A stretch of DNA, generally several hundred to several thousand bases long, that is relatively rich in dinucleotides containing the bases cytosine and guanine, with the cytosines (C) located ‘upstream’ (i.e. 5′) of a guanine (G) residue. The ‘p’ denotes a phosphodiester bond, meaning that the C and G residues are joined along the same strand of DNA, not paired between complementary strands. An example is 5′-CGCG-3′. CpG sequences are generally about five times less frequent than would be expected throughout the genome—a phenomenon called CG suppression–but their abundance is much greater in the promoter region near the start site of many eukaryote genes. Methylation of the (normally unmethylated) cytosine residues in the promoter suppresses transcription of the gene; this DNA methylation, which is transmitted through subsequent cell divisions, is significant in the process of ageing and, by switching off tumour suppressor genes, the formation of tumours. It is also a feature of heritable suppression of paternal or maternal genes (see gene imprinting). CpG islands outside promoters are hotspots for mutation, because methylcytosine undergoes spontaneous deamination to thymidine, which creates a T–G mismatch in the base pairing. The DNA repair machinery cannot resolve which is the correct template base, T or G, so in some cases makes the repair T–A instead of the correct C–G.