Strategies toward predicting peptide cellular permeability from computed molecular descriptors

Area of Science:

• Pharmacokinetics
• Computational Chemistry
• Drug Delivery

Background:

• Drug efficacy depends on pharmacological and pharmacokinetic properties, including access to the target site.
• Physicochemical and biological barriers are critical in drug candidate selection and development.
• Cellular permeability, a key drug delivery component, is influenced by desolvation potential and lipophilicity.

Purpose of the Study:

• To investigate the relationship between computationally derived molecular descriptors and experimentally measured cellular permeability.
• To evaluate the utility of various geometric descriptors (e.g., molecular volume, surface areas) in predicting drug transport.
• To explore the role of solvation and computational factors in developing structure-based models of cellular permeability.

Main Methods:

• Calculated molecular geometric descriptors (volume, surface areas, cross-sectional areas) for peptides and peptidomimetics.
• Correlated computed descriptors with experimentally measured hydrogen-bond potentials, lipophilicity, and cellular permeabilities.
• Analyzed solvation treatments and computational factors influencing descriptor generation and transport mechanism deconvolution.

Main Results:

• No significant correlation was found between computed molecular descriptors and cellular permeability.
• The study highlighted the inability to identify reliable surrogates for hydrogen-bond desolvation potential among the tested descriptors.
• Existing computational descriptors did not adequately capture the factors governing cellular transport for the studied molecules.

Conclusions:

• Current computational geometric descriptors are insufficient for accurately modeling cellular permeability, particularly for molecules with significant hydrogen-bonding potential.
• Accurate prediction of drug delivery requires improved computational methods that account for desolvation effects.
• Further research is needed to develop more mechanistically meaningful models for predicting drug transport and optimizing drug candidates.