Comparative Methylation Hybridization

For over 50 years, we have known the structure of the DNA molecule, thanks to James Watson, Francis Crick, and the often-overlooked Rosalind Franklin (Watson & Crick, 1953). More recently, the sequencing phase of the Human Genome Project was completed in 2004 (International Human Genome Sequencing Consortium, 2004). Since then, the contents of our genome—in other words, the genes "spelled out" by the sequence, the variations in this sequence that differentiate individuals, and various other features at the DNA level—have been extensively annotated and described (International HapMap Consortium, 2003; Feuk et al., 2006; Hurles et al., 2008). The function of many human genes remains unknown, however, especially genes that do not encode proteins (Pennisi, 2007).

This wealth of genomic knowledge begs an important question: Why don't scientists better understand how genes contribute to common, complex human diseases such as cancer, heart disease, and diabetes? Environmental risk factors certainly play a significant role in these conditions, but genetic predisposition is also a causal player. Merely studying the DNA sequence has, however, produced more questions than answers, causing researchers to broaden their focus to ask what controls how and when genes are expressed. The emerging field of epigenetics (literally, "above genetics") seeks to answer such queries by focusing on nongenetically inherited changes that alter gene expression independent of the DNA sequence itself (Allis, 2007).

One important source of epigenetic modification in humans and many other species is DNA methylation, or the addition of a methyl (-CH₃ ) group to the cytosine of a CG pair in a sequence of DNA. Although the mechanisms of methylation are not entirely understood, the methyl group is usually thought to prevent transcription by recruiting protein complexes that either condense chromatin or act as a physical barrier against assembly of the transcription complex (Bird & Wolffe, 1999). The ability to alter gene expression via DNA methylation is key to maintaining the specificity and identity of different cell types in the body. After all, if cells were not selective in their gene products, this would result in a chaos of proteins and an enormous waste of cellular resources.

Figure 3: Whole genome microarray.

Blue bars along the idiogram represent tiled bacterial artificial chromosome (BAC) clones, each 100-150 kb with 50 kb overlapping between adjacent clones. Each of the approximately 32,000 spots on the sub-array, or “chip,” represents a single BAC clone that, when combined, cover >99% of the euchromatic human genome. The BAC clones are mapped and annotated in public genome databases such as Ensembl and UCSC Genome Browser.

© 2008 Nature Education All rights reserved.

Although there are many types of genomic microarrays, representational oligonucleotide microarray analysis (ROMA) and NimbleGen high-density arrays, which use short DNA sequences ranging from 50 to 100 nucleotides in length to represent features across the genome, are best suited for high-density methylome studies (Feuk et al., 2006). A more cost-effective initial screening step may involve use of arrays derived from larger pieces of DNA.

The genomic microarray used for this experiment was built from bacterial artificial chromosome (BAC) clones also used in the sequencing of the human genome. These BAC clones are composed of 100- to 150-kilobase pieces of the human genome that can be ordered along each chromosome in a tiled fashion with 50-kilobase overlapping ends (Figure 3). It takes approximately 32,000 100-kilobase BAC clones to cover the euchromatic human genome in what is called a tiling genomic array (Fiegler et al., 2003). Each spot on the array contains many copies of the BAC clone to detect differences in the test DNA samples. To reduce the complexity of the DNA sequence for hybridization, the clones are copied using a series of degenerate oligonucleotide-primed polymerase chain reactions (DOP-PCRs) (Figure 4). Each DOP-PCR primer is flanked by known 5′ and 3′ anchor sequences with six random nucleotides inserted between them, thus ensuring that the entire clone sequence will be amplified. The 5′ anchor sequence further serves to flag the amplified segments for an additional PCR step in which an amino (-NH₃ ) group is linked to the end and used to affix the clone segment onto the microarray slide (Fiegler et al., 2003). To prevent false positive hybridization of repetitive DNA elements present in both the clone and the test DNA, repetitive Cot-1 DNA is first hybridized to the array and also added to the test DNA samples. Cot-1 DNA is the highly repetitive fraction of the human genome that, in the single-stranded form, is known to quickly "find" and reassociate with a complementary sequence elsewhere in the genome (Strachan & Read, 1999).

