The evolution of the mitochondrial proteome in animals
Carolyn J. Lawrence-Dill
Mitochondria are subcellular organelles in eukaryotes which possess their own genome. While they are most well-known for their role in energy metabolism via oxidative phosphorylation, research has shown that mitochondria are involved in diverse critical cellular functions like Fe/S cluster biosynthesis, apoptosis, signaling, etc. In mammals, over 1,500 proteins carry out these functions in the mitochondria. A small portion of these proteins ( ~ 1%) is contributed by the mitochondrial genome, whereas the vast majority (~ 99%) are encoded in the nuclear genome and transported into the organelle. This set of nuclear-encoded mitochondrial proteins is defined as the "mitochondrial proteome". The primary objective of my research is to analyze the evolution of the mitochondrial proteome in animals, and to develop tools for facilitating the comparative analysis of animal mitochondrial proteomes.
For obtaining a broad picture of animal mitochondrial proteome evolution, it is necessary to examine the mitochondrial proteomes of both bilaterian and non-bilaterian animals. All experimentally-characterized mitochondrial proteomes in animals are from Bilateria. This is unfortunate, since the comparative analysis of animal mitochondrial genomes has shown that most of the mitochondrial genomic diversity in animals can be found in the four phyla of non-bilaterian animals (Porifera, Cnidaria, Ctenophora, and Placozoa). In this dissertation, we carry out the first comparative analysis of mitochondrial proteomes from non-bilaterian animals.
We use bioinformatic techniques to predict the mitochondrial proteomes in the four phyla of non-bilaterian animals. We detect a large variation in the size and content of the inferred mitochondrial proteomes of non-bilaterian animals. The size of the inferred mitochondrial proteomes ranges from 454 proteins in Kudoa iwatai to 2,119 proteins in Leucosolenia complicata. We find that much of the variation in the size of the mitochondrial proteomes in non-bilaterian animals is due to the number of proteins with a mitochondrial targeting signal, but no ortholog to any human or yeast protein. Additionally, we also identify several instances of mitochondrial neolocalization in the non-bilaterian mitochondrial proteomes. Conversely, ~ 2.5% of the human mitochondrial proteome has no ortholog in any non-bilaterian species, representing potential bilaterian mitochondrial innovations. Next, through a comparative analysis of the experimentally-characterized mitochondrial proteomes of bilaterian animals, we investigate the causes and functional implications of the variation in size and content of the animal mitochondrial proteomes. We find that the animal mitochondrial proteome is a dynamic entity, with a small core of mitochondrial proteins that are conserved in all four animals, and a large number of lineage-specific gains and losses. Of the several factors responsible for the size-variation in the four animal mitochondrial proteomes, we find that the gain of novel mitochondrial proteins in mammals and loss of conserved mitochondrial proteins in the two ecdysozoans are the main contributors. Interestingly, while nearly one-fifth of each animal mitochondrial proteome consists of proteins that underwent mitochondrial neolocalization in animals, the majority of these neolocalized proteins lack a canonical mitochondrial targeting signal.
While carrying out comparative analysis of mitochondrial proteomes in animals, researchers encounter two main challenges: 1) data on experimentally-characterized animal mitochondrial proteomes is scattered across several databases, and 2) most animal phyla lack a species with an experimentally-characterized mitochondrial proteome. To address these challenges, we develop two tools to facilitate the comparative analysis of mitochondrial proteomes in animals- 1) the Metazoan Mitochondrial Proteome database, which consolidates data on animal mitochondrial proteomes from various sources, and 2) MitoPredictor, a novel machine-learning tool to predict mitochondrial proteins in animals, using three sources of information: orthology, mitochondrial-targeting signal prediction and protein-domain information.