The mechanism of interaction of filaggrin with intermediate filaments. The ionic zipper hypothesis.

Filaggrins of mammalian epidermis represent archetypical examples of intermediate filament-associated proteins that can bind large numbers of intermediate filaments in vitro (and keratin filaments in vivo) into macrofibrils. To explore the mechanism of this interaction, the secondary structures of filaggrins were analyzed. As much as 80% of mouse and human filaggrins consist of multiple repeating elements. The first level consists of a tetrapeptide beta-turn motif in which about 35% of the turns are positively charged and about 10% are negatively charged. At the next level, triplets of this motif form segments 13 to 14 residues in length, which in turn are repeated two to six times into blocks separated by short hydrophobic sequences to constitute a complete filaggrin molecule. Thus, filaggrins evolved by frequent duplications of a primordial repeat unit of about 13 to 14 residues with subsequent retention of the conserved beta-turn and charge characteristics. To test how these features bind filaments, two approaches were used. Of a series of synthetic peptides, those of 20 to 26 residues (about 2 segments) containing at least five beta-turns with a net charge of +2 (that is, about 40% of the turns are positively charged) were as effective as full length filaggrin in binding large numbers of both type I/II keratin and type III vimentin/desmin filaments, as judged by electron microscopy. Secondly, macrofibrils formed from unlabeled filaggrin and keratin filaments labeled in vivo with [1-13C]glycine or L-[4,4,5,5-2H4]lysine were probed by nuclear magnetic resonance. The effective isotropy and time scale of mobilities of the glycine-labeled end domains were essentially identical in keratin filaments alone and those bound in macrofibrils, suggesting that filaggrins do not bind filaments by way of their end domains. However, the lysine-labeled rod domains of the filaments in macrofibrils were considerably more constrained than in filaments alone. These data support the hypothesis that filaggrins bind filaments by way of simple ionic and/or H-bonding interactions between the conserved positive and negative charges on the beta-turns of filaggrins and the conserved distributions of negative and positive charges along the packed rod domains of intermediate filaments, as in an ionic zipper.