Differentiation requires continuous active control.

The problems posed by differentiation that appear most soluble by a passive control mechanism can readily be solved by an active mechanism. Given the need for plasticity in gene expression in different cell types at different stages, an active mechanism may be advantageous, even essential. It is striking how few changes during differentiation are completely irreversible, the gene rearrangements leading to immunoglobulin expression being one clear exception. Indeed, a prediction of the active-control hypothesis is that any nucleus exposed to the appropriate constellation of proteins at the appropriate concentration should be able to perform functions typical of any given differentiated cell type. In the next decade, the elucidation of novel memory mechanisms, or feedback loops, will substantially increase our understanding of how stable differentiated states can be maintained by continuous active control.