Background

DNA methylation is an important mechanism for epigenetic regulation of gene expression (Greenberg and Bourc’his, 2019). It has a wide range of effects on genes via several biological processes. DNA methylation is usually detected and analyzed using bisulfite sequencing short reads. However, it is difficult to align these short reads (~150 bp) to some chromosomal regions, such as repetitive sequences and structural variants (Goerner-Potvin and Bourque, 2018). Similarly, short reads also constrain detection of chromosome-specific methylation patterns, such as imprinted regions, in polyploid organisms (Akbari et al., 2021). A comprehensive understanding of epigenetic regulation by DNA methylation will therefore require complementary methods.

The bisulfite method distinguishes between unmethylated and methylated cytosine (C vs. mC), by chemically converting unmethylated C to uracil (U) and to thymine (T), by subsequent amplification (Lister et al., 2009). However, since this reaction is carried out under chemically severe conditions, a large proportion of the DNA in the reaction is fragmented and degraded. The genomic regions that undergo this degradation show biased representation (Olova et al., 2018), further limiting the experimental conclusions this method can provide. The read length of the previous method combining bisulfite conversion and long-read sequencing was only ~1.5 kb, even using targeted PCR (Yan et al., 2015). Recently, enzymatic methyl sequencing (EMseq) was developed as an alternative to the bisulfite method of base conversion (Vaisvila et al., 2021). EMseq involves the oxidation of mC by ten-eleven translocation (TET) enzymes to protect them, followed by base conversion of unmethylated C to U, by APOBEC enzymes. Via the amplification process, U is converted to T, as in the bisulfite method. Because this reaction is performed under milder chemical conditions than with the bisulfite method, longer DNA fragments are obtainable. In fact, a previous study showed that DNA fragments over 5 kb long can be obtained using EMseq and target-specific PCR, and that these fragments can be successfully sequenced in a long-read sequencer (Sun et al., 2021).

Nanopore sequencers read nucleic acid sequences by measuring the change in electric current while the nucleic acids are passing through the nanopore. The maximum read length of nanopore sequencing is over 100 kb (Sakamoto et al., 2020). By recognizing specific electrical patterns for modified bases, base modifications can also be detected (Rand et al., 2017; Simpson et al., 2017). However, while the base-reading accuracy of nanopore sequencers is currently up to 90%, this is not quite high enough to accurately infer methylation patterns (Sakamoto et al., 2020). Furthermore, it requires about 500 ng–1 µg of DNA input, reducing its practical utility for rare samples, such as clinical specimens and biopsies. Although several methods combining base-conversion and long-read sequencing have been developed, thus far all have employed gene-specific amplification (Yang et al., 2015; Liu et al., 2020; Sun et al., 2021). A method for whole-genome methylation analysis by this method, and a bioinformatic pipeline to process the sequence data it generates, have not heretofore been developed.

Here, we report a method for whole-genome long-read methylation sequencing, using a relatively small amount of input DNA, for nanopore sequencing of base-converted DNA by APOBEC enzymes (Figure 1) (Sakamoto et al., 2021). Our method, which we designate nanoEM, allows for whole-genome long-read methylation analysis with 10–100 ng of DNA. In addition, we have developed a data analysis pipeline for nanoEM reads by adopting a three-letter alignment approach to long-read alignment. NanoEM is an useful approach for detecting methylation status of structural variants (SVs), repetitive regions, and imprinting regions, which are difficult to analyze using short read sequencing (Sakamoto et al., 2021).

Figure 1. Flow chart of the experimental procedure.