From Genes to Ecosystems: Resource Availability and DNA Methylation Drive the Diversity and Abundance of Restriction Modification Systems in Prokaryotes

Together, prokaryotic hosts and their viruses numerically dominate the planet and are engaged in an eternal struggle of hosts evading viral predation and viruses overcoming defensive mechanisms employed by their hosts. Prokaryotic hosts have been found to carry several viral defense systems in recent years with Restriction Modification systems (RMs) were the first discovered in the 1950s. While we have biochemically elucidated many of these systems in the last 70 years, we still struggle to understand what drives their gain and loss in prokaryotic genomes. In this work, we take a computational approach to understand the underlying evolutionary drivers of RMs by assessing ‘big data’ signals of RMs in prokaryotic genomes and incorporating molecular data in trait-based mathematical models. Focusing on the Cyanobacteria, we found a large discrepancy in the frequency of RMs per genome in different environmental contexts, where Cyanobacteria that live in oligotrophic nutrient conditions have few to no RMs and those in nutrient-rich conditions consistently have many RMs. While our models agree with the observation that increased nutrient inputs make the selective pressure of RMs more intense, they were unable to reconcile the high numbers of RMs per genome with their potent defensive properties- a situation of apparent overkill. By incorporating viral methylation, an unavoidable effect of RMs, we were able to explain how organisms could carry over 15 RMs. With this discovery, we then tried and reassess the distribution of methyltransferases, an essential component of RMs that can also have alternate physiological rolls in the cell. We expand on conventional wisdom, that methyltransferases that are widely phylogenetically conserved are associated with global cellular regulation. However, we also find that organisms with high numbers of RMs also have a surprising amount of conservation in the methyltransferases that they carry. This data suggests caution should be used in associating phylogenic signals with functional rolls in methyltransferases as different functional rolls seem to overlap in their phylogenetic signal. Indeed, we suggest trait-based modeling may be the best tool in elucidating why organisms with a high selective pressure to maintain RMs appear to have conserved methyltransferase.