Scrambled and not-so-tiny genomes of fungal endosymbionts.

Symbiosis, defined as prolonged and close association of two or more different organisms, is a major driving force for expansion of biological diversity from genes to ecosystems. In the most intimate form, the symbiont lives inside the cells of its host, is heritable, and the two parties could no longer survive independently. Such heritable endosymbiosis blurs the line of what constitutes an individual organism and is considered the origin of organelles in present-day eukaryotic cells (1). Over the past decade, our understanding of the mutualistic endosymbionts of insects has been greatly expanded through examination of their genomes (2–4). It seemed that once a stable obligate symbiosis is developed, the endosymbiont genomes would undergo extreme levels of reduction and become highly static. These observations parallel those found in animal mitochondria (5) and angiosperm chloroplasts (6), which begs the question of how universal these evolutionary trends are among heritable endosymbionts. In PNAS two companion articles (7, 8) provide a first glimpse of the endosymbiont genomes from a group of ancient and diverse fungi. The fungal hosts discussed in these papers, arbuscular mycorrhizal fungi (AMF) belonging to the phylum Glomeromycota (9) (Fig. 1), are themselves obligate symbionts of land plants. In their mutualistic relationships these root-associated fungi provide the plants with mineral nutrients and water and in return receive carbohydrates derived from photosynthesis. These fungal–plant associations could be traced back to >400 Ma and are thought to be pivotal in the origin of the land plants and the functioning of terrestrial ecosystems (10). Most AMF harbor heritable endosymbionts named Mollicutes-related endobacteria (MRE) (11), which have unknown biological roles. To infer the functions of MRE, the authors of these two studies (7, 8) performed genome sequencing of MRE from four diverse AMF species, including Dentiscutata heterogama (Dh), Racocetra verrucosa (Rv), Claroideoglomus etunicatum (Ce), and Rhizophagus clarus (Rc) (Fig. 1 and Table 1). MRE are related to Mycoplasma in the class Mollicutes (11), which is a group of wall-less bacteria that are mostly animal pathogens. These four MRE genomes have a size range of ∼663–1,228 kilobase pairs (kbp) and a GC content of ∼32–34%, which are comparable to those found among horizontally transmittedMycoplasma (Table 1). The extreme genome reduction (i.e., down to the size range of ∼112–641 kbp), such as that observed in the heritable endosymbionts of insects (2–4), has not occurred in these MRE despite their long evolutionary history of being heritable endosymbionts (>400 Ma) in comparison with those insect endosymbionts (<300 Ma) (12). Other than not having experienced extreme genome reduction, these MRE are different from most insect endosymbionts by possessing highly dynamic genomes. For the MRE associated with Rv and Rc AMF, a single phylotype was found based on the 16S rRNA gene sequences. However, multiple recombinant types of chromosomal contigs were found, which prevents the assembly of these contigs into a single sequence representing the chromosome of each species. This intraspecific genome heterogeneity, in which strains share identical 16S rRNA but have diverged in their gene content, have been reported in a few recent studies on insect symbionts (13–15). These findings challenge the common practice of using 16S rRNA sequences for defining operational taxonomic units and also our definition of species in bacteria. The case of Ce MRE, for which multiple 16S phylotypes were found within individual fungal spores, seems to support the interpretation of this genome heterogeneity as a form of sympatric speciation (14). Curiously, the Dh MRE (7) differs from the other three MRE (8) in lacking intraspecific genome heterogeneity (such that the genome is well-assembled) and having a much lower gene density (Table 1). The explanations for these differences remain to be investigated. Despite these differences, both studies agreed on several key findings. First, ∼3–5% of the protein-coding sequences (CDS) in these MRE genomes seem to have been acquired from their fungal hosts. Intriguingly, most of these genes contain protein domains involved in eukaryotic cell processes, such as leucine-rich repeats (LRR; protein–protein interactions), heterokaryon incompatibility (HET; nonself recognition and programmed cell death in fungi), and sentrin/small ubiquitin-like modifier (SUMO; eukaryotic posttranslational modification). It is conceivable that these genes are likely involved in the MRE–AMF interactions and contributed to the evolutionary success of MRE. Second, Diversispora Redeckera Acaulospora