Was the Last Bacterial Common Ancestor a Monoderm after All?

Abstract

The very nature of the last bacterial common ancestor (LBCA), in particular the characteristics of its cell wall, is a critical issue to understand the evolution of life on earth. Although knowledge of the relationships between bacterial phyla has made progress with the advent of phylogenomics, many questions remain, including on the appearance or disappearance of the outer membrane of diderm bacteria (also called Gram-negative bacteria). The phylogenetic transition between monoderm (Gram-positive bacteria) and diderm bacteria, and the associated peptidoglycan expansion or reduction, requires clarification. Herein, using a phylogenomic tree of cultivated and characterized bacteria as an evolutionary framework and a literature review of their cell-wall characteristics, we used Bayesian ancestral state reconstruction to infer the cell-wall architecture of the LBCA. With the same phylogenomic tree, we further revisited the evolution of the division and cell-wall synthesis (dcw) gene cluster using homology- and model-based methods. Finally, extensive similarity searches were carried out to determine the phylogenetic distribution of the genes involved with the biosynthesis of the outer membrane in diderm bacteria. Quite unexpectedly, our analyses suggest that all cultivated and characterized bacteria might have evolved from a common ancestor with a monoderm cell-wall architecture. If true, this would indicate that the appearance of the outer membrane was not a unique event and that selective forces have led to the repeated adoption of such an architecture. Due to the lack of phenotypic information, our methodology cannot be applied to all extant bacteria. Consequently, our conclusion might change once enough information is made available to allow the use of an even more diverse organism selection.

Keywords: Bayesian inference (BI); ancestral traits; bacterial evolution; cell-wall; comparative genomics; outer membrane (OM); phylogenomics.

Figure 1

Phylogenomic tree of the bacterial domain based on a supermatrix concatenating 117 single-copy orthologous genes chosen for their broad conservation across Bacteria. The tree was rooted on Terrabacteria. The supermatrix had 85 species and 19,959 unambiguously aligned amino-acid positions (<5% missing character states). The tree was inferred from amino-acid sequences using PhyloBayes MPI and the CAT+GTR+Γ model of sequence evolution. Open symbols at the nodes are posterior probabilities (PP), and nodes without a symbol correspond to maximum statistical support for phylogenetic inference (posterior probabilities of 1.0; averaged over two MCMC chains). The length of the branch marked with “//” has been reduced by 50% for the sake of clarity. Colour key is red = Terrabacteria, orange = Proteobacteria, green = FBC superphylum, blue = PVC superphylum. Outer circles stand for the status of the peptidoglycan (PG) and of the outer membrane in the organisms, according to our literature survey: red = thick PG, orange = thin PG, yellow = no PG, dark blue = diderm, light blue = monoderm, white = no information. Alternating white and grey backgrounds highlight the alternance between differentially coloured groups or phyla.

Figure 2

Cladogram derived from the tree of Figure 1 featuring the cell-wall architecture inferred for selected last common ancestors among Bacteria. Colour key is red = Terrabacteria, orange = Proteobacteria, green = FBC superphylum, blue = PVC superphylum Branches ending with a triangle represent collapsed groups (for details, see Figure 1 or Table S3). The pie chart sectors correspond to the posterior probabilities of the model reverse-jump hyperprior exponential 0 to 100 (R2). Colour key is red = thick PG, orange = thin PG, yellow = no PG, dark blue = diderm, light blue = monoderm.

Figure 3

Overview of gene distribution and synteny analyses. (A) ProCARs results for dcw cluster organization in selected LCA among Bacteria. Full rectangle = gene present and in the main cluster; empty circle in rectangle = gene present but in a sub-cluster; empty rectangle = gene present but outside of any cluster. Note that the reconstruction procedure prevents the complete lack of a gene in an ancestral genome. (B) Recurring distribution patterns at the phylum level for the proteins involved with the outer membrane. Full circle = gene present in the group; empty circle = gene absent in the group; “?” in a circle = potential presence of the gene in the group; /// = presence in a sub-group only (i.e., Deinococcus-Thermus). Numbers in bold are the pattern numbers. Names written in bold are the names of groups regrouping several phyla.