CHEMICAL SYNTHESIS, STRUCTURAL MODELING, AND BIOLOGICAL-ACTIVITY OF THE EPIDERMAL GROWTH FACTOR-LIKE DOMAIN OF HUMAN CRIPTO

Cripto, also known as human teratocarcinoma-derived growth factor 1 (T DGF-1), contains a 40 amino acid region with some similarity to the ep idermal growth factor (EGF) domain. However, sequence homology is larg ely restricted to the classical cysteine/glycine motif with only limit ed similarities in other regions. Significant differences to human EGF include the absence of all seven residues between the two N-terminal half-cystines and a five-residue shorter loop between the third and fo urth half-cystines. We examine the hypothesis that, in spite of these differences, cripto can adopt the characteristic EGF-like 1-3, 2-4, 5- 6 disulfide bond pattern. A comparative structural model of the growth factor cripto was constructed on the basis of its similarity to EGF, transforming growth factor alpha (TGF-alpha), and the EGF-like domain of human clotting factor IX. The predicted disulfide bridges and disul fide-bridged loops were analyzed and appear viable in the modeled stru cture. Moreover, to ascertain the importance of disulfide arrangement for cripto bioactivity, two 47-residue peptides were synthesized and t hen refolded using either a simple oxidative or a controlled sequentia l refolding protocol. The cripto peptides were tested for their abilit y to stimulate MAP-kinase activity, for inhibition of beta-casein indu ction, and for Shc phosphorylation in MDA-MB 453 human mammary carcino ma cells and HC-11 mouse mammary epithelial cells. Data suggest that c ripto does adopt the 1-3, 2-4, 5-6 disulfide pattern and thus forms th e classical EGF-like fold in spite of the significant deletions within the folding domain. The predicted structure of cripto shows some of t he characteristics of both the ErbB1- and ErbB3/ErbB4-binding growth f actors.