New Analyses Favour Mitochondria as Sister to Known Alphaproteobacteria

Guest post by Andrew J. Roger (@andrewjroger)

I'm really excited to share our paper that, we hope, represents an important step in understanding the origin of mitochondria in eukaryotic cells: Site-and-branch-heterogeneous analyses of an expanded dataset favour mitochondria as sister to known Alphaproteobacteria.

This study, led by Sergio Muñoz-Gómez, represents a truly interdisciplinary collaboration involving statistical phylogenetic modeling by Edward Susko, who developed the site-and-branch heterogeneous phylogenetic model and phylogenomics/early evolution expertise contributed by Sergio, Laura Eme, Kelsey Williamson, David Moreira, Purificación López-García (DEEM Team), Claudio Slamovits and metagenomics data and expertise of DEEM Team.

Our study robustly places mitochondria as the immediate sister group of Alphaproteobacteria using phylogenomic analysis of the largest data set to date using Susko's novel site-and-branch heterogeneous model GFmix. A major problem in placing mitochondria in phylogenomic analyses has been the extreme heterogeneity in GC content in the data sets.

Mitochondrial sequences and some alphaproteobacterial sequences (Rickettsiales, Holosporales, and Pelagibacterales) are very A+T-rich. This leads to an enrichment in amino acids specified by AT-rich codons. In contrast, many other alphaproteobacterial and related taxa are either balanced in G+C or G+C-rich -- the latter are enriched in amino acids with GC-rich codons.

Previous analyses have often placed mitochondria as sister to Rickettsiales, Holosporales, and/or Pelagibacteriales and it was never clear whether or not this was the 'true' phylogenetic position or an artefact of shared amino acid composition bias.

Through analyses with compositional bias filtering approaches and GFmix, we show that mitochondria instead emerge immediately outside of a strongly-supported clade of all Alphaproteobacteria (which includes Rickettsiales, Holosporales, and Pelagibacterales), consistent with a recent analysis by Joran Martijn, Thijs Ettema, and colleagues: Deep mitochondrial origin outside the sampled alphaproteobacteria.

In my view, our paper showcases the importance of biologically-realistic phylogenetic models for inferring ancient divergences in the tree of life and the power of combining such models with large, carefully curated data matrices that include novel prokaryotic metagenomic data.

Our groups at Dalhousie University and our collaborators (DEEM team Orsay) are following in the footsteps of Mike Gray and W. Ford Doolittle who were amongst the first to generate molecular data in support of the endosymbiotic origin of mitochondria (see: Wheat embryo mitochondrial 18S ribosomal RNA: evidence for its prokaryotic nature).

On a personal level, I was inspired to work on mitochondrial evolution and eukaryogenesis by Gray and Doolittle who were my mentors at Dalhousie University and by Tom Cavalier-Smith, who sadly passed away last year. I'm deeply grateful to these wonderful scientists! I doubt our paper will be the last word on mitochondrial origins, but I do hope that we've made an important contribution in this exciting field!

Paper: Site-and-branch-heterogeneous analyses of an expanded dataset favour mitochondria as sister to known Alphaproteobacteria.