Second genesis of a plastid organelle.

O ne of the more remarkable discoveries of 20th century biology is that all eukaryotes are chimeras of two or more different cells (1). The tree of life does not just bifurcate through successive speciation events; its branches also graft, one onto another, merging lineages to produce novel fruit. Our mitochondria, and the photosynthetic plastids of plants and algae, are proof of the success of these new combinations. Both organelles were acquired from separate bacterial lineages, where a useful bacterium was internalized and maintained within the host cell— a process known as endosymbiosis. These organelles are no mere hitchhikers. They have been elaborately integrated with their hosts, so intimately that even their genes have been relinquished to the host cell to allow these partnerships to truly operate as one. To achieve this, the host cell learned to take responsibility for expressing the organelle genes and delivering the protein products back into the organelle according to its requirements. Our understanding of how these seminal achievements of organellogenesis occurred, however, is obscured by their antiquity, both organelles arising ∼1+ billion years ago (2). It is akin to studying modern jet aircraft in the hope of reconstructing the Wright brothers’ first daring attempts at flight. Iterative leaps of technology often mask the formative innovations. However, now we have a chance to revisit this process, as Nowack and Grossman in PNAS (3) show that the new photosynthetic endosymbiont of Paulinella chromatophora has also commenced this journey of genetic integration. Plastids show remarkable diversity across plants and the variously pigmented algae, but it is now generally agreed that they all stem from a single endosymbiotic gain of a β-cyanobacterium (4, 5). This single origin of plastids implies that establishing a novel organelle from a bacterium is a challenging undertaking and not frequently repeated, despite the obvious benefits of acquiring photosynthesis. To understand possible bottlenecks in this process it is useful to consider the sequence of events required to establish a genetically integrated organelle (Fig. 1). Any endosymbiotic relationship is initially cemented by establishment of some mutually beneficial metabolic sharing and coordinated division of the endosymbiont to ensure its persistence in the host cell. A great variety of endosymbiotic relationships have achieved these initial tasks, implying that these partnerships are not so difficult (6). For genetic integration, the first step is the transfer of copies of genes from the endosymbiont to the host nucleus (Fig. 1B). Whereas such meddling in the genomic integrity of the nucleus and endosymbiont might seem unlikely, we know that it happens a lot. Both experimental and anecdotal evidence tells us that chunks of organelle DNA are frequently transferred to the nucleus (7). The next step is to express the nuclear copy of the endosymbiont gene, to start synthesizing this endosymbiont protein on cytosolic ribosomes (Fig. 1C). This step might require only the “dumb luck” of the gene integrating within an appropriate nuclear sequence to promote expression. Repetitive gene transfers might be enough to ensure that this integration is eventually achieved. However, now the cell faces the further challenge of transferring the protein across the membrane barriers surrounding the endosymbiont, back to its functional location. How proteins first achieve translocation across the endosymbiont membranes is an open question (Fig. 1C) (8, 9). They might pass through existing pore-forming complexes, already in place to handle some other cargo (10). Or they might be delivered by membrane vesicle, if by chance they are routed into the host cell endomembrane system (11). In either case (or a combination of these), only once they are transported into the endosymbiont in sufficient quantity to provide adequate function can the endosymbiont-encoded copy of the gene finally be lost. If regulatory control of this protein is necessary for function, then the nucleus-encoded gene too must achieve this. Loss of the endosymbiont gene copy is the final act in genetic integration of the endosymbiont with its host, and to many this act is the threshold for achieving organelle status (12). However, if early trafficking of proteins initially used existing transport machinery designed for other molecules, it probably would not have been very efficient or specific. Refinement of a dedicated route for organelle proteins would likely be required. This would have the further advantage that proteins of other transferred organelle genes could adopt this route (or routes) and obviate the need to reinvent targeting with every transfer (Fig. 1D). In modern plastids and mitochondria, the majority of organelle proteins bear a characteristic peptide extension that acts as a targeting signal A