Table 1 Summary of common methodologies for methylation analysis

Direct bisulfite sequencing

Sanger sequencing of bisulfite-modified DNA sequences

Allows semi-quantitative to quantitative detection of DNA methylation as an average for each CpG position. Possible to sequence longer sequence lengths compared with pyrosequencing.

Difficulty in obtaining clean reads at start of sequence. Heterogeneous methylation associated with poor peak quality in four-dye electropherograms and underestimation of total DNA methylation.

10%-20% variant base pairs

Frommer et al. [80] (1992), Clark et al. [81] (1994)

Bisulfite pyrosequencing

Sequencing by synthesis technique that allows quantification of methylation at individual CpG positions

Quantitative method. Clean reads at beginning ofsequence.

Able to sequence relatively short sequence lengths of about 80 bp. Can identify but not quantify heterogeneous methylation.

10% at each CpG position

Colella et al. [82] (2003), Tost et al. [83] (2003)

Mass spectrometry

Bisulfite-modified DNA amplified using methylation-independent primers followed by base-specific cleavage of nucleic acids. Methylated and unmethylated fragments differ in mass and separated by mass spectrometry.

Detects both methylated and unmethylated sequences. Quantitative method which gives an average reading for each CpG site or region. Possible to analyze longer sequence lengths compared with pyrosequencing.

Interrogation of individual CpG sites not always possible with fragments which contain several CpG sites.

5% at each CpG position

Ehrich et al. [84] (2005), Coolen et al. [85] (2007)

Methylation-specific PCR (MSP)

Methylation-specific primers amplify methylated bisulfite-modified DNA

Very high sensitivity

Detects methylated sequences only. False- positive results may occur from poor primer design, amplification of a minor methylated subpopulation, and from incomplete bisulfite modification. Non-quantitative.

0.01%

Herman et al. [86] (1996)

MethylLight

MSP combined with Taqman probe to allow quantification of amplification in real time

Allows quantification of DNA methylation in homogeneously methylated samples. Reduced false positives due to incomplete bisulfite conversion compared with MSP.

Detects methylated sequences only. Reduced sensitivity with heterogeneous methylation and only semi-quantitative at best in context of heterogeneous methylation.

<0.01%

Eads et al. [87] (2000)

Sensitive melting analysis after real-time methylation-specific PCR (SMART-MSP)

Methylation-specific primers amplify methylated bisulfite-modified DNA with quantification of amplification by use of fluorescent dye. PCR amplification followed by melt step which allows detection of false-positive results.

Quantitative for homogenous methylation. Melt step allows detection of false positives.

Detects methylated sequences only. Cannot quantify heterogeneous methylation.

0.10%

Kristensen et al. [88] (2008)

Methylation-sensitive high-resolution melting (MS-HRM)

Methylation-independent primers amplify bisulfite-modified DNA sequences. Methylated and unmethylated sequences differentiated based on differing melting profiles. Process can be monitored in real time by use of a fully saturating double-stranded DNA-binding dye and can be semi-quantitative by comparing the melting profile of the sample with controls of known methylation levels.

Semi-quantitative. Detects both methylated and unmethylated sequences.

Can detect presence of, but not quantify, heterogeneous methylation

0.1%-1.0%

Wojdacz and Dobrovic [89] (2007)

Methylation-sensitive single-strand conformation analysis (MS-SSCA)

Bisulfite-modified DNA amplified by using methylation-independent primers. Methylated and unmethylated amplicons form different conformers and separated by electrophoresis.

Detects both methylated and unmethylated sequences

Non-quantitative method. False positives and negatives can occur from co-migration of conformers.

5%

Bianco et al. [90] (1999)

Methylation-sensitive single-nucleotide primer extension (MS-SNuPE)

Nucleotide incorporated to extend primer placed immediately adjacent to C of CpG used to calculate average methylation at a given CpG position.

Quantitative method. Detects both methylated and unmethylated sequences.

Investigates only one CpG site with each primer. Limited ability to place primers in regions of high CpG density.

Gonzalgo and Jones [91] (1997), Gonzalgo and Jones [92] (2002)

Combined bisulfite restriction analysis (COBRA)

Amplified bisulfite-modified DNA is digested with restriction endonucleases and the fragments separated by gel electrophoresis.

Detects both methylated and unmethylated sequences

Analysis possible for up to 2 CpG sites only due to each restriction endonuclease having limited number of cutting sites. Reduced cutting efficiency of restriction endonuclease leads to underestimation of DNA methylation level. Heterogeneous methylation results in underestimation of methylation levels due to formation of heterodimers.

1%

Xiong and Laird [93] (1997)

Methylation-specific multiplex ligation-dependent probe amplification (MS-MLPA)

Probe hybridized to denature DNA. Probe-DNA complex simultaneously ligated and digested by methylation-specific endonucleases. Complexes with a methylated target CpG site will not be digested, resulting in an amplification product. Complexes with an unmethylated target CpG site are digested and no amplification product results.

Semi-quantitative technique able to analyze multiple sites simultaneously. Avoids bisulfite modification, and can be used for single-stranded, short (50-60 bp) DNA sequences.

Methylation analysis restricted to methylation-sensitive restriction sites and dependent on enzyme efficiency. Fixation can reduce enzyme cleavage efficiency resulting in false positives.

Nygren et al. [94] (2005)

Digital techniques

Techniques which use limiting dilution to allow analysis of single template epialleles.

Avoids potential PCR amplification bias. Accurate quantification of heterogeneous methylation.

Requires appropriate instrumentation

Candiloro et al. [95] (2008), Candiloro and Dobrovic [96] (2009), Li et al. [97] (2009), Mikeska et al. [30]