DiGReF that allowed us to comprehensively screen available databases for DGRs. We identified 155 DGRs which are found in all major classes of bacteria, though exhibiting sporadic distribution across species. Phylogenetic analysis and sequence comparison showed that DGRs move between genomes by associating with various mobile elements such as phages, transposons and plasmids. The DGR cassettes exhibit high flexibility in the arrangement of their components and easily acquire additional paralogous target genes. Surprisingly, the genomic data alone provide new insights into the molecular mechanism of DGRs. Most notably, our data suggest that the template RNA is transcribed separately from the rest of the element.

Conclusions: DiGReF is a valuable tool to detect DGRs in genome data. Its output allows comprehensive analysis of various aspects of DGR biology, thus deepening our understanding of the role DGRs play in prokaryotic genome plasticity, from the global down to the molecular level.

Figure 1

Mode of action of diversity-generating retroelements (DGRs). DGRs always comprise an ORF encoding a reverse transcriptase (RT), a template repeat (TR) and at least one target ORF harboring the variable region (VR), which corresponds to the TR. First, an RNA transcript is made from the TR, which is then reversely transcribed by the RT in an error-prone fashion. In a process termed mutagenic homing, the mutagenized cDNA replaces the parental VR in the target ORF, thereby altering the host gene.

Figure 2

Phylogenetic Tree of DGR RTs (representative selection). A phylogenetic tree was compiled using a Neighbour-Joining algorithm, bootstrapping was done with 1000 replications using PHYLIP. Groups of DGR RTs that form highly uniform clades with high bootstrapping values are shown collapsed. The complete tree is supplied online as Additional file 2.

Figure 3

Core DGR element found in Vibrio cholerae HE-09. A 1.5 kb fragment of the Vibrio phage kappa was replaced by a 2.9 kb fragment in V. cholerae HE-09, including an RT ORF, a template repeat (TR), an atd ORF and an mtd ORF including a variable region (VR), The inserted fragment bears high homology to corresponding elements in the Bordetella phage BPP-1. Sequences upstream and downstream of the 2.9 kb element correspond to homologous sequences in Vibrio phage kappa.

Figure 4

Inactivated or unusual DGRs. We encountered several examples of DGRs with inactivated RTs, but intact VR/TR repeats. This includes (A) nonsense mutations as in Ruminococcus gnavus ATCC 29149, where a premature stop codon truncates the RT ORF, (B) disruption of the RT ORF by insertion elements as in Acaryochloris marina MBIC11017, or (C) frameshift mutations as in Bacteroides sp. 9_1_42FAA. (D) Some DGRs contain full-length RT ORFs, but several B-to-N mutations in their VR. VRs of the respective elements are shown on the left and depicted in yellow (A-to-B mutations: orange vertical bars; B-to-N mutations: black vertical bars). The corresponding RTs are schematically shown on the right (not to scale). The 5’ part of the RT is shown as green box, inactivated parts in gray.

Figure 5

Sequence logo of motif 4 of DGR RTs. A total of 155 substrings comprising motif 4 of the DGR RTs were taken from our result set and a sequence logo was created using WebLogo [40,41]. Numbers 1 to 23 on the x-axis indicate the relative position in the substring. The height of the symbols denotes the relative frequency of each amino acid at the respective position, while the overall height of the stack represents the degree of conservation measured in bits.

Figure 6

Statistical analysis of adenine substitutions. Frequencies of the four bases were determined for each adenine-corresponding position in the VRs of our data set. Bars represent the mean values of these frequencies over the total results set (A) or the Bacteroides cluster 1 (B). Error bars are +/− standard deviation. The frequencies of the three nucleotides C, T and G in (A) were significantly different from each other (p < 0.001, χ² test).

Figure 7

Structural diversity of DGRs. Arrangements of DGR elements was determined and DGRs were classified into four main groups and respective subgroups. The number and fraction of hits per subgroup is given in parentheses. RT ORFs are shown in green and target ORFs in blue, TRs are depicted in pink and VRs in yellow. Some subgroups (1e, 1f, 3b, 3e, etc.) were not observed in our data set and are thus missing in this figure.