Changing perspectives on the origin of eukaryotes Laura A. KatzA Trends in Ecology and Evolution 1998, 13:493-497 Dept of Biological Sciences, Smith College, Northampton, MA 01063, USA, and is a member of the Program in Organismic and Evolutionary Biology, University of Massachusetts at Amherst, Amherst, MA 01003, USA A lkatz@smith.edu

Abstract

From the initial application of molecular techniques to the study of microbial organisms, three domains of life emerged, with eukaryotes and archaea as sister taxa. However, recent analyses of an expanding molecular data set reveal that the eukaryotic genome is chimeric with respect to archaea and bacteria. Moreover, there is now evidence that the primitive eukaryotic group `Archezoa' once harbored mitochondia. These discoveries have challenged the traditional stepwise model of the evolution of eukaryotes, in which the nucleus and microtubules evolve before the acquisition of mitochondria, and consequently compel a revision of existing models of the origin of eukaryotic cells.

The past 20 years have seen dramatic changes in our view of the origin and evolution of eukaryotes as the so-called `Woesian revolution' has resulted in a reclassification of life into three domains [Archaebacteria (Archaea), Eubacteria (Bacteria) and Eukaryotes (Eukarya)] based on variation in the small subunit ribosomal RNA (ssu-rRNA) gene 1. The ssu-rRNA genealogies indicate that eukaryotes, defined by the presence of nuclei and microtubules, are a genetically diverse group of organisms2. Both fossil3 and molecular4 data establish the origin of eukaryotes at approximately two billion years ago. Analyses of the ssu-rRNA gene have been particularly important for interpreting relationships among organisms where morphology and ultrastrucuture have failed to provide consistent hypotheses, and for providing a framework to test hypotheses on the origin and diversification of eukaryotes25. These analyses supported a model of the evolution of eukaryotic cells in which mitochondria were acquired only after the divergence of several extant eukaryotic lineages (Fig 1).

Figure 1. `Tree of life' based on the small subunit ribosomal RNA (ssu-rRNA) gene indicating the stepwise model of the origin of eukaryotes. (1) Evolution of nuclei and microtubules; (2) acquisition of mitochondria. Genealogy redrawn, with permission, from M.L. Sogin (unpublished).

Although the ssu-rRNA genealogy alone fails to provide a rooting for the tree of life, analyses of ancient gene duplications of protein-coding genes (ATPases and elongation factors) supported the sister status of archaea and eukaryotes5. Hence, under the Woesian revolution, a picture emerged of living organisms representing three major lineages (two prokaryotic and one eukaryotic), in which the eukaryotes and archaea share a more recent common ancestor with each other than either does with bacteria. However, recent analyses of single genes characterized from divergent taxa and complete genome sequences of a handful of taxa, coupled with changing perspectives on the interactions of early organisms, demand a revision of our ideas on the origin of eukaryotes.


Chimeric nature of eukaryotic genomes

The recent accumulation of DNA sequence data from numerous genes has made it clear that eukaryotic genomes are chimeric with respect to archaea and bacteria678910. A genome is chimeric when a gene (or genes) within an organism does not simply trace the history of vertical transmission of genetic information from one generation to the next. Instead, chimeric genomes are the result of lateral transmission of genes (or genomes) across species boundaries. As a consequence of lateral gene transfer, phylogenetic analyses of multiple genes generate conflicting gene genealogies, obscuring the evolutionary history of the organisms. Some eukaryotic genes are more similar to archaeal genes, whereas others appear to share a more recent common ancestor with bacteria.

Evidence of chimerism first arose from analysis of the DNA-dependent RNA polymerase gene family6. This observation was supported by analysis of biochemical and molecular features (e.g. gene order, promoter types and sequence features), with sequences from glycolytic enzymes generally uniting bacteria and eukaryotes, and features associated with the genetic machinery of eukaryotic cells uniting eukaryotes and archaea7. Subsequent analyses of 24 protein coding genes also supported the chimeric nature of genomes, with nine genealogies uniting archaea and eukaryotes, seven supporting the sister status of eukaryotes and gram negative bacteria, and eight genealogies being unresolved8.

