Abstract: Cyclic peptides possess unique conformational stability, diverse biological activities, and favorable target specificity, making them important lead compounds in drug development. Rapid and accurate prediction of their conformational stability not only aids in uncovering the molecular mechanisms of protein misfolding but also provides a theoretical basis for target identification and intervention. This is of great significance for the rational design of structurally stable and highly active cyclic peptide-based drugs. In this paper, the polar chemical bonds C=O, N-H, Cα-H, and C-O, O-H in cyclic peptides are regarded as bond dipoles. The permanent dipole-permanent dipole interaction is used to describe the electrostatic interaction in the system, the permanent dipole-induce dipole interaction and induce dipole-induced dipole interaction are used to describe the polarization. The bonded terms, including the bond-stretching, angle-bending, and dihedral torsion, are also introduced. The polarizable dipole-dipole interaction model is thus developed into a potential function that can be used to rapidly calculate the relative energies of different conformations of cyclic peptides. The potential function is applied to 9 cyclic peptides, total 33 different conformations to rapidly predict the conformational energies of these conformations. The conformational energies of these conformations is also calculated using the AMOEBA and DLPNO-MP2/aug-cc-pVTZ methods. The calculation results show that, compared with the DLPNO-MP2/aug-cc-pVTZ conformational energies, the linear correlation coefficient R2 of our model is 0.9784, and the root mean square deviation is 13.43 kJ·mol-1, slightly better than the linear correlation coefficient 0.9682 and root mean square deviation 16.28 kJ·mol-1 of the AMOEBA method. The results of structural optimization and frequency calculation further suggest the rationality of our model. Furthermore, compared with the AMOEBA polarizable force field, our polarizable model significantly reduces the number of electrostatic terms. The model proposed in this paper may provide a new tool for the research and development of novel cyclic peptides as drug candidate molecules.

Key words: Cyclic peptide, Conformational energy, Bond dipole, Induced dipole, Polarizable potential energy function