Comparative analysis of proteomes and functionomes provides insights into origins of cellular diversification

Abstract

Reconstructing the evolutionary history of modern species is a difficult problem complicated by the conceptual and technical limitations of phylogenetic tree building methods. Here, we propose a comparative proteomic and functionomic inferential framework for genome evolution that allows resolving the tripartite division of cells and sketching their history. Evolutionary inferences were derived from the spread of conserved molecular features, such as molecular structures and functions, in the proteomes and functionomes of contemporary organisms. Patterns of use and reuse of these traits yielded significant insights into the origins of cellular diversification. Results uncovered an unprecedented strong evolutionary association between Bacteria and Eukarya while revealing marked evolutionary reductive tendencies in the archaeal genomic repertoires. The effects of nonvertical evolutionary processes (e.g., HGT, convergent evolution) were found to be limited while reductive evolution and molecular innovation appeared to be prevalent during the evolution of cells. Our study revealed a strong vertical trace in the history of proteins and associated molecular functions, which was reliably recovered using the comparative genomics approach. The trace supported the existence of a stem line of descent and the very early appearance of Archaea as a diversified superkingdom, but failed to uncover a hidden canonical pattern in which Bacteria was the first superkingdom to deploy superkingdom-specific structures and functions.

Figure 1

Overview of the comparative proteomics and functionomics methodology. Proteomes and functionomes were scanned for the occurrence and abundance of FSFs and GO terms (i.e., traits). This information was represented in data matrices that were analyzed for trends of trait sharing and traces of vertical and horizontal inheritance. Inferences were drawn regarding superkingdom diversification and were confirmed with previously published phylogenetic studies.

Figure 2

Global trends of trait sharing in Venn taxonomic groups. (a) Venn diagram displaying the distribution of 1,733 FSF domains in 981 completely sequenced proteomes sampled from 652 Bacteria, 70 Archaea, and 259 Eukarya. This constituted the structure dataset. (b) Venn diagram displaying the distribution of 1,924 terminal-level GOs in 249 free-living functionomes corresponding to 183 Bacteria, 45 Archaea, and 21 Eukarya. This constituted the function dataset.

Figure 3

The spread of FSF domain structures (a) and GO terminal terms (b) in the proteomes and functionomes of each member of the superkingdom in the seven Venn taxonomic groups (panels ABE, AB, AE, BE, A, B, and E). Shaded regions indicate that FSFs or GOs were present in >80% of the proteomes (f > 0.8), and their numbers, n ₁ and n ₂. Numbers in boxplots of each distribution indicate group medians. Numbers in red suggest the strongest vertical evolutionary trace.

Figure 4

Boxplots comparing the log-transformed abundance values of structural (a) and functional (b) traits in the proteomes and functionomes of the seven Venn taxonomic groups. Italicized characters identify outliers with maximum and minimum abundance of traits in each group: a, Takifugu rubripes; b, Cand. Hodgkinia cicadicola Dsem; c, Mycoplasma genitalium G37; d, Zea mays; e, Mycobacterium marinum; f, Guillardia theta; g, Homo sapiens; h, Rhodospirillum rubrum; i, Desulfurococcus kamchatkensis; j, Ralstonia eutropha; k, Thermosipho africanus.

Figure 5

Bar plots illustrating the breakdown of terminal GOs in the seven taxonomic groups for level 1 GO terms. A total of 1,871 out of 1,924 GOs (97.24%) could be reliably mapped to their parents. Level 1 GOs that could not be mapped include “D-alanyl carrier activity [GO:0036370],” “electron carrier activity [GO:0009055],” “chemoattractant activity [GO:0042056],” “chemorepellent activity [GO:0045499],” and “nutrient reservoir activity [GO:0045735].” Note that terminal GOs may have more than one parent. The Venn diagram shows that none of the A, B, AB, and AE taxonomic groups uniquely code for any level 1 GO term.

Figure 6

Evolutionary timelines highlighting the abundance of FSFs in superkingdom taxonomic groups. Evolutionary age (nd) was calculated from a phylogenetic tree of protein domains describing the evolution of 1,733 FSFs (taxa) in 981 organisms (characters) (see [26, 28, 35] for technical details). SCOP alphanumeric identifiers were used to identify the most ancient FSF in each taxonomic group. In case of multiple FSFs of same age, only the FSF with maximum abundance was labeled. c.37.1 is the P-loop containing NTP hydrolase FSF; b.34.1 is the C-terminal domain of transcriptional repressors FSF; a.267.1 is the topoisomerase V catalytic domain-like FSF; a.253.1 is the AF0941-like FSF; d.2.1 is the Lysozyme-like FSF; a.47.5 is the FlgN-like FSF; b.6.2 is the major surface antigen p30, SAG1.