Mitochondria are not captive bacteria.

Lynn Sagan's conjecture (1967) that three of the fundamental organelles observed in eukaryote cells, specifically mitochondria, plastids and flagella were once free-living primitive (prokaryotic) cells was accepted after considerable opposition. Even though the idea was swiftly refuted for the specific case of origins of flagella in eukaryotes, the symbiosis model in general was accepted for decades as a realistic hypothesis to describe the endosymbiotic origins of eukaryotes. However, a systematic analysis of the origins of the mitochondrial proteome based on empirical genome evolution models now indicates that 97% of modern mitochondrial protein domains as well their homologues in bacteria and archaea were present in the universal common ancestor (UCA) of the modern tree of life (ToL). These protein domains are universal modular building blocks of modern genes and genomes, each of which is identified by a unique tertiary structure and a specific biochemical function as well as a characteristic sequence profile. Further, phylogeny reconstructed from genome-scale evolution models reveals that Eukaryotes and Akaryotes (archaea and bacteria) descend independently from UCA. That is to say, Eukaryotes and Akaryotes are both primordial lineages that evolved in parallel. Finally, there is no indication of massive inter-lineage exchange of coding sequences during the descent of the two lineages. Accordingly, we suggest that the evolution of the mitochondrial proteome was autogenic (endogenic) and not endosymbiotic (exogenic).

Reductive evolution of proteomes and protein structures.

The lengths of orthologous protein families in Eukarya are almost double the lengths found in Bacteria and Archaea. Here we examine protein structures in 745 genomes and show that protein length differences between superkingdoms arise as much shorter prokaryotic nondomain linker sequences. Eukaryotic, bacterial, and archaeal linkers are 250, 86, and 73 aa residues in length, respectively, whereas folded domain sequences are 281, 280, and 256 residues, respectively. Cryptic domains match linkers (P < 0.0001) with probabilities ranging between 0.022 and 0.042; accordingly, they do not affect length estimates significantly. Linker sequences support intermolecular binding within proteomes and they are probably enriched in intrinsically disordered regions as well. Reductively evolved linker sequence lengths in growth rate maximized cells should be proportional to proteome diversity. By using total in-frame coding capacity of a genome [i.e., coding sequence (CDS)] as a reliable measure of proteome diversity, we find linker lengths of prokaryotes clearly evolve in proportion to CDS values, whereas those of eukaryotes are more randomly larger than expected. Domain lengths scarcely change over the entire range of CDS values. Thus, the protein linkers of prokaryotes evolve reductively whereas those of eukaryotes do not.

The RNA dreamtime: modern cells feature proteins that might have supported a prebiotic polypeptide world but nothing indicates that RNA world ever was.

Modern cells present no signs of a putative prebiotic RNA world. However, RNA coding is not a sine qua non for the accumulation of catalytic polypeptides. Thus, cellular proteins spontaneously fold into active structures that are resistant to proteolysis. The law of mass action suggests that binding domains are stabilized by specific interactions with their substrates. Random polypeptide synthesis in a prebiotic world has the potential to initially produce only a very small fraction of polypeptides that can fold spontaneously into catalytic domains. However, that fraction can be enriched by proteolytic activities that destroy the unfolded polypeptides and regenerate amino acids that can be recycled into polypeptides. In this open system scenario the stable domains that accumulate and the chemical environment in which they are accumulated are linked through self coding of polypeptide structure. Such open polypeptide systems may have been the precursors to the cellular ribonucleoprotein (RNP) world that evolved subsequently.

Compensatory gene amplification restores fitness after inter-species gene replacements.

Genes introduced by gene replacements and other types of horizontal gene transfer (HGT) represent a significant presence in many archaeal and eubacterial genomes. Most alien genes are likely to be neutral or deleterious upon arrival and their long-term persistence may require a mechanism that improves their selective contribution. To examine the fate of inter-species gene replacements, we exchanged three native S. typhimurium genes encoding ribosomal proteins with orthologues from various other microbes. The results show that replacement of each of these three genes reduces fitness to such an extent that it would provide an effective barrier against inter-species gene replacements in eubacterial populations. However, these fitness defects could be partially ameliorated by gene amplification that augmented the dosage of the heterologous proteins. This suggests that suboptimal expression is a common fitness constraint for inter-species gene replacements, with fitness costs conferred by either a lower expression level of the alien protein compared with the native protein or a requirement for an increased amount of the alien protein to maintain proper function. Our findings can explain the observation that duplicated genes are over-represented among horizontally transferred genes, and suggest a potential coupling between compensatory gene amplification after HGT and the evolution of new genes.

Deletion rate evolution and its effect on genome size and coding density.

