Rooting for the root of elongation factor-like protein phylogeny

Abstract

Lateral gene transfer (LGT) may play a pivotal role in the evolution of elongation factor-like (EFL) genes in eukaryotes. To date, numbers of putative cases for lateral transfer of EFL genes have been postulated based on unrooted EFL phylogenies. Nevertheless, the root position in EFL phylogeny is important to validate lateral EFL gene transfer: for instance, a clade of two EFL homologs from distantly related organisms in an unrooted EFL tree does not necessarily confirm the LGT, since the possibility that the root may locate in this clade cannot be excluded. Cocquyt et al. (2009, p. 39) recently demonstrated that a putative case of lateral EFL gene transfer, which was originally proposed based on an unrooted phylogeny, could not be endorsed by the corresponding rooted analysis. Although rooting EFL phylogeny is indispensable to elucidate various aspects in EFL gene evolution, we suspected that the outgroup clade comprised of EF-1alpha and eukaryote-specific EF-1alpha paralogs erroneously attached to long EFL branches in Cocquyt et al. (2009) - a typical long branch attraction (LBA) artifact. Here, we systematically assessed the putative LBA artifact between the branch leading to the outgroup clade and long ingroup branches by analyzing the original dataset used in Cocquyt et al. (2009) with and without modifying ingroup-sequence sampling. A series of the rooted EFL analyses indicated that the root inference was highly susceptible to presence and absence of long-branched ingroup-sequences, suggesting that the rooted EFL phylogenies cannot be free from severe LBA artifact. We also discussed a new aspect in EFL gene evolution in stramenopiles identified in the course of the EFL analyses described above. Finally, the relative timing of the first emergence of EFL gene in eukaryotes was contemplated based on the current EF-1alpha/EFL distribution.