Chemoenzymatic labeling of DNA methylation patterns for single-molecule epigenetic mapping

Abstract

DNA methylation, specifically, methylation of cytosine (C) nucleotides at the 5-carbon position (5-mC), is the most studied and among the most significant epigenetic modifications. Here we developed a chemoenzymatic procedure to fluorescently label non-methylated cytosines in the CpG context allowing epigenetic profiling of single DNA molecules spanning hundreds of thousands of base pairs. For this method, a CpG methyltransferase was used to transfer an azide to cytosines from a synthetic S-adenosyl-l-methionine cofactor analog. A fluorophore was then clicked onto the DNA, reporting on the amount and position of non-methylated CpGs. We found that labeling efficiency was increased two-fold by the addition of a nucleosidase that degrades the inactive by-product of the azide-cofactor after labeling, and prevents its inhibitory effect. We first used the method to determine the decline in global DNA methylation in chronic lymphocytic leukemia patients and then performed whole genome methylation mapping of the model plant Arabidopsis thaliana. Our genome maps show high concordance with published methylation maps produced by bisulfite sequencing. Although mapping resolution is limited by optical detection to 500-1000 base pairs, the labeled DNA molecules produced by this approach are hundreds of thousands of base pairs long, allowing access to long repetitive and structurally variable genomic regions.

Long reads capture simultaneous enhancer–promoter methylation status for cell-type deconvolution

“…These results constitute a first step toward harnessing enhancer–promoter linked methylation for deconvolution of cell populations with highly similar cell-type methylomes. Despite focusing on Bionano Genomics’ optical methylation mapping ( Gabrieli et al , 2021 ; Sharim et al , 2019 ), which currently provides the highest coverage of long reads, the principles are valid to other future datasets such as those produced by Oxford Nanopore ultralong-read sequencing protocol. Further exploration of these linkages, including the joint effects of multiple enhancers per promoter may shed light on insightful cellular transformations regulated by long-range epigenetic interactions.…”

DeepOM: Single-molecule optical genome mapping via deep learning

“…The accuracy of OGM can be defined as the expected fraction of imaged molecules that are aligned with high confidence to the reference genome. This accuracy is extremely important for applications where target DNA quantity in the sample is limited, such as cultivation-free pathogen identification [7], or where maximal coverage of the genome is required per mapping experiment, such as in rare variant detection [14] or epigenetic mapping [15,16]. The current computational approaches are limited in accuracy, since they are unable to extract all the available information from the image of the DNA molecule.…”