Deletion rates are thought to be important factors in determining the genome size of organisms in nature. Although it is indisputable that deletions, and thus deletion rates, affect genome size, it is unclear how, or indeed if, genome size is regulated via the deletion rate. Here, we employ a mathematical model to determine the evolutionary fate of deletion rate mutants. Simulations are employed to explore the interactions between deletions, deletion rate mutants, and genome size. The results show that, in this model, the fate of deletion rate mutants will depend on the fraction of essential genomic material, on the frequency of sexual recombination, as well as on the population size of the organism. We find that there is no optimal deletion rate in any state. However, at one critical coding density, all changes in deletion rate are neutral and the rate may drift either up or down. As a consequence, the coding density of the genome is expected to fluctuate around this critical density. Characteristic differences in the impact of deletion rate mutations on prokaryote and eukaryote genomes are described.

The modern RNP world of eukaryotes.

Eukaryote gene expression is mediated by a cascade of RNA functions that regulate, process, store, transport, and translate RNA transcripts. The RNA network that promotes this cascade depends on a large cohort of proteins that partner RNAs; thus, the modern RNA world of eukaryotes is really a ribonucleoprotein (RNP) world. Features of this "RNP infrastructure" can be related to the high cytosolic density of macromolecules and the large size of eukaryote cells. Because of the densely packed cytosol or nucleoplasm (with its severe restriction on diffusion of macromolecules), partitioning of the eukaryote cell into functionally specialized compartments is essential for efficiency. This necessitates the association of RNA and protein into large RNP complexes including ribosomes and spliceosomes. This is well illustrated by the ubiquitous spliceosome for which most components are conserved throughout eukaryotes and which interacts with other RNP-based machineries. The complexes involved in gene processing in modern eukaryotes have broad phylogenetic distributions suggesting that the common ancestor of extant eukaryotes had a fully evolved RNP network. Thus, the eukaryote genome may be uniquely informative about the transition from an earlier RNA genome world to the modern DNA genome world.

Akaryotes and Eukaryotes are independent descendants of a universal common ancestor.

We reconstructed a global tree of life (ToL) with non-reversible and non-stationary models of genome evolution that root trees intrinsically. We implemented Bayesian model selection tests and compared the statistical support for four conflicting ToL hypotheses. We show that reconstructions obtained with a Bayesian implementation (Klopfstein et al., 2015) are consistent with reconstructions obtained with an empirical Sankoff parsimony (ESP) implementation (Harish et al., 2013). Both are based on the genome contents of coding sequences for protein domains (superfamilies) from hundreds of genomes. Thus, we conclude that the independent descent of Eukaryotes and Akaryotes (archaea and bacteria) from the universal common ancestor (UCA) is the most probable as well as the most parsimonious hypothesis for the evolutionary origins of extant genomes. Reconstructions of ancestral proteomes by both Bayesian and ESP methods suggest that at least 70% of unique domain-superfamilies known in extant species were present in the UCA. In addition, identification of a vast majority (96%) of the mitochondrial superfamilies in the UCA proteome precludes a symbiotic hypothesis for the origin of eukaryotes. Accordingly, neither the archaeal origin of eukaryotes nor the bacterial origin of mitochondria is supported by the data. The proteomic complexity of the UCA suggests that the evolution of cellular phenotypes in the two primordial lineages, Akaryotes and Eukaryotes, was driven largely by duplication of common superfamilies as well as by loss of unique superfamilies. Finally, innovation of novel superfamilies has played a surprisingly small role in the evolution of Akaryotes and only a marginal role in the evolution of Eukaryotes.

Empirical genome evolution models root the tree of life.

A reliable phylogenetic reconstruction of the evolutionary history of contemporary species depends on a robust identification of the universal common ancestor (UCA) at the root of the Tree of Life (ToL). That root polarizes the tree so that the evolutionary succession of ancestors to descendants is discernable. In effect, the root determines the branching order and the direction of character evolution. Typically, conventional phylogenetic analyses implement time-reversible models of evolution for which character evolution is un-polarized. Such practices leave the root and the direction of character evolution undefined by the data used to construct such trees. In such cases, rooting relies on theoretic assumptions and/or the use of external data to interpret unrooted trees. The most common rooting method, the outgroup method is clearly inapplicable to the ToL, which has no outgroup. Both here and in the accompanying paper (Harish and Kurland, 2017) we have explored the theoretical and technical issues related to several rooting methods. We demonstrate (1) that Genome-level characters and evolution models are necessary for species phylogeny reconstructions. By the same token, standard practices exploiting sequence-based methods that implement gene-scale substitution models do not root species trees; (2) Modeling evolution of complex genomic characters and processes that are non-reversible and non-stationary is required to reconstruct the polarized evolution of the ToL; (3) Rooting experiments and Bayesian model selection tests overwhelmingly support the earlier finding that akaryotes and eukaryotes are sister clades that descend independently from UCA (Harish and Kurland, 2013); (4) Consistent ancestral state reconstructions from independent genome samplings confirm the previous finding that UCA features three fourths of the unique protein domain-superfamilies encoded by extant genomes.

