In the behavioural sciences, imprinting is the process through which offspring learn to recognize their parents. In the classical example, newly hatched geese imprint on the first moving object that they see during a crucial time period after hatching. Imprinting also has a logical counterpart in the biological sciences: genomic imprinting. The latter is the differential marking of the genome according to whether it was derived from the maternal contribution in the egg or the paternal contribution in the sperm.

The first evidence for the non-equivalence of the parental genomes came in 1984 when two landmark papers reported that mouse embryogenesis could not proceed to completion without the contribution of both maternal and paternal genomes. Surani and colleagues, using activated eggs as recipients for a maternal or paternal pronucleus, showed that only those that received a paternal pronucleus could develop to term. In similar experiments, transplanting pronuclei between one-cell-stage embryos, McGrath and Solter reported that only those that received both a maternal and a paternal pronucleus developed properly.

This early work based in experimental embryology showed in broad terms that the maternal and paternal genomes are not equivalent. Evidence of imprinting at the level of individual genes came in 1991, when a series of papers from the Tilghman, Barlow and Efstratiadis groups reported that, in the mouse, three different genes are specifically and consistently expressed from only one parental copy: H19 and Igf2r from the maternal copy and Igf2 from the paternal copy. One correlate of Mendel's laws of inheritance is that the sex of the parent contributing a given gene is irrelevant; however, these papers showed that in the case of imprinted genes, it does matter — a mutation has to be passed from a specific parental gender for the offspring to have the phenotype. This non-Mendelian pattern of inheritance has also been observed in studies of human genetic diseases that result from mutations in imprinted genes.

Delving deeper, the first clue to the mechanisms used to distinguish the parental copies came in 1993, when the Tilghman and Surani groups reported studies of the epigenetic modifications that govern H19 imprinting. Both groups showed that the promoter region of H19 is subject to CpG methylation and compacted chromatin on the silenced paternal copy, whereas the expressed maternal copy is relatively unmethylated and has open chromatin. These and other epigenetic marks that distinguish the maternal and paternal genome contributions must be erased in the germ cells of each generation, and then re-established to reflect the sex of the individual in which they reside. Moreover, once these marks are established, they must be maintained in somatic cells throughout the process of cell division.

At present, we know of approximately 80 imprinted genes in the human genome and there might be many more. They have important roles in mammalian physiology, growth and behaviour, and numerous diseases are associated with imprinting defects. Beyond their function, imprinted genes are appreciated as a useful and important model for the study of epigenetic regulation of gene expression, because they provide a natural system in which epigenetic modifications are the main determinants of a functional state.