In an extremely extensive study, Brown and Doolittle9 analysed genealogies from 66 protein-coding genes for which sequences had been characterized from members of all three domains of life. Of the 56 genealogies with significantly different interdomain distances, 31 supported the sister status of eukaryotes and archaea, whereas seven united eukaryotes with bacteria9. In all these analyses6789 (and other comparisons from complete genome sequences), genes that support specific relationships between prokaryotes and eukaryotes were classed by function: genes that support the sister status of archaea and eukaryotes tend to be involved in the genetic machinery of the cell, whereas genes that unite bacteria and eukaryotes are more likely to regulate metabolic processes8911. These conflicting genealogies (Fig 2) challenge traditional views in which vertical transmission of genetic material from one generation to the next is the predominant force in evolution.

Figure 2. Representative gene genealogies indicating the chimeric nature of eukaryotes (data compiled from Refs 48 and 9). Genealogies in the first column group the eukaryotes with archaea and those in the second column unite the eukaryotes with bacteria.

There are three explanations for conflicting gene trees and species trees: (1) failed methodology, (2) maintenance of ancestral polymorphism resulting from ancient gene duplications, and (3) lateral gene transfers. On a gene by gene basis, it might be possible to argue for one of the first two explanations, but, overall, lateral gene transfer is necessary to explain the growing number of conflicting gene genealogies.

Although biases and inadequacies of individual alignment and phylogenetic algorithms can obscure the reconstruction of gene trees12, the multiple approaches taken by researchers make it unlikely that failed methodology explains the growing number of conflicting gene trees. Gene genealogies can also be misleading if the proteins sampled represent paralogs rather than orthologs; that is, if the analysis fails to account for ancient gene duplication13. Given the emerging data from complete genome sequences, as well as the increased sampling of individual genes from distantly related taxa, the notion that the apparent chimerism is caused solely by inadequate identification of duplicated gene families is also unlikely.

This leaves the final explanation—that eukaryotic genomes are chimeric because of lateral transfer across species boundaries of either individual genes or even portions of (or complete) genomes. The large number of genes that support differing relationships between eukaryotes and prokaryotes (Fig 2) imply a significant level of lateral transfer among lineages14. Abundant evidence for the occurrence of lateral gene transfers comes from data on the secondary transfer of plastid genomes among eukaryotes15, movement of transposable elements among taxa16 and the lateral transfer of individual genes1718. Because of extensive lateral transfer, eukaryotes cannot be considered a bifurcating lineage emerging through vertical transmission of genetic material from a common ancestor of either archaea or bacteria. Rather, eukaryotes appear to contain a mixture of genes from the genomes of both of these lineages. Instead of the three `domains' of life that emerged from the Woesian revolution, we must change our views to accept a tangled web of genes within genomes, particularly with respect to early diverging eukaryotes (Fig 3).

Figure 3. Tangled web of life: a revised model for the origin of eukaryotes indicating a fusion of (or symbiosis between) archaeal and bacterial lineages, and the potential simultaneous origin of nuclei, microtubules and mitochondria in eukaryotes. Bold lines represent lineage boundaries; gray lines are gene genealogies; broken arrows are possible lateral transfers of individual genes. The later acquisition of chloroplasts is also indicated.

Changing perspectives on Archezoa

In addition to changing our views on the nature of eukaryotic genomes, we must also contend with the abandonment of the eukaryotic group `Archezoa' (not to be confused with the prokaryotic group archaea). Based on analysis of the ssu-rRNA gene219, several amitochondrial taxa were shown to represent the basal lineages of eukaryotes (Fig 1), including diplomonads (e.g. Giardia lamblia), trichomonads (e.g. Trichomonas vaginalis) and microsporidians (e.g. Vairimorpha necatrix). These amitochondrial taxa have been grouped together as the kingdom Archezoa—eukaryotes that had evolved a nucleus and microtubules but had yet to acquire mitochondria20. Several recent lines of evidence are responsible for the demise of Archezoa: (1) genes of mitochondrial origin have been identified within the nucleus of Archezoa; (2) phylogenetic analyses of numerous loci suggest that some putative members of the Archezoa could be more recently derived eukaryotes; and (3) biochemical evidence indicates that hydrogenosomes (hydrogen-producing organelles found in some Archezoa) and mitochondria are homologous.

