A diversity of uncharacterized reverse transcriptases in bacteria

Abstract

Retroelements are usually considered to be eukaryotic elements because of the large number and variety in eukaryotic genomes. By comparison, reverse transcriptases (RTs) are rare in bacteria, with only three characterized classes: retrons, group II introns and diversity-generating retroelements (DGRs). Here, we present the results of a bioinformatic survey that aims to define the landscape of RTs across eubacterial, archaeal and phage genomes. We identify and categorize 1021 RTs, of which the majority are group II introns (73%). Surprisingly, a plethora of novel RTs are found that do not belong to characterized classes. The RTs have 11 domain architectures and are classified into 20 groupings based on sequence similarity, phylogenetic analyses and open reading frame domain structures. Interestingly, group II introns are the only bacterial RTs to exhibit clear evidence for independent mobility, while five other groups have putative functions in defense against phage infection or promotion of phage infection. These examples suggest that additional beneficial functions will be discovered among uncharacterized RTs. The study lays the groundwork for experimental characterization of these highly diverse sequences and has implications for the evolution of retroelements.

Figure 1.

RTs identified in eubacterial, archaeal and phage genomes. RTs were classified based on limited phylogenetic analyses, BLAST scores and sequence examination for synapomorphies. The number of RTs per group is indicated in parentheses and by triangle size. Domain structures are depicted in black (alignable with group II introns), gray (uncertain alignment) and white (lack of alignability). Additional domains, including domains of unknown function (DUFs), are in light gray squares, with parentheses indicating that some members lack the domain. In this figure, the designation of an extension as a DUF indicates that the amino acid sequence is conserved within the group. Groups corresponding to lineages in Kojima and Kanehisa (38) are as follows: G2L1 [H1, H2, H3 in (38)]; G2L2, (H4); UG1 (F1); UG3 (G); UG5 (F2); UG6 (E); UG8 (G); UG9 (D); retrons (A); and DGRs (B, C).

Figure 2.

A maximum likelihood tree of group II introns and group II-like (G2L) RTs. Group II introns include representatives of the major intron classes [B–F, mitochondria-like (ML), chloroplast-like (CL)] as well as one that is currently unclassified (G2). Group II-like classes (G2L1–G2L5) are highlighted with gray ovals. Nodes that have a bootstrap value ⩾70 are indicated with black circles. The tree file with GenBank gene identification numbers is available in Supplementary Data S2.

Figure 3.

A maximum likelihood tree of unique retron and retron-like RTs. Black dots indicate nodes with bootstrap support >70. Gray circled numbers designate sequences that have previously been identified as retrons [see Ref. (15)]. The GenBank gene identification number for each circled number in the figure is as follows: 1 (94310415), 2 (8100799), 3 (1519451), 4 (15925199), 5 (134078), 6 (28882460), 7 (15642382), 8 (39546377), 9 (42501), 10 (22036084), 11 (42775), 12 (19703506), 13 (113475791), 14 (23130511), 15 (115376572), 16 (556277), 17 (17530191), 18 (108757956), 19 (134075), 20 (21233055), 21 (17230453) and 22 (39996465). Those sequences with experimental support are shown with a gray circle and their previously assigned name (15). The tree file with GenBank gene identification numbers is available in Supplementary Data S3.

Figure 4.

Alignment of RT domains. An alignment of RT domains 0–7 is shown for representative bacterial and eukaryotic RTs. The predicted three catalytic residues at the RT active site are indicated with asterisks. Residues judged to be alignable across all bacterial RTs (and used in distance calculations for Table 1) are indicated with a plus sign at the bottom of the alignment. Absence of sequence in the figure indicates the lack of alignability rather than the absence of amino acids. Sequences considered ambiguously aligned (based on a more comprehensive set of RTs than those depicted in the figure) are indicated with light shading.

Figure 5.

Phylogenetic distribution of RTs in eubacteria (A) and archaea (B). The size of each sector represents the number of sequenced genomes in GenBank as of 31 July 2007, with actual numbers shown in parentheses beside the phyla listed to the right. Sector colors specify the bacterial group classification. Numbers beside each sector indicate unique full-length RTs within each bacterial phylum (Supplementary Table S3 and footnotes).