Selection of DNA methylation methods

In general, DNA methylation methods can be divided into three broad categories: global DNA methylation, genome-wide methylation and site-specific methylation. The selection of which method used to analyse methylation is dependent on a range of different factors including the aim of the study, the quantity of DNA, available equipment and the cost involved. This section reviews some of the different methods which can be used to detect DNA methylation to select the methods which will be utilised throughout this thesis.

4.1.1.1 Global DNA methylation methods

Global DNA methylation can be used to gain insight into the impact of an intervention on DNA methylation patterns. Global based methods do not allow the investigation of specific biological pathways; however, altered global DNA methylation has been associated with several disease states including various forms of cancer (Gao et al. 2014; Joyce et al. 2016), rheumatoid arthritis (Liu et al. 2011) and cardiovascular disease (Kim et al. 2010). Aside from disease models, global methylation has also been used to determine the impact of lifestyle interventions including exercise (see section 2.2) and nutrition (see section 2.3).

The gold-standard method for the highly sensitive determination of global methylation is using high-performance liquid chromatography (HPLC; Kuo et al. 1980). Although the gold-standard method, HPLC requires the use of specialist equipment and large quantities of DNA making it an impractical option to be used in the present thesis. Another highly sensitive method to determine global methylation is using liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS; Song et al. 2005). Unlike HPLC, LC-MS/MS only requires a small quantity of DNA to determine methylation; however, the requirement of specialist equipment prevented the use of this method. Unlike HPLC and LC-MS/MS, the use of enzyme-linked immunosorbent assay (ELISA) based DNA methylation detection does not require any specialist equipment to determine global DNA methylation (Kurdyukov and Bullock 2016). Despite the lack of specialist equipment required, So et al. (2014) report ELISA based detection may only be suitable when identifying large variations in global DNA

methylation because of high assay variability; therefore, the use of ELISA for the may lack the sensitivity to determine small changes in methylation which are expected in this thesis.

After ruling out the use of the methods above, the remaining options to determine global DNA methylation are via the methylation of the LINE-1 sequences or LUMA as surrogates of global DNA methylation. LINE-1 sequences comprise ~17% of the human genome and hypomethylation of LINE-1 sequences have previously been associated with cancer (Gao et al. 2014). LINE-1 DNA methylation is strongly correlated with the gold-standard global methylation technique HPLC suggesting it is a good estimate for global DNA methylation (Lisanti et al. 2013). Initially pyrosequencing of LINE-1 elements was selected as the method to determine global methylation; however, the commercially available LINE-1 pyrosequencing failed the validation procedure because of low peak heights (Appendix I).

The lack of a functioning LINE-1 assay meant LUMA was selected as the method to determine global DNA methylation in this thesis. LUMA quantifies the methylation of internal CpG sites within CCGG sequences throughout the genome (Karimi et al. 2006). In total, CCGG sequences account for ~8% of the total CpG sites throughout the human genome (Fazzari and Greally 2004). These CpG sites occur in both repetitive element and protein coding regions (Fazzari and Greally 2004), whereas, LINE-1 sequences are only situated in nongenic regions. To perform LUMA, DNA is digested in two reactions; one incubated with the methylation-sensitive enzyme HpaII and the other with MspI which digests regardless of methylation status. EcoRI is added to each reaction to control for the input quantity of DNA. The digested reactions are pyrosequenced, and the overhangs filled by nucleotides producing light. DNA methylation is then determined by the HpaII / MspI ratio (Karimi et al. 2006).

4.1.1.2 Gene-specific DNA methylation methods

The gene-specific determination of DNA methylation can be divided into two general categories, genome-wide methods and candidate-gene focussed approaches. Genome-wide based detection methods are useful for the determination of differentially methylated regions of the genome at single-base resolution; however,

52 these methods are highly expensive and require the use of specialist equipment which was not available for this thesis. Candidate-gene based approaches allow the determination of methylation at specific CpG sites using PCR to amplify the region of interest; however, DNA needs to be processed (bisulfite converted) before PCR amplification because methylation is not maintained during PCR.

