The proline-rich glycoprotein from human parotid saliva has a common heptapeptide sequence around four of six N-glycosylation sites (Maeda, N., H. S. Kim, E. A. Azen, and O. J. Smithies, 1985, J. Biol. Chem., 20:11123-11130). A synthetic model of the heptamer protein sequence, NH2-Q(1)-G(2)-G(3)-N(4)-Q(5)-S(6)-Q(7)-CONH2, was examined by nuclear magnetic resonance (NMR) spectroscopy and the ECEPP/2-VAO4A (Empirical Conformation Energy Program for Peptides) energy minimization computer algorithm (Scheraga, H. A., 1982, Quantum Chemistry Program Exchange, 454; Powell, M. J. D., 1964, Quantum Chemistry Program Exchange, 60). The NMR spectrum was almost completely assigned in dimethylsulfoxide-d6 (DMSO), and the amide chemical shift temperature dependence, phi dihedral angles, and chi 1 rotamer populations elucidated. These data indicated that a significant population of the heptamer could exist as a type I beta-turn [4----1 between Q(5) and G(2)] and/or a type II' beta-turn [4----1 between (Q)5 and G(2) and/or a gamma-turn [3----1 between Q(5) and G(3)] with the amino acid chi 1 torsion angles weighted toward the gauche- conformation. Starting from these three possible conformations, the ECEPP/2-VAO4A rigid geometry energy minimization program was used to find the localized predominant in vacuo structures of this heptapeptide sequence. The type II' beta-turn conformation best fits the data based on internuclear hydrogen-bonding distances, minimum potential energy considerations, and the NMR parameters.