Homologous sequences in steroidogenic enzymes, steroid receptors and a steroid binding protein suggest a consensus steroid-binding sequence.

The amino acid sequences of two steroidogenic enzymes, P450c17 (steroid 17 alpha-hydroxylase/17,20 lyse) and P450c21 (steroid 21-hydroxylase), are only 28.9% identical. However, these proteins share a region of 21 amino acids bearing 17 identical residues, which we previously suggested may represent the steroid binding site. We assembled a sequence database of known steroid-binding proteins and searched this with the sequence of this 21 amino acid region. The steroidogenic enzymes, P450c17, P450c21, P450scc (the cholesterol side-chain cleavage enzyme), and P450c11 (steroid 11 beta/18-hydroxylase) share a subregion of 17 amino acids having at least 15 identical residues. Related sequences were identified in a computerized search of the available sequences of steroid hormone receptors and binding proteins. These sequences were invariably found within larger domains previously associated with steroid binding. From these we propose a more general consensus sequence of LPLLL +/- 000KDRE0LKRL +/- PV, where +/- refers to any charged amino acid, and 0 refers to an uncharged amino acid. This consensus sequence predicts 147 or 187 total amino acids in 11 human proteins examined (78.6%). An equivalent degree of sequence identity, 178 of 221 amino acids (80.5%) was found among 13 animal homologs of these human proteins. The ability of this consensus sequence to predict 325 of 408 amino acids (79.7%) strongly suggests this sequence is necessary, if not sufficient, for a steroid binding site in many proteins. Lecithin-cholesterol acetyl transferase, cholesterol ester transfer protein, and steroid sulfatase did not have sequences similar to our consensus sequence.