19.3: Epigenomic Assays

ChIP: a method for determining where proteins bind to DNA or where histones are modified

Given the importance of epigenomic information in biology, great efforts have been made to study signals that quantify this information. One common method for epigenomic mark measurement is called chromatin immunoprecipitation (ChIP). ChIP technology yields fragments of DNA whose location in the genome denote the positions of a particular histone modification or transcription factor. The procedures of ChIP are described as follows and are depicted in Figure 19.2:

1. Cells are exposed to a cross-linking agent such as formaldehyde, which causes covalent bonds to form between DNA and its bound proteins (e.g., histones with specific modifications).
2. Genomic DNA is isolated from the cell nucleus.
3. Isolated DNA is sheared by sonication or enzymes.
4. Antibodies are grown to recognize a specific protein, such as those involved in histone modification. The antibodies are grown by exposing the proteins of interest to mammals, such as goats or rats, whose immune response then causes the production of the desired antibodies.
5. Antibodies are added to the solution to immunoprecipitate and purify the complexes.
6. The cross-linking between the protein and DNA is reversed and the DNA fragments specific to the epigenetic marks are purified.

After a ChIP experiment, we have short sequences of DNA that correspond to places where histones were bound to the DNA. To identify the location of these DNA fragments in the genome, one can hybridize them to known DNA segments on an array or gene chip and visualize them with fluorescent marks; this method is known as ChIP-chip. Alternatively, one can do massive parallel next-generation sequencing of these fragments; this is known as ChIP-seq. The latter approach, ChIP-seq, is a newer approach that is used much more frequently. It is preferred because it has a wider dynamic range of detection and avoids problems like cross-hybridization in ChIP-chip.

Each sequence tag is 30 base pairs long. These tags are mapped to unique positions in the reference genome of 3 billion bases. The number of reads depending on sequencing depth, but typically there are on the order of 10 million mapped reads for each ChIP-seq experiment.

There is a fairly standard pipeline used to infer the enrichment of the protein of interest at each site in the genome given a set of short sequencing reads from a ChIP-seq experiment. First, the DNA fragments must be mapped to the DNA (called read mapping). Next, we must determine which regions of the genome have statistically significant enrichment of the protein of interest (called peak calling). After these preprocessing steps, we can build different supervised and unsupervised models to study chromatin states and their relation to biological function. We look at each of these steps in turn.

Bisulfite Sequencing: a method for determining where DNA is methylated

DNA methylation was the first epigenomic modification to be discovered and is an important transcrip- tional regulator in that the methylation of cytosine residues in CpG dinucleotides results in “silencing,” or repression, of transcription. Bisulfite sequencing is a method by which DNA is treated with bisulfite before sequencing, allowing the precise determination of the nucleotides at which the DNA had been methylated. Bisulfite treatment converts unmethylated cytosine residues to uracil, but does not affect methylated cy- tosines. Thus, genomic DNA can be sequenced with or without bisulfite treatment, and the sequences can be compared, and the sites at which cytosine has not been converted to uracil in the treated DNA (or, equiv- alently, sites at which there is bisulfite-generated difference between the treated and untreated sequences) are sites at which cytosine was methylated. This analysis assumes complete conversion of unmethylated cytosine residues to uracil, so incomplete conversion can result in false positives (i.e., nucleotides identified as methylated but which in fact were not methylated) [11].