The conflict theory of genomic imprinting predicts that imprinted genes are
growth enhancing when paternally expressed and growth suppressing when mat
ernally expressed. The expression pattern of autosomal imprinted genes gene
rally fits these predictions. However, the conflict theory cannot easily ac
count for the pattern of X-linked imprinting in humans and mice. This has l
ed us to propose a novel hypothesis that X-linked imprinting has evolved to
control sex specific gene expression in early embryos. The hypothesis link
s paternal X-imprinting (i.e. paternal copy silencing) to random X-inactiva
tion and the retention of Y-linked copies, and links maternal X-imprinting
to escape from random X-inactivation and the loss of Y-linked copies. The h
ypothesis offers a good explanation of the existing data on X-imprinted gen
es. (C) 1999 Academic Press.