DNA methylation is an indispensible, tissue-specific epigenetic mark occurring predominantly at CG dinucleotides (CpGs) in mammalian genomes. It refers to the post-replication maintenance or de novo
addition of a methyl group to the carbon-5 position of the cytosine pyrimidine ring by DNA methyltransferases to form 5-methylcytosine (5mC). DNA methylation has important consequences for gene expression by contributing to the remodelling of chromatin via recruitment of methyl-CpG-binding domain (MBD) protein complexes and subsequent chromatin modifiers. Disease phenotypes have been shown to arise when these activities are perturbed, resulting in undesired gene expression or silencing; this is most notable in carcinogenesis (for reviews, see [1,2]
Although DNA methylation was discovered 60 years ago – and has been shown to be involved in human disease some 25 years ago – efficient technologies for genome-wide methylation analysis have only become available fairly recently with the introduction of the microarray and, more recently, second-generation sequencing, exemplified by Solexa (Illumina), SOLiD (Applied Biosystems) and PSQ (Roche) chemistries. To date, most studies have focussed on DNA methylation at gene promoters and CpG islands, as methylation at these regions defines cellular identity and aberrant methylation here is likely to contribute rare but significant changes to the phenotype, however; recent evidence suggests that regions most likely to undergo change in DNA methylation may occur at loci outside gene promoters and CpG islands – so called CpG island shores 
. Changes here are thought to be more frequent but also more subtle, thus contributing synergistically to disease phenotypes.
From a methodological point of view, assaying these putatively functional regions presents two issues. Firstly, analyses need to encompass the whole genome: there are >28 million CpG sites scattered throughout the haploid human genome (the methylome) and it remains technically challenging to capture the majority of these (for an overview of these methodologies, see [4,5]
). Secondly, these techniques need to be scalable for use with large cohorts to maximise statistical power to elucidate reliable differentially methylated regions (DMRs).
Second-generation sequencing has eased the former burden; for example, Solexa sequencing coupled with bisulfite conversion (MethylC-seq) recently provided the first human DNA methylome at single CpG resolution 
. At the current cost however, MethylC-seq is precluded from routine use on large cohorts and leaves the latter burden unresolved.
One approach that strikes a good balance between coverage, resolution, throughput and cost is to combine second-generation sequencing with the methylated DNA immunoprecipitation (MeDIP) assay 
. MeDIP is an enrichment-based technique 
that uses an antibody against 5mC to capture the methylated fraction of a genome suitable for array-based analysis (MeDIP-chip; [9–11]
) or sequencing-based analysis (MeDIP-seq; 
). The main advantage of MeDIP is that genomic capture is relatively unbiased and not limited to restriction sites or CpG islands. At ~100 bp resolution, MeDIP-seq offers comparable coverage at much lower cost than MethylC-seq. However, throughput is currently limited due to the manual procedure of the MeDIP step, an issue we address here.
Traditionally, issues with throughput have been alleviated using automation – an approach used here. The recent development of the SX-8G IP-Star by Diagenode opens the way to perform up to 16 automated MeDIP (AutoMeDIP) or ChIP (AutoChIP) assays per run that, with some considerations, are compatible with subsequent second-generation sequencing. In this study, we focus on AutoMeDIP and describe a quality control (QC) procedure using in vitro methylated lambda (λ) DNA sequences to demonstrate reliability, sensitivity and specificity of the method. In addition, we provide practical considerations for combining AutoMeDIP with Illumina Genome Analyzer II (GAII) library preparation for subsequent sequencing (AutoMeDIP-seq). We believe AutoMeDIP-seq offers a competitive approach to high-throughput whole genome methylation (methylome) analysis of medium to large cohorts.