Evolution of genome base composition and genome size in bacteria

In bacteria and archaea, genome size and guanine–cytosine (GC) content are correlated (Bentley and Parkhill, 2004; Musto et al., 2006; Mitchell, 2007; Suzuki et al., 2008; Guo et al., 2009). These parameters show greater correlation in bacteria (Pearson’s correlation coefficient r = 0.46) than in archaea (r = 0.195) (Nishida, 2012a). The GC content in bacteria varies widely from 13.5% in “Candidatus Zinderia insecticola” (McCutcheon and Moran, 2010) to 74.9% in Anaeromyxobacter dehalogenans (Thomas et al., 2008). Although the GC content is similar among closely related bacteria, sometimes, the GC content is similar in phylogenetically distant bacteria. The distribution of GC content in bacterial genomes differs from a Gaussian distribution with multiple peaks. Bacterial chromosome organization is mediated by nucleoid-associated proteins (NAPs) (Wang et al., 2011). The specificity of NAP–DNAbinding is determined by the differences in the GC content in specific regions of the DNA (Lucchini et al., 2006; Navarre et al., 2006; Castang et al., 2008; Smits and Grossman, 2010; Yun et al., 2010; Gordon et al., 2011). For example, the Salmonella NAP specifically binds to DNA regions with low GC content and inhibits expression of the genes present in these regions (Lucchini et al., 2006; Navarre et al., 2006). NAPs vary among bacteria (Ali et al., 2012). In addition, the NAP genes are located in the plasmid as well as in the chromosome, suggesting that these genes have been distributed via plasmids (Takeda et al., 2011). I hypothesize that the GC content distribution may be related to the variation in bacterial NAPs (Nishida, 2012b, 2013). However, the correlation between the genome size and GC content in bacteria is poorly understood. The obligate host-associated bacteria contain short genomes with low GC content (Mira et al., 2002; Moran, 2002; McCutcheon and Moran, 2012). Insertion sequence elements play an important role in the genome reduction of the hostassociated bacteria (Song et al., 2010), whose small population size and asexual mode of reproduction lead to reduction of the genome size. In addition, deletion of the genes involved in DNA repair may contribute to a GC-poor genome (Moran et al., 2008). However, genome size reduction is not limited to the obligate hostassociated bacteria (Nilsson et al., 2005). Generally, bacteria show a bias toward genomic deletions than insertions (Mira et al., 2001). Thus, bacteria must acquire additional genes to adapt to different environments. Some bacterial lineages, for example, Actinobacteria, have maintained long genomes with high GC content. Plasmids (and viruses) have played an important role in additional gene uptake into chromosomes (Davison, 1999; Sørensen et al., 2005; Harrison and Brockhurst, 2012). Occasionally, the plasmid DNA gets integrated into the host chromosomal DNA (Harrison and Brockhurst, 2012). In addition, viral DNA occasionally remains in the chromosome as a prophage. Horizontally transferred DNA, plasmid DNA, and virus DNA have lower GC content than host chromosome DNAs (Rocha and Danchin, 2002). In a previous study, I compared the GC content across 953 pairs of bacterial chromosomes and plasmids. Among the 953 pairs, 746 (78.3%) pairs showed <10% difference in the GC content of the plasmid DNA and the host chromosomal DNA (Nishida, 2012a). Probably, most bacteria are unable to maintain and regulate plasmids that show very different GC content from their own. However, why most bacteria have not acquired DNAs with GC content higher than that of their own chromosome, but have acquired DNAs with lower GC content is not clear. During evolution, DNA base mutations occurred intracellulaly and not in extracellular environments. The DNA polymerase components that are involved in DNA replication directly influence the base composition of the genome (Zhao et al., 2007; Wu et al., 2012). Variations in the bacterial genome DNA sequences cannot be fully explained on the basis of neutralmutations alone (Sueoka, 1988). In bacterial genomes, mutations from GC to adenosine–thymine (AT) are more common than mutations from AT to GC (Lind and Andersson, 2008; Hershberg and Petrov, 2010; Hildebrand et al., 2010; Rocha and Feil, 2010). I hypothesize that plasmids (and viruses) have been generated from the chromosome (Frontiers Research Topics, “Evolution and function of bacterial and archaeal genome sequences”). The host bacterial genome DNA has undergone a series of changes during evolution to become AT rich. On the other hand, the GC content in a plasmid that is in an extracellular environment would not change. Such a plasmid will not be accepted by the original host bacterium because most bacteria appear unable to acquire DNAs with GC content higher than that in their own chromosome. Such plasmids are transferred to bacteria that contain genomes with GC content higher than that in the plasmid (Figure 1). This natural system may be effective for obtaining useful genetic information (DNA fragments) from phylogenetically distant bacteria. I propose that the genome size and GC content