Association of charge clusters with functional domains of cellular transcription factors.

Using rigorous statistical methods, we have identified and evaluated unusual properties of the distribution of charged residues within the sequences of eukaryotic cellular transcription factors. Virtually all transcription factors, including GAL4, c-Jun, C/EBP, CREB, Oct-1, Oct-2, Sp1, Egr-1, CTF-1, steroid and thyroid hormone receptors, and others, carry one or more highly significant charge clusters. For the most part these clusters (conserved within families of homologous proteins) are of positive net charge but contain also substantial numbers of acidic residues. Predominantly basic charge clusters are often, but not exclusively, associated with DNA-binding domains, and vice versa. Negative charge clusters of note occur only in the yeast protein PHO4 and in the proteins encoded at the Drosophila loci zeste (zeta) and knrl. This dearth of statistically significant negative charge clusters raises questions with respect to the generality of acidic activation domains. A number of sequences (Oct-1, Oct-2, zeste, Dhr23, E75, and knrl) contain multiple charge clusters together with one or more significantly long uncharged regions. The occurrence of multiple charge clusters is a rare phenomenon (found in less than 3% of all proteins, mainly in Drosophila developmental control proteins and in transactivators of eukaryotic DNA viruses). Most of the proteins with zinc-binding "fingers" carry a mixed charge cluster centered at the zinc-finger motif preceded by a long uncharged stretch, suggestive of a modular structure for these proteins.