4.1.1.2.1 Bisulfite conversion of DNA for methylation analysis

The bisulfite conversion of DNA involves the chemical modification of cytosine into uracil via a 3-step process of sulfonation, deamination and desulfonation (Figure 4.1). Bisulfite conversion does not alter methylated DNA, therefore, creating sequence differences between methylation and unmethylated DNA (Table 4.1; Hernández et al. 2013; Delaney, Garg, and Yung 2015). During PCR uracil pairs with adenine in the first cycle, then in the remaining cycles adenine will bind with thymine; therefore, the uracil bases are amplified as thymine, whereas, methylated cytosines remain as cytosine (Table 4.1). The proportion of cytosine/thymine can then be used to determine DNA methylation percentage (Hernández et al. 2013).

Figure 4.1 - Stages involved in the bisulfite conversion process.

Several PCR-based methods have been developed to determine DNA methylation including methylation-specific PCR (Herman et al. 1996), COBRA (Xiong and Laird 1997), MethyLight (Eads et al. 2000), and methylation-sensitive high-resolution melt (Wojdacz and Dobrovic 2007). The drawback of many of these PCR based techniques is they can only determine the overall methylation of a PCR product (i.e. they do not allow the investigation of methylation of individual CpG site). Sequencing of PCR products can be used to determine the methylation status of individual CpG sites within the PCR product (Kurdyukov and Bullock 2016). The limitation of sequencing-based methods is the access to a DNA sequencer; however, within our laboratory, we have

Desulphonation H2O NH4+ Deamination Sulphonation HSO3- OH- OH- HSO3- NH2 N H N Cytosine O NH2 N H N Cytosine sulphonate O SO3- O N H NH Uracil sulphonate O SO3- O N H NH Uracil O

access to a pyrosequencer which can be used to determine the methylation percentage of individual CpG sites within a PCR product.

Table 4.1 - Bisulfite conversion induce sequence differences between methylated and unmethylated

DNA.

Unmethylated DNA Methylated DNA

Original Sequence T-G-A-C-C-G-A-C-G-C T-G-A-C-m_C-G-A-m_C-G-C

Bisulfite converted sequence T-G-A-U-U-G-A-U-G-U T-G-A-U-m_C-G-A-m_C-G-U

PCR product T-G-A-T- T-G-A-T-G-T T-G-A-T - C -G-A- C -G-T

4.1.1.2.2 Pyrosequencing based DNA methylation detection

Pyrosequencing is one of the most commonly used methods to determine DNA methylation accurately at specific CpG sites. Originally pyrosequencing was developed to identify SNPs; however, after bisulfite conversion, pyrosequencing can be used to determine the ratio of the bisulfite-converted C/T SNP (Delaney et al. 2015; Tost, Dunker, and Gut 2003)

Following bisulfite conversion, DNA undergoes PCR with one of the primers labelled with biotin. The biotin labelled PCR product then binds to streptavidin-coated beads and DNA is denatured creating single-stranded DNA. The unlabelled DNA strand is washed away, leaving the labelled strand free to be sequenced. A sequencing primer is then introduced, and nucleotides are dispensed in a specific order to be incorporated by DNA polymerase (Harrington et al. 2013). The incorporation of nucleotides releases pyrophosphatase which can be converted into light by an enzyme cascade (Figure 4.2). The amount of light produced is proportional to the amount of pyrophosphatase generated and therefore the number of nucleotides incorporated (Delaney et al. 2015; Harrington et al. 2013; Tost and Gut 2007). DNA methylation can be quantified by the amount of light produced by the incorporation of cytosine (methylated DNA) compared to thymine (unmethylated bisulfite-converted cytosine).

54 Figure 4.2 – Pyrosequencing enzyme cascade. Biotin labelled primer binds to streptavidin-coated

beads, DNA is then denatured, and the sequencing primer binds to the labelled DNA strand. Nucleotides are then dispensed and incorporated into the sequenced DNA strand releasing PPi. PPi is then converted into ATP in a reaction using the substrate APS catalysed by the enzyme Sulfurylase. ATP in the presence of the substrate luciferin is then converted into oxyluciferin and light by the enzyme luciferase. The amount of light produced is proportional to the number of nucleotides incorporated. Any unincorporated nucleotide is degraded by apyrase before the dispensation of the next nucleotide. dNTP, nucleotides; PPi, pyrophosphate; APS, adenosine phosphosulfate; ATP, adenosine triphosphate.