Soluble State of Villin Headpiece Protein as a Tool in the Assessment of MD Force Fields

Abstract

Protein self-assembly plays an important role in cellular processes. Whereas molecular dynamics (MD) represents a powerful tool in studying assembly mechanisms, its predictions depend on the accuracy of underlying force fields, which are known to overly promote protein assembly. We here examine villin headpiece domain, HP36, which remains soluble at concentrations amenable to MD studies. The experimental characterization of soluble HP36 at concentrations of 0.05 to 1 mM reveals concentrationindependent 90% monomeric and 10% dimeric populations. Extensive allatom MD simulations at two protein concentrations, 0.9 and 8.5 mM, probe the HP36 dimer population, stability, and kinetics of dimer formation within two MD force fields, Amber ff14SB and CHARMM36m. MD results demonstrate that whereas CHARMM36m captures experimental HP36 monomer populations at the lower concentration, both force fields overly promote HP36 association at the higher concentration. Moreover, contacts stabilizing HP36 dimers are force-field-dependent. CHARMM36m produces consistently higher HP36 monomer populations, lower association rates, and weaker dependence of these quantities on the protein concentration than Amber ff14SB. Nonetheless, the highest monomer populations and dissociation constants are observed when the TIP3P water model in Amber ff14SB is replaced by TIP4P/2005, showcasing the critical role of the water model in addressing the protein solubility problem in MD.

Conformational Manifold Sampled by Two Short Linear Motif Segments Probed by Circular Dichroism, Vibrational, and Nuclear Magnetic Resonance Spectroscopy

“…Andrews et al also showed that Amber ff14SB performs better when one replaces TIP3P with TIP4P-2005 – Overall, our prior work demonstrated that CHARMM36m performs better than OPLS-AA/M and Amber ff14SB in several aspects of conformational dynamics. − Consistent with these observation, we applied CHARMM36m simulations to examine conformational ensembles of polyarginine peptides . It is also worth noting that CHARMM36m is routinely used to examine dynamics of one of the most notorious intrinsically disordered proteins, amyloid β-protein (Aβ), and its interactions with lipids , because it outperforms other MD force fields when modeling peptide self-assembly …”

How do salt and lipids affect conformational dynamics of Aβ42 monomers in water?

“…We here examine the effect of salt and DMPC lipids on fully atomistic Aβ42