CpG Islands 

 In some animals, cytosine bases occurring in the sequence CG are frequently methylated to 5-methylcytosine. Yet even in species where most such residues are methylated, there are genome segments, known as CpG islands, which are conspicuously not methylated. These are short, dispersed regions of DNA with a high frequency of CpG dinucleotides relative to the bulk genome, but they are not methylated. In the haploid human genome, it is estimated that there are some 30 to 45 × 10³ CpG islands. Such regions of DNA are at least 200 bp in length and have a G + C content greater than 50%. Outside these islands, the frequency of CpG sequences is depleted to about only 20% of that expected on a random basis. This bias against CpG sequences that would be methylated is believed to be caused by methylcytosine bases that deaminate spontaneously 10 to 20 times more readily than normal cytosine bases. Furthermore, the product of deamination of 5-methylcytosine is thymine, a normal base, so this mutagenic event is more difficult to repair than usual.

 CpG islands are associated with the 5′ ends of housekeeping genes and of a few tissue-specific genes (highly tissue-specific genes usually lack islands). They have an open chromatin structure, and it has been postulated that they are sites of interaction between transcription factors and promoters. All known widely expressed genes are associated with more than one CpG island, which usually includes the transcription start site, and a few of these genes have an additional island in the 3′ direction. The average length of CpG islands is about 240 bp in widely expressed genes, but there is no typical size. Most are between 200 and 1400 bp, and a majority of islands are 200 to 400 bp. The average size of islands associated with genes with limited expression is slightly smaller.

Less than half of tissue-specific expressed genes and of those with limited expression are associated with CpG islands that occur in one or more exons. This implies that CpG islands may be used to identify transcripts, since they occur in exons. Furthermore, CpG islands cover the whole or part of the promoter regions containing canonical TATA box and CAAT box, and they include GC-rich promoters lacking a typical TATA box. Such promoters are considered typical of housekeeping genes.