Measuring DNA methylation

3.2.1 Method selection

DNA methylation was first discovered in 1948 (Hotchkiss, 1948), since then there has been an evolution in the methods that can be used to measure this epigenetic mark. Measurement of cytosine methylation levels involves either the use of methylation- specific restriction enzymes, bisulphite conversion of DNA, or enrichment with methylation targeting enzymes. Methodologies fall into two categories, epigenome - wide or locus-specific.

Levels of methylation across the genome can be measured in a non-locus specific manner by global methylation analysis techniques which include luminometric methylation assay (LUMA), enzyme-linked immunosorbent assays (ELISAs), high performance liquid chromatography (HPLC), repetitive element DNA methylation analysis (e.g. LINE1) (Kurdyukov and Bullock, 2016). However, these techniques do not provide information on the levels of methylation at certain loci. This may lead to losing key information on the methylation patterns in certain loci of interest. Measurement techniques which allow for the identification of methylation at individual loci include bisulphite whole genome sequencing (BS-WGS), which provides details of methylation levels across the whole genome; or regional levels can be measured via pyrosequencing. Epigenome-wide methylation analysis determines global methylation levels, examples of such methods include high performance liquid chromatography (HPLC) and bisulphite whole genome sequencing (BS-WGS); the latter of which also provides detail on the location of the methylation marks. There are methods, such as array based chips, which give levels of methylation from loci spread across the genome, representing approximately 2% of the sites that could be methyalted.

Methods which do not allow for measurement of methylation at individual loci can be difficult to link changes to biologically significant outcomes, however, the cost of the global methylation methods of £5 per sample is significantly cheaper than the genome wide loci specific methods. The use of Illumina Infinium HumanMethylation450 BeadChip ArraysTM _{(approx. £250/sample) and BS-WGS (approx. £3,600/sample) do,}

however, provide information on individual CpG sites. These methods are often used to scan the genome for novel DMRs. However, in addition to their high cost, they require large amounts of input DNA and even then, significant results found using the 450k array should be validated using sequencing-based methods. Microarrays, such as the 450k array, are susceptible to batch effects and require comprehensive analytical techniques to correct for this. In addition, due to the requirement to have two different probes types on the same array, to be able to still measure methylation in areas of high CpG density, data processing can be problematic as a result of the differences in the performance of the probes (Pidsley et al., 2013).

An alternative to epigenome-wide methylation analysis, which can be used when a candidate gene has been identified, is the measurement of methylation levels of smaller regions of DNA. The methylation of these smaller regions can be quantified after bisulphite conversion using polymerase chain reaction (PCR)-based methods, which include methylation sensitive PCR (MSP), MethyLight (utilising real-time PCR), methylation-specific high-resolution melting and bisulphite pyrosequencing.

Methylation sensitive PCR permits screening of methylation at a specific site; however, it is not quantitative. Real-time PCR methods do allow quantitative measurement at specific sites and utilise primer sets designed to amplify methylated or unmethylated regions; however, this will only allow for the detection of completely methylated or unmethylated sites within the target region.

Having regard to the advantages and disadvantages of the various available methods, including budgetary constraints, and having considered the number of samples that would have to be analysed during the course of the proposed research the decision was made to use a PCR-based method for analysing DNA methylation at specific identified CpG sites of interest.

Methylation levels of CpG sites within a specific region can be assessed by using the primer extension sequencing method, pyrosequencing. Pyrosequencing uses bisulphite-converted amplified DNA and provides sequence data over short DNA stretches. Methylation can be quantified by calculating the ratio of unconverted cytosine to converted thymine at the CpG loci (see below for further details on bisulphite conversion). Benefits of the pyrosequencing method include the inclusion of

an internal bisulphite conversion control and the ability to measure individual methylation values of CpG sites in the short stretch of DNA sequenced. As the length of DNA sequenced is relatively short, less than 60 base pairs (bp), this limits the number of CpG sites that can be assessed in one assay.

Bisulphite conversion

In order to assess the DNA methylation of a target CpG site when using PCR-based methods, the DNA must first undergo bisulphite conversion.

