An excavate root for the eukaryote tree of life

Abstract

Much of the higher-order phylogeny of eukaryotes is well resolved, but the root remains elusive. We assembled a dataset of 183 eukaryotic proteins of archaeal ancestry to test this root. The resulting phylogeny identifies four lineages of eukaryotes currently classified as "Excavata" branching separately at the base of the tree. Thus, Parabasalia appear as the first major branch of eukaryotes followed sequentially by Fornicata, Preaxostyla, and Discoba. All four excavate branch points receive full statistical support from analyses with commonly used evolutionary models, a protein structure partition model that we introduce here, and various controls for deep phylogeny artifacts. The absence of aerobic mitochondria in Parabasalia, Fornicata, and Preaxostyla suggests that modern eukaryotes arose under anoxic conditions, probably much earlier than expected, and without the benefit of mitochondrial respiration.

Fig. 1.. A rooted phylogeny of eukaryotes based on eukaryotic proteins of archaeal descent.

A concatenated alignment of 183 proteins with 45,443 aligned positions and 85% overall data occupancy (tables S1 and S2 and data S5) was analyzed by maximum likelihood with eight different evolutionary models. The tree shown was derived using the deduced protein structure partitioned model (6 STR + G). Solid circles indicate nodes with 100% bootstrap support from all models and controls, and branch lengths are drawn to scale as indicated by the scale bar. Bootstrap support values for the major nodes are shown in the table at the top. Controls 1 to 3 used stepwise reduction in ingroup-outgroup distances, while controls 4 to 6 used stepwise reduction in individual tree support for early branching excavates (table S4). Taxon names and all bootstrap values for the 6 STR + G tree are shown in fig. S1.

Fig. 2.. Comparative fit of evolutionary models for the six main protein structure elements.

Goodness of fit based on Bayesian information criteria [BIC; (40)] was calculated using model fit [IQTREE.v1.6.12; (41)] on a 183-protein alignment partitioned into six predicted structural elements using NetSurfP-3.0 (16) and evaluated on the best global tree (Fig. 1). Bars show improvement in BIC scores for various models relative to the LG model (without gamma) and are colored according to their general model type as indicated by the key to the right. The analyses were run on a 20-core CPU and include optimizing branch lengths and model parameters. The numbers of categorical mixture components for each model and CPU time in minutes are shown at the far left of each bar, to the left and right of a slash, respectively. All phylogenetic analyses of the 183 euArc data with these models produce the same tree (Fig. 1 and data S2). Raw values are provided in table S5.

Fig. 3.. A proposed stepwise scenario for the origin of mitochondria and mitochondria-like organelles.

A schematic version of the rooted euArc phylogeny in Fig. 1 is shown with two proposed endosymbiotic events, the earlier most likely involving a γ- and/or δ-proteobacterium (γ-/δ-proteo) and the second an α-proteobacterium (α-proteo).