The discovery of genes of mitochondrial origin in microsporidians2122, trichomonads2324 and diplomonads252627 indicates that Archezoa once harbored mitochondria, and that before the loss of the organelle, mitochondrial-specific genes were transferred into the nucleus. In addition, analyses of protein coding genes have challenged the phylogenetic position of at least one type of Archezoan, the microsporidia. For example, genealogies constructed with both -tubulin and -tubulin sequences2829 revealed that microsporidia are highly-derived fungi rather than early-diverging eukaryotes. Finally, analyses of hydrogenosomes have led to the hypothesis that these organelles are highly modified relatives of extant mitochondria303132.

Taken together, these observations dispute the hypothesis that the earliest diverging lineages of eukaryotes had yet to acquire mitochondria33. The implications of the loss of the kingdom Archezoa are profound: instead of a stepwise model of the evolution of eukaryotes, in which the nucleus and microtubules evolved before the acquisition of mitochondria, we must now consider the possibility that the nuclei, mitochondria and microtubules of eukaryotes had a simultaneous origin.


Theories on the origins of eukaryotes

Models of the origin of eukaryotic cells have focused on the evolution of the nucleus and microtubules. The data on the chimeric nature of eukaryotic genomes, combined the possibility that the acquisition of mitochondria occurred simultaneously with the emergence of eukaryotes, require that we re-evaluate these models for their ability to explain four characteristics of the last common ancestor of all extant eukaryotes: the presence of a nucleus, microtubules, mitochondria and a chimeric genome (1).

Table 1. Representative models of the origin of eukaryotesa
Model	Organism 1	Organism 2	Mode of interaction	Resulting organism	Origin of nucleus	Origin of microtubules	Origin of mitochrondria	Origin of genome	Refs

Perhaps the simplest models of the origin of eukaryotes are those that focus solely on the chimeric nature of eukaryotic genomes. Without specifically addressing the origin of the nucleus, mitochondria or microtubules, Zillig7 suggested a `fusion' of an archaeon and a bacterium in the generation of the chimeric eukaryotic genome (1a). Similarly, Woese34 argued for a model of `genetic annealing', in which, early on when evolutionary temperatures were `high,' extensive lateral transfers among lineages created the observed chimerism. Under this model, the `cooling' of evolutionary temperatures, which corresponds to the increased complexity of the evolving lineages, dramatically slowed the rate of transfers among lineages34.

Genealogical analyses, first of heat shock proteins35 and then of 24 protein coding genes8, led Gupta and colleagues to argue that eukaryotes represented a fusion of a gram negative bacterium and an archaeon (most likely an eocyte). According to their model, the membranes used by a gram negative bacterium to engulf an archaeon were eventually internalized to form the endoplasmic reticulum and nucleus36 (1b). Gupta37 further explained the observed association of genealogies with function (e.g. that genes involved in DNA processing tend to unite archaea with eukaryotes78911) by arguing that after the fusion of the bacterium and archaeon, selection acted to sort genes by function, such that genes involved in specific functions (e.g. translation or transcription) were inherited as a block from one of the fusion partners.

Even more intriguing is the theory that the DNA-based eukaryotic genome is a relatively recent innovation that resulted from an endosymbiosis involving organisms with RNA-based genomes38. Comparisons of conflicting gene genealogies led Sogin39 to argue that the eukaryotic genome was acquired through endosymbiosis between the proto-eukaryote (still harboring an RNA-based genome) and an archaeon with a DNA-based genome39. This endosymbiosis gave rise to both the chimeric genome, with sequences of genes involved in DNA processing uniting eukaryotes and archaea, and the nucleus (1c). Although Sogin noted that the cytoskeleton is required for the host to engulf the prokaryote, no explanation was given for the evolution of the cytoskeleton itself.

