A new physics-based computational method TMPfold has been developed and implemented in a public web server for assessment of stability of TM α-helical complexes (accessible at https://opm.phar.umich.edu/tmpfold_server). Using the available 3D structure of an arbitrary TM α-bundle formed by a single or multiple subunits as input, the method calculates the free energies of subunit-subunit and helix-helix association (ΔGasc), identifies stable two-helical units that can initiate folding, and predict the plausible folding pathways for multi-pass transmembrane (TM) proteins. The free energy components include van der Waals, hydrogen bonding and dipole interactions, side-chain conformational entropy, and solvation energy in the anisotropic lipid environment. The method was verified using the experimental ΔGasc values for 36 TM complexes, including dimers of 10 glycophorin A mutants. TMPfold was applied for evaluation of TM helix association energies in 554 PDB structures of 85 seven-helical TM proteins and identification of their stable two-helical folding intermediates. The proposed tentative paths of co-translational helix assembly of several polytopic proteins were qualitatively consistent with experimental studies of their folding. The assessment of free energy changes caused by mutations in TM helices demonstrated the good correlation between experimental ΔΔGfolding and calculated ΔΔGasc values for 42 mutants of bacteriorhodopsin (R2 of 0.41, RMSE of 0.81 kcal/mol) and 25 mutants of rhomboid protease (R2 of 0.43, RMSE of 0.40 kcal/mol).
Read full abstract