Validating CHARMM parameters and exploring charge distribution rules in structure-based drug design.

Area of Science:

• Computational chemistry
• Molecular modeling
• Drug discovery

Background:

• Non-nucleoside inhibitors of HIV-1 reverse transcriptase are crucial for AIDS therapy.
• Accurate ligand parameters are essential for reliable molecular simulations in drug development.
• CHARMM22 is a widely used molecular mechanics force field.

Purpose of the Study:

• To systematically evaluate recently developed ligand parameters for CHARMM22.
• To assess the accuracy of these parameters for TIBO compounds, a series of HIV-1 reverse transcriptase inhibitors.
• To investigate the impact of different charge assignment schemes on binding affinity calculations.

Main Methods:

• Thermodynamic integration simulations were performed for 44 pairs of TIBO compounds.
• Ligand parameters were evaluated against the CHARMM22 force field.
• CHELPG charges were adopted from localized regions of model compounds for charge assignment.

Main Results:

• An overall average unsigned error (AUE) of 1.3 kcal/mol was achieved for relative binding affinities.
• Accuracy was dependent on substituent size and functional group class.
• Thioether derivatives exhibited large systematic errors, while alkyl, allyl, aldehyde, nitrile, trifluoromethyl, and halide derivatives showed low errors.
• CHELPG charges from model compounds provided reliable results in the absence of specific bond-charge increments.

Conclusions:

• The CHARMM22 force field, with appropriate ligand parameterization, can accurately predict binding affinities for certain TIBO derivatives.
• Systematic errors observed for thioethers necessitate further refinement of parameters for these functional groups.
• Using model compounds to derive charge distributions and bond-charge increments is advantageous for expanding fragment libraries in drug development.