Experimental and Computational Approaches for Non-CpG Methylation Analysis

Abstract

Cytosine methylation adjacent to adenine, thymine, and cytosine residues but not guanine of the DNA is distinctively known as non-CpG methylation. This CA/CT/CC methylation accounts for 15% of the total cytosine methylation and varies among different cell and tissue types. The abundance of CpG methylation has largely concealed the role of non-CpG methylation. Limitations in the early detection methods could not distinguish CpG methylation from non-CpG methylation. Recent advancements in enrichment strategies and high throughput sequencing technologies have enabled the detection of non-CpG methylation. This review discusses the advanced experimental and computational approaches to detect and describe the genomic distribution and function of non-CpG methylation. We present different approaches such as enzyme-based and antibody-based enrichment, which, when coupled, can also improve the sensitivity and specificity of non-CpG detection. We also describe the current bioinformatics pipelines and their specific application in computing and visualizing the imbalance of CpG and non-CpG methylation. Enrichment modes and the computational suites need to be further developed to ease the challenges of understanding the functional role of non-CpG methylation.

Keywords: computational approaches; mCpH; methods; non-CpG methylation; techniques.

Figure 1

Enrichment strategies that are involved in non-CpG methylation detection and analysis. (A) Enzyme-based enrichment techniques: Enzyme-based enrichment approaches are classified based on the methylation-sensitive and -insensitive restriction enzymes. The targeted endonuclease activity eliminates CpG methylation and enriches non-CpG methylation that can be deduced by other methods such as semi-quantitative PCR, luminescence, quantitative PCR, and also read through Sanger sequencing. (B) Antibody-based enrichment techniques: Antibody-based enrichment strategies exploit the use of 5-methylcytidine antibodies that are specific to 5-mC to capture the methylated cytosines irrespective of their adjacent moiety. The captured methylated cytosines have both CpG and non-CpG methylation. Algorithmic filtering should be employed to the reads that are obtained from sequencing to differentiate non-CpG methylation from CpG methylation. Also, methods such as semi-quantitative PCR, luminescence, quantitative PCR, and Sanger sequencing can be used to detect non-CpG methylation post-enrichment for global scale non-CpG detection. (C) Single-molecule real-time sequencing: The recent development of single molecule real-time (SMRT) sequencing detects non-CpG methylation at single-base resolution without any enrichment techniques. Coupling SMRT with an algorithm that is specific for non-CpG methylation helps in detecting the CA/CT/CC methylation from CG methylation.

Figure 2

Computational approaches that are involved in non-CpG methylation analysis. DNA methylation analysis uses a computational pipeline to distinguish between non-CpG methylation and CpG methylation. There are three major bioconductor packages such as Methyl Kit, Methyl Pipe, and DMR caller that could identify non-CpG methylation with different types of factors such as input sequence (SAM/txt/tab-limited), different aligners (BS-Specific/general), and various base-calling strategies. All of these factors help in distinguishing non-CpG methylation with multiple advantages.