Figure 4

Figure Detail

Once the microarrays are ready, the next step in CMH is to prepare the test DNA samples that will be hybridized to the array. In order to maintain cell growth and division, cells are passaged in vitro; the more passages a cell undergoes, the more times it has divided, and the "older" it is. Passaging SMCs from the aorta (aoSMCs) in culture results in cells with characteristics similar to migratory SMCs (Ying et al., 2000). Furthermore, several genes in these cells exhibit altered methylation statuses in early versus late passages (Kim et al., 2007; Post et al., 1999). These changes are also seen in atherosclerotic lesions, which suggests that epigenetic changes in proliferating SMCs could indeed be involved in the initiation and progression of heart disease. For this experiment, passage 5 (p5) aoSMCs will serve as the control, and passage 8 (p8) aoSMCs will be used as the disease sample, representing aoSMCs more prone to proliferation and migration as compared to p5 cells. Identifying differences in methylation patterns between p5 and p8 aoSMCs would support the hypothesis that changes in the epigenome contribute to myocardial infarction.

To flag methylated sites across the genome in both sets of aoSMCs, CMH uses restriction enzyme biology followed by ligation-mediated PCR. First, the DNA is treated with the restriction enzyme Sma I, which recognizes and cuts the DNA sequence CCC^GGG (Figure 5B, part i). The presence of a methyl group at the CG residue will prevent Sma I from cutting at that site. The DNA is then subjected to a second digestion by Xma I (Figure 5B, part ii), which recognizes C^CCGGG and is undeterred by a methylated CG (Inazawa et al., 2004). The resulting test DNA samples are cut into pieces, the ends of which are methylated sites with five base pairs of "sticky overhang" (-CCGGG). A linker sequence is next ligated to the sticky ends of the methylated DNA (Figure 5B, part iii). This product is then amplified using PCR so that each digested segment will be detectable on the array. PCR for the control sample, p5 aoSMCs, uses a nucleotide mixture in which the cytosines are labeled with the green fluorescent tag Cy3. On the other hand, the disease sample, p8 aoSMCs, undergoes PCR with red Cy5-labeled cytosine residues (Figure 5B, part iv). These different fluorophores are what will allow the identification of control versus disease DNA.

Figure 5

Figure Detail

The control and test DNA samples are then both hybridized to the microarray slide and scanned with a microarray reader (Figure 5C). The next challenge is to translate the red and green spots into a meaningful set of results that will inform researchers of what role epigenetic changes might play in causing heart attacks. This analysis consists of comparing the relative hybridization of the Cy3-labeled control DNA (p5 aoSMCs) to the Cy5-labeled test, or disease, DNA (p8 aoSMCs). Any spot on the array in which the intensity of the green fluorescent tag Cy3 is significantly greater than the fluorescence of the red Cy5 tag represents a clone in which the control DNA is more methylated (hypermethylated) compared to the disease DNA. Conversely, any spot in which the red fluorescent signal is greater indicates a clone segment that is hypermethylated in the disease DNA and less methylated (hypomethylated) in the control. The researchers must decide how different (what level of significance) the fluorescent signals need to be to warrant further investigation, a process that involves subtracting out the noisy background and false signals that accompany most microarray experiments. The final product is a list of DNA segments differentially methylated in p5 aoSMCs versus p8 aoSMCs.

So, what comes next? One hundred thousand base pairs of DNA is a vast genetic landscape, comprising anywhere from one to many genes, and it is no minor challenge to identify exactly which gene or genes are changing in methylation status. Researchers thus turn to bioinformatics resources such as genetic databases to search for likely candidate genes within the regions identified by a microarray, usually by identifying Sma I sites in the clone and finding the genes associated with those sites. Alternative lines of evidence, such as gene expression experiments and known candidates from the literature, may also be incorporated. For example, if one clone segment showing differential methylation contains a gene involved in regulating the cell cycle (i.e., cyclin-dependent kinase) and was found to have different expression levels in the p5 versus p8 samples, that region may be of particular interest for follow-up studies. The hypothesis moving forward might be that an aoSMC begins to proliferate and form plaques due to a change in methylation status of that cell's cell cycle control gene. Further experiments might include CMH using a microarray of smaller clones in selected regions to provide a higher resolution, or perhaps a single nucleotide polymorphism (SNP) array that has the power to detect methylation changes at a single nucleotide (Suzuki & Bird, 2008).

Epigenomic profiling holds great promise for understanding, predicting, and treating numerous human diseases, especially in its ability to account for the influence of environmental factors on disease states (Feinberg, 2007). Several trials of methylation-altering cancer drugs are already underway, and the potential to epigenetically treat other diseases is the next great frontier in this "post-genome" age. These discoveries will rely on successful experimental methods such as CMH to study genome-wide methylation patterns. As with many experiments, CMH yields answers and questions in roughly equal proportion. However, increased knowledge of how methylation changes in disease states informs the next set of experimental questions, thereby leading researchers closer to the ultimate goal of unlocking the epigenome.