Silk fibroin: structural implications of a remarkable amino acid sequence

The amino acid sequence of the heavy chain of Bombyx mori silk fibroin was derived from the gene sequence. The 5,263-residue (391-kDa) polypeptide chain comprises 12 low-complexity "crystalline" domains made up of Gly-X repeats and covering 94% of the sequence; X is Ala in 65%, Ser in 23%, and Tyr in 9% of the repeats. The remainder includes a nonrepetitive 151-residue header sequence, 11 nearly identical copies of a 43-residue spacer sequence, and a 58-residue C-terminal sequence. The header sequence is homologous to the N-terminal sequence of other fibroins with a completely different crystalline region. In Bombyx mori, each crystalline domain is made up of subdomains of approximately 70 residues, which in most cases begin with repeats of the GAGAGS hexapeptide and terminate with the GAAS tetrapeptide. Within the subdomains, the Gly-X alternance is strict, which strongly supports the classic Pauling-Corey model, in which beta-sheets pack on each other in alternating layers of Gly/Gly and X/X contacts. When fitting the actual sequence to that model, we propose that each subdomain forms a beta-strand and each crystalline domain a two-layered beta-sandwich, and we suggest that the beta-sheets may be parallel, rather than antiparallel, as has been assumed up to now.