Genome-Based Taxonomic Classification of the Phylum Actinobacteria

Abstract

The application of phylogenetic taxonomic procedures led to improvements in the classification of bacteria assigned to the phylum Actinobacteria but even so there remains a need to further clarify relationships within a taxon that encompasses organisms of agricultural, biotechnological, clinical, and ecological importance. Classification of the morphologically diverse bacteria belonging to this large phylum based on a limited number of features has proved to be difficult, not least when taxonomic decisions rested heavily on interpretation of poorly resolved 16S rRNA gene trees. Here, draft genome sequences of a large collection of actinobacterial type strains were used to infer phylogenetic trees from genome-scale data using principles drawn from phylogenetic systematics. The majority of taxa were found to be monophyletic but several orders, families, and genera, as well as many species and a few subspecies were shown to be in need of revision leading to proposals for the recognition of 2 orders, 10 families, and 17 genera, as well as the transfer of over 100 species to other genera. In addition, emended descriptions are given for many species mainly involving the addition of data on genome size and DNA G+C content, the former can be considered to be a valuable taxonomic marker in actinobacterial systematics. Many of the incongruities detected when the results of the present study were compared with existing classifications had been recognized from 16S rRNA gene trees though whole-genome phylogenies proved to be much better resolved. The few significant incongruities found between 16S/23S rRNA and whole genome trees underline the pitfalls inherent in phylogenies based upon single gene sequences. Similarly good congruence was found between the discontinuous distribution of phenotypic properties and taxa delineated in the phylogenetic trees though diverse non-monophyletic taxa appeared to be based on the use of plesiomorphic character states as diagnostic features.

Keywords: G+C content; Genome BLAST Distance Phylogeny; chemotaxonomy; genome size; morphology; phylogenetic systematics; phylogenomics.

Figure 1

First part of the phylogenomic tree inferred with GBDP. Tree inferred with FastME from GBDP distances calculated from whole proteomes. The numbers above branches are GBDP pseudo-bootstrap support values from 100 replications. Tip colors indicate type species of genera, colors to the right of the tips indicate, from left to right, class, order and family (see the embedded legend for details, Figure 2 for the families). The blue gradient toward the far right indicates the exact G+C content as calculated from the genome sequences, followed by black bars indicating the (approximate) genome size in bp. The parts of the tree which have been collapsed here are shown in Figures 2–8.

Figure 2

Second part of the phylogenomic tree inferred with GBDP. A detailed description is provided in the caption of Figure 1. The parts of the tree which have been collapsed here are shown in other figures as indicated.

Figure 3

Third part of the phylogenomic tree inferred with GBDP. A detailed description is provided in the caption of Figure 1. The parts of the tree which have been collapsed here are shown in other figures as indicated.

Figure 4

Fourth part of the phylogenomic tree inferred with GBDP. A detailed description is provided in the caption of Figure 1. The parts of the tree which have been collapsed here are shown in other figures as indicated.

Figure 5

Fifth part of the phylogenomic tree inferred with GBDP. A detailed description is provided in the caption of Figure 1. The parts of the tree which have been collapsed here are shown in other figures as indicated.

Figure 6

Sixth part of the phylogenomic tree inferred with GBDP. A detailed description is provided in the caption of Figure 1. The parts of the tree which have been collapsed here are shown in other figures as indicated.

Figure 7

Seventh part of the phylogenomic tree inferred with GBDP. A detailed description is provided in the caption of Figure 1. The parts of the tree which have been collapsed here are shown in other figures as indicated.

Figure 8

Eighth part of the phylogenomic tree inferred with GBDP. A detailed description is provided in the caption of Figure 1. The parts of the tree which have been collapsed here are shown in other figures as indicated.

Figure 9

(A) Percent G+C content in dependence on genome size (in numbers of base pairs). (B) Phylogenetically independent contrasts (PICs) derived from logit-transformed proportions of G+C content in dependence on PICs derived from log-transformed genome sizes. PICs were calculated to account for the significant effect of the phylogeny on both G+C content and genome size. In both panels, Loess-based bootstrap aggregation (100 replicates) was applied; the resulting red curve denotes the average from 100 smoothers, each fitted to a subset of the original data set, to increase stability and reduce overfitting.

Figure 10

Hypothetical trees and character-state distributions illustrating the impossibility to detect “diagnostic” character states for monophyletic groups even in the case of a perfect fit of the character to the phylogeny. Most-parsimonious reconstructions of changes between character states are indicated by arrows. (A) Maximally symmetric tree with eight tips and a binary character without homoplasies. Each of the two character states is diagnostic for a clade. The orange-red state is actually plesiomorphic but accidentally happens to occur in a single clade only. (B) Same tree as before together with another binary character without homoplasies. Only one of the two character states is diagnostic for a monophyletic group. The orange-red state is again plesiomorphic, hence it comes as no surprise that it characterizes a paraphyletic group. (C) Same tree as before in conjunction with a multi-state character without homoplasies. Only two of the four states, light blue and dark green, are diagnostic, each for a distinct clade. (C) Maximally asymmetric tree with eight tips in conjunction with another multi-state character without homoplasies. Only one of the four states, light blue, is diagnostic for a clade.