Under the `serial endosymbiosis theory' (SET), championed by Margulis4041, endosymbiosis explains the origin of both microtubules and mitochondria. Margulis argued that eukaryotes originated first from a symbiotic relationship between a host archaeon (with a nucleoid membrane) and an endosymbiotic spirochete, with the spirochete providing mobility to the archaeon4041 (1d). The chimeric organisms generated by this symbiosis only later acquired a respiring bacterial symbiont that eventually evolved into extant mitochondria4041 (1d). This second endosymbiosis is well supported by the structure of mitochondria, the presence of a circular mitochondrial genome (in many eukaryotes) and the sequence of mitochondrial DNA.

Comparative biochemistry of early diverging eukaryotes suggests a novel mechanism for the initial association between organisms that led to the acquisition of mitochondria. Linking the metabolic properties of extant amitochondrial eukaryotes and the bacteria harbored within them, Martin and Müller42 argued that the original symbiotic relationship that led to mitochondria was between a bacterium that generated hydrogen as a waste product and a hydrogen-dependent archaean (1e). Over time, the `host cell' engulfed the bacterium and the genetic system of the symbiont shifted to the host to allow for the coordinated control of metabolism42. Likewise, comparisons of extant symbioses caused Searcy43 to argue that the original symbiosis that gave rise to mitochondria was driven by sulfur-based metabolism. In both cases, the theories account only for the acquisition of mitochondria and do not discuss the evolution of other eukaryotic-specific features.


Conclusions and perspectives

Recent data on the nature of eukaryotic genomes and the timing of the acquisition of mitochondria compel us to transform our views on the origin of eukaryotes. It is now possible to assume that the original endosymbiotic event that gave rise to the mitochondria (whether for respiration40, hydrogen-dependent metabolism42 or sulfur-dependent metabolism43) occurred in the ancestor of all extant eukaryotes, and that this endosymbiosis explains both the chimeric nature of eukaryotic genomes and the origin of mitochondria. From comparative studies of genomes4445 and experimental studies in yeast46, we know that genes can readily move from organelles to the nucleus. Consequently, following the endosymbiosis that gave rise to mitochondria, the transfer of a significant portion of the symbiont's genome into the nucleus can explain the chimeric nature of contemporary eukaryotic genomes10. Although subsequent lateral transfers may well have obscured this early event, it is possible to test this theory by extensive sampling of both prokaryotic and eukaryotic genomes, and by focusing on potentially early-diverging eukaryotes and on prokaryotes presumed to be within the same clade as the ancestors of mitochondria.

Less clear is the origin of two of the other key characteristics of eukaryotes, the nucleus and microtubules. There are at least three explanations for the evolution of the nucleus: (1) the nucleus is derived from a nucleoid structure found in archaea40; (2) the nucleus evolved through invagination of membranes within the lineage that gave rise to eukaryotes1947; or (3) the nucleus directly resulted from the engulfment of one organism by the other3639. Although further research into the nature of archaea might allow testing of the first theory, distinguishing between the other hypotheses is likely to prove more difficult.

Weaker still are data on the evolution of microtubules. The origin of this defining eukaryotic feature can be hypothesized as the result of either endosymbiosis40 or autogenic processes1920. However, neither theory adequately explains the observed level of divergence between microtubule proteins and any potential prokaryotic homolog. A possible homolog for tubulin (one of the major proteins in microtubules) has been found in prokaryotes48. However, the level of similarity between the protein and its eukaryotic counterpart is low, which implies a dramatic acceleration in the rate of evolution of microtubular proteins as they first evolved in eukaryotes relative to their subsequent slow rate of evolution within eukaryotes48. Regardless of whether microtubules were obtained through endosymbiosis or evolved directly within eukaryotes, there is no satisfying explanation for the observed pattern of microtubular protein evolution.

We now know that individual gene genealogies alone cannot provide a `tree of life'. Instead, these genealogies spin a tangled web of the history of genes within organisms. The next steps are to develop more sophisticated methods to interpret conflict among multiple gene genealogies, and to augment molecular data with studies of the cell biology of early diverging eukaryotes.



Acknowledgements

I wish to thank J.T. Bonner and D.H. Berger and three anonymous reviewers for their comments on this article. The work was supported by a grant to the Smith College Dept of Biological Sciences from the Albert F. Blakeslee Fund (administered by the National Academy of Sciences, USA).