| DNA methylation is an important biological signal in prokaryotic genomes, responsible for regulating many critical processes, including innate immunity, mismatch repair, replication, and virulence. Recent developments in third generation sequencing technologies have enabled comprehensive studies of the most prevalent types of methylation in prokaryotic genomes, but existing detection methodologies suffer from limitations that prevent the high-resolution study of methylation dynamics. This thesis describes efforts to develop single molecule level methods for DNA methylation detection in prokaryotic genomes. I first describe a framework called SMALR (single molecule modification analysis of long reads) comprised of two methods for high-resolution methylation detection. These methods can assess methylation status on single molecules, both at the level of individual nucleotides and at all phased instances of methylation motifs. I demonstrate how these two methods can be used to accurately quantify the methylated fraction of motif sites in a sample, track the progression of the DNA replication fork during a round of cell division, and detect the ON/OFF activity switching of phase-variable methyltransferases (MTases). Driven by stochastic frameshift mutations, the activity switching of the observed phase-variable MTases can induce dramatic shifts in the program of gene transcription.;The concepts underlying SMALR do not require sample homogeneity and are, as a result, applicable beyond single cultured isolates. The single molecule level detection procedure was therefore modified for metagenomic applications in the form of a novel tool, mBin. mBin can not only identifying methylated motifs, but also treat the methylated motifs as discriminatory features for binning individual genomes and linking mobile genetic elements (MGEs) to their host organism. After verifying the performance of mBin on multiple synthetic communities of known composition, I apply it to characterize a complex adult mouse gut microbiome. By identifying methylated motifs in the metagenomes, I demonstrate how the unique methylation profiles uncovered by mBin in the mouse gut can be used to bin seven Bacteroidales genomes in the mouse gut that cannot be segregated using existing binning tools. The methods described in this thesis can provide unique epigenomic insights ultimately leading toward greater understanding of the mechanisms at work in microbial communities. |