The phylogenomics of protein structures: The backstory.

In this introductory retrospective, evolution as viewed through gene trees is inspected through a lens compounded from its founding operational assumptions. The four assumptions of the gene tree culture that are singularly important to evolutionary interpretations are: a. that protein-coding sequences are molecular fossils; b. that gene trees are equivalent to species trees; c. that the tree of life is assumed to be rooted in a simple akaryote cell implying that akaryotes are primitive, and d. that the notion that all or most incongruities between alignment-based gene trees are due to horizontal gene transfer (HGT), which includes the endosymbiotic models postulated for the origins of eukaryotes. What has been unusual about these particular assumptions is that though each was taken on board explicitly, they are defended in the face of factual challenge by a stolid disregard for the conflicting observations. The factual challenges to the mainstream gene tree-inspired evolutionary view are numerous and most convincingly summarized as: Genome trees tell a very different story. Phylogeny inferred from genomic assortments of homologous protein structural-domains does not support any one of the four principle evolutionary interpretations of gene trees: a. 3D protein domain structures are the molecular fossils of evolution, while coding sequences are transients; b. Species trees are very different from gene trees; c. The ToL is rooted in a surprisingly complex universal common ancestor (UCA) that is distinct from any specific modern descendant and d. HGT including endosymbiosis is a negligible player in genome evolution from UCA to the present.

Rooted phylogeny of the three superkingdoms.

The traditional bacterial rooting of the three superkingdoms in sequence-based gene trees is inconsistent with new phylogenetic reconstructions based on genome content of compact protein domains. We find that protein domains at the level of the SCOP superfamily (SF) from sequenced genomes implement with maximum parsimony fully resolved rooted trees. Such genome content trees identify archaea and bacteria (akaryotes) as sister clades that diverge from an akaryote common ancestor, LACA. Several eukaryote sister clades diverge from a eukaryote common ancestor, LECA. LACA and LECA descend in parallel from the most recent universal common ancestor (MRUCA), which is not a bacterium. Rather, MRUCA presents 75% of the unique SFs encoded by extant genomes of the three superkingdoms, each encoding a proteome that partially overlaps all others. This alone implies that the common ancestor to the superkingdoms was very complex. Such ancestral complexity is confirmed by phylogenetic reconstructions. In addition, the divergence of proteomes from the complex ancestor in each superkingdom is both reductive in numbers of unique SFs as well as cumulative in the abundance of surviving SFs. These data suggest that the common ancestor was not the first cell lineage and that modern global phylogeny is the crown of a "recently" re-rooted tree. We suggest that a bottlenecked survivor of an environmental collapse, which preceded the flourishing of the modern crown, seeded the current phylogenetic tree.

What tangled web: barriers to rampant horizontal gene transfer.

Dawkins in his The Selfish Gene(1) quite aptly applies the term "selfish" to parasitic repetitive DNA sequences endemic to eukaryotic genomes, especially vertebrates. Doolittle and Sapienza(2) as well as Orgel and Crick(3) enlivened this notion of selfish DNA with the identification of such repetitive sequences as remnants of mobile elements such as transposons. In addition, Orgel and Crick(3) associated parasitic DNA with a potential to outgrow their host genomes by propagating both vertically via conventional genome replication as well as infectiously by horizontal gene transfer (HGT) to other genomes. Still later, Doolittle(4) speculated that unchecked HGT between unrelated genomes so complicates phylogeny that the conventional representation of a tree of life would have to be replaced by a thicket or a web of life.(4) In contrast, considerable data now show that reconstructions based on whole genome sequences are consistent with the conventional "tree of life".(5-10) Here, we identify natural barriers that protect modern genome populations from the inroads of rampant HGT.

On the origin of mitochondria: a genomics perspective.

The availability of complete genome sequence data from both bacteria and eukaryotes provides information about the contribution of bacterial genes to the origin and evolution of mitochondria. Phylogenetic analyses based on genes located in the mitochondrial genome indicate that these genes originated from within the alpha-proteobacteria. A number of ancestral bacterial genes have also been transferred from the mitochondrial to the nuclear genome, as evidenced by the presence of orthologous genes in the mitochondrial genome in some species and in the nuclear genome of other species. However, a multitude of mitochondrial proteins encoded in the nucleus display no homology to bacterial proteins, indicating that these originated within the eukaryotic cell subsequent to the acquisition of the endosymbiont. An analysis of the expression patterns of yeast nuclear genes coding for mitochondrial proteins has shown that genes predicted to be of eukaryotic origin are mainly translated on polysomes that are free in the cytosol whereas those of putative bacterial origin are translated on polysomes attached to the mitochondrion. The strong relationship with alpha-proteobacterial genes observed for some mitochondrial genes, combined with the lack of such a relationship for others, indicates that the modern mitochondrial proteome is the product of both reductive and expansive processes.