Without bisulphite conversion, during PCR amplification, methylated cytosines are replaced by unmethylated cytosine (He and Cole, 2015), and therefore the methylation is lost after the first cycle of PCR.

Bisulphite conversion is the deamination of unmethylated cytosine to uracil. This reaction does not occur if the cytosine is methylated. Hence after bisulphite conversion the only cytosine remaining in the DNA is methylated cytosine. During PCR these methylated cytosines are converted to unmethylated cytosine and the uracil is converted to thymine. This preserves the methylation information of the original DNA sequence after PCR amplification (Grunau, Clark and Rosenthal, 2001).

The original method for bisulphite conversion was proposed by Frommer et al. (1992). Briefly, precipitated DNA is incubated with sodium bisulphite hydroquinone at pH5 & 50o_{C for 16 or 40 hours, after which remaining bisulphite is removed via dialysis and}

the resulting solution is neutralized, desalted and re-suspended in storage solution and is ready for PCR (Frommer et al., 1992). The harsh conditions of this original method have been shown to degrade the DNA during treatment. New commercial kits are not as problematic as several chemical improvements have been made to the reaction and therefore there is improved recovery. Hernández et al. (2013) compares the commercially available kits. The most commonly used kits in research are EpiTectTM _{Bisulphite kit produced by Qiagen (Qiagen N.V. and its subsidiaries), or EZ}

3.2.2 Tissue Selection

In DNA methylation research, the tissue of interest must be carefully chosen as more variability is observed between tissues from the same individual, than within tissue between individuals (Kochanek et al., 1990). The ability to obtain tissue from metabolically relevant tissues within humans, such as muscle, liver and adipose tissue can be problematic in healthy individuals that have not required surgery. One key tissue used in genetics research, due to the less invasive collection method, is whole blood. This source of DNA is most beneficial when looking for genetic polymorphisms. Other cells used for this line of work include buccal epithelial cells and hair. However, it is not going to be relevant to use hair follicles or buccal swabs for many epigenetic studies where epigenetic signatures are specific to cell type. For example, when considering the role of DNA methylation in gene regulation, there is greater variance between tissues in the same individual than that observed between individuals within a single tissue (Eckhardt et al., 2006; Byun et al., 2009; Davies et al., 2012). Consideration must also be given for samples derived from multiple cell types that have the potential for different methylation patterns (Siegmund and Laird, 2002; Wu et al., 2011). The implication when designing a research study investigating epigenetic modifications is that the samples studied must be from the same tissue and chosen so that any changes that are observed in DNA methylation in a specific cell type are biologically relevant to the research question. In addition to epigenetic signatures depending on the type of cell, they are also influenced by the metabolic processes within the cell and the environment the cell is exposed to.

In regards to inflammation, within the blood are white blood cells that are important in systemic function and inflammation (Afman and Müller, 2012). These cells contain DNA that can be easily extracted from small volumes of blood. PBMCs isolated from blood are a heterogeneous mix of white blood cells and therefore will have varying epigenetic profiles. It is also important to consider that the blood composition, leukocyte composition of the whole blood was found to correlate with DNA methylation (Lam et al., 2012), an issue that may influence associations between environmental factors and DNA methylation when using blood samples. Therefore, composition of whole blood should be determined for each individual and used in statistical models to correct the methylation values or complete the DNA methylation analysis on specific

blood cells types. Differences in DNA methylation in PBMCs have also been associated with demographic factors such as sex, age and ethnicity (Lam et al., 2012). In this thesis all blood samples have had the cell composition analysed on a Beckman Coulter Counter (Beckman, USA) to allow for appropriate adjustments to be made for the methylation values. In this thesis, the method presented by Jones et al., (2015) has been used to adjust DNA methylation values to account for the white blood cell composition. This method uses a sum of the mean methylation for the site and the unstandardised residual from a linear regression between DNA methylation (dependent variable) and the individual white blood cell counts (independent variables) to give an adjusted methylation value.