Predicting Distribution Coefficients (LogD) of Cyclic Peptides Using Molecular Dynamics Simulations.

Abstract

Purpose: The distribution coefficient (LogD) is a critical property for oral peptide drug design. In this study, we focused on cyclic peptides (octreotide and its analogs) and aimed to determine their LogD values at four pHs using both the simulation and experimental approaches.

Methods: For the experimental approach, the shake-flask method with LCMS quantification was employed to determine LogD values. For the simulation approach, the partition coefficient (LogP) was obtained from the solvation free energy calculations using molecular dynamics (MD) simulation. The LogD values were then calculated from the obtained LogP values considering the predicted pKa and ionization states of each peptide residue. More peptide properties such as polar surface area (PSA), number of intramolecular hydrogen bonds, solvent accessible surface area (SASA), and radius of gyration (R_g) were also calculated with the aid of MD simulation.

Results: For a total of 28 LogD values across four pHs, the predicted values from the simulation under the OPLS-AA forcefield agreed with the experimental values, with an average deviation of 1.39 ± 0.86 log units, displaying better predictions compared to the data generated under the CHARMM forcefield or using commercial software. In addition, the analysis of PSA, SASA, and R_g data suggested the peptides exhibited some conformational flexibility in both aqueous and organic phases.

Conclusions: The method developed in this study can predict the LogD values at a wide pH range covering multiple formulation/physiological conditions and therefore can provide insights into designing oral peptide drugs, especially for early-stage projects.