The genomics of symbiosis: hosts keep the baby and the bath water.

L ife as we know it is the result of symbiosis. Mitochondria, the energy generators of our own bodies, are the end result of the symbiosis of a bacterium with a host eukaryote. Visible around us are the land plants that we eat to fuel our mitochondria. Land plants are the beneficiaries of a second symbiosis, that of a cyanobacterium with a mitochondriate host, a relationship that subsequently developed into the light-harvesting chloroplast. Perhaps because of the ubiquitousness of these two symbioses and their smooth functioning, research has focused on the process by which the bacterial symbiont was reduced and tamed. The chloroplast can no longer grow without its host, and the chloroplast’s genetic material is reduced dramatically from that of the original cyanobacterium. Amazingly some of the chloroplast’s own genes were not simply lost from its genome but moved to the host nuclear genome. The evolutionary introduction of a transit peptide sequence into these genes resulted originally in cyanobacterial gene products ending up back in the chloroplast, carrying out their original function but under host control. In this issue of PNAS, Martin et al. (1) argue forcefully, however, that this focus on the reduction of the land-plant chloroplast genome has ignored the other bonanza from this symbiosis: the large number of genes from the original cyanobacterium that have ended up in the host nuclear genome without a role in chloroplast maintenance. These genes provided the raw genetic material for plant diversification and competition, possibly against the original cyanobacterium itself. Some early but largely ignored insights into the role of bacteria in chloroplast evolution can be found in the literature. Schimper in 1885 (2) suggested that chloroplasts were derived from symbiotic microorganisms, and Mereschkowsky in 1905 argued more extensively for the development of different types of chloroplasts from different types of cyanobacteria (3, 4). Much later, the role of bacteria in symbiogenesis was championed by Margulis (5). Many scientists have provided the crucial data that convincingly show the close relationship between chloroplasts and cyanobacteria including similar ribosomes, RNA polymerases, and other cellular machinery (briefly reviewed in ref. 1). Current research suggests that the cyanobacterial eukaryote symbiosis occurred only once but diverged rapidly to three major lineages: the greens, the reds, and the rather odd lineage called glaucocystophytes (reviewed in ref. 6). The greens are the green algal land-plant group. The reds are the red algae, now producing much of the agarose consumed by molecular biology labs. The chloroplasts in the many algal groups show diverse pigments, morphology, and storage products. Gibbs (7, 8) was the first to suggest that the origin of algal chloroplasts did not end with the original cyanobacterium and eukaryote symbiosis. Like a crazed alchemist trying improbable combinations of elements, there are ongoing evolutionary experiments in which the original red and green eukaryotic algal lineages are being tested as symbionts inside new eukaryotic hosts. These symbioses have been referred to as secondary and tertiary symbioses. Thus far, these experiments have been spectacularly successful. The diatoms are a group of small algae in lakes and oceans, the chloroplasts of which evolved from a secondary or tertiary symbiosis involving a red alga (9); they are nearly as important as land plants in the global carbon cycle. A relatively abundant group called the cryptophytes have a chloroplast derived from a red alga (10). One characteristic of bacterial symbiosis (and in some cases pathogenesis) is the reduction of the genome size of the symbiont pathogen. This reduction is seen in such diverse examples as Mycoplasma genitalium (641 kb), an intracellular parasite of epithelial cells (11), and Buchnera, the intracellular bacterial symbiont of aphids (450 kb; ref. 12). Smaller still, chloroplast genomes are 120–190 kb. Work by Martin et al. (1) and work summarized previously (9) show that chloroplast gene number varies from 58 to 200. The total number of different genes found to date in chloroplasts is 210, but only 45 of these are found in all chloroplasts from diverse lineages. Cyanobacterial genomes, in contrast, can range from 1,694 (Prochlorococcus) to 7,281 (Nostoc) predicted genes (www.jgi.doe.gov JGI microbial html index.html). Clearly the genome of the original cyanobacterial symbiont was reduced drastically to arrive at the current chloroplast genome. This process seems to be continuing (13). Using the complete genome of Arabidopsis for the first time (24,990 predicted genes) and comparing it to the genomes of 3 cyanobacteria, 16 other prokaryotes, and yeast, Martin et al. (1) were able to define a subset of 9,368 genes that could be analyzed phylogenetically. Of these, they detected 1,700 genes of likely cyanobacterial origin, which is surprisingly close to the size of some recently sequenced, possibly minimal cyanobacterial genomes such as Prochlorococcus (www.jgi.doe.gov JGI microbial html index.html). The most controversial part of their paper is perhaps their subsequent extrapolation using this percentage (1,700 9,368 18%) to calculate the fraction of the highly divergent remainder of the Arabidopsis genome (15,622 genes) that is also derived from cyanobacteria. They thus estimate that a total of 4,500 genes were derived from the original symbiont. These include genes homologous to ones in the original cyanobacterium and genes derived by processes such as gene duplication, gene shuff ling, and rapid divergence. Because