A molecular atlas of plastid and mitochondrial evolution from algae to angiosperms

Abstract

Algae and plants carry two organelles of endosymbiotic origin that have been co-evolving in their host cells for more than a billion years. The biology of plastids and mitochondria can differ significantly across major lineages and organelle adaptation likely accompanied the conquering of new ecological niches including land. Based on organelle proteome data, and the genomes of 168 phototrophic (Archaeplastida) versus a broad range of 518 non-phototrophic eukaryotes, we screened for changes in plastid and mitochondrial biology across one billion years of evolution. Taking into account 331,571 protein families (or orthogroups), we identify 31,625 protein families that are unique to primary plastid-bearing eukaryotes. 1961 and 846 protein families are predicted to operate in plastids and mitochondria, respectively. Tracing the evolutionary history of these protein families through evolutionary time uncovers the significant remodelling the organelles experienced from algae to land plants. The analyses of gained orthogroups identifies molecular adaptations of organelle biology that connect to the diversification of major lineages and facilitated major transitions from chlorophytes en route to the global greening and origin of angiosperms.

Performance of localization prediction algorithms decreases rapidly with the evolutionary distance to the training set increasing

“…The endosymbiotic organelles of algae and plants have been co-evolving for over a billion-years and their proteomes continue to change and adapt 69,71,72 . During plant terrestrialization for instance, the plastid proteome of the algal ancestor expanded from a few hundred to that of the angiosperm plastid housing about 1500 proteins 69 .…”

“…With the growing notion of organelle 'pan-proteomes', i.e. organelle proteins present in selected species or organelle sub-types 69,70,72,[74][75][76][77][78] , our analysis shows that algorithms are inadequate at capturing this pan-proteome or even the distant homologues of conserved proteins. Training the future algorithms on these missed proteins from across (Fig.…”

“…To expand benchmarking across a larger evolutionary scale, we used TargetP (the best performing among the three) to predict organelle proteomes for 173 photosynthetic eukaryotes, for which genomes are available. Since organelle proteomes are scarce, we utilised orthology inferred organelle proteomes based on their sequence similarity with experimentally validated organelle proteomes 69 . Around 70-80% of predicted plastid proteins in eudicots could be validated by orthology based predictions (Fig.…”

“…The endosymbiotic organelles of algae and plants have been co-evolving for over a billion-years and their proteomes continue to change and adapt 69,71,72 . During plant terrestrialization for instance, the plastid proteome of the algal ancestor expanded from a few hundred to that of the angiosperm plastid housing about 1500 proteins 69 . The algorithms predict there to be 1000 to 2000 plastid (and mitochondrial) organellar proteins even outside of angiosperms, 25% or less of which appear to be true positives (Fig.…”

“…We made use of available, experimentally verified plant proteomes of mitochondria and plastids as well as protein clustering to investigate the reliability of these algorithms across species ranging from algae to angiosperms. Our analysis brings forth deficiencies of these algorithms, caused by a combination of their inherent modus operandi , a lack of training on a diverse dataset, the complex biology of the plant cell, and the evolutionary dynamic nature of the plant organelles 69 . Tracing the error sources allows to sketch an approach towards developing better algorithms that are capable of serving the diversity of the plant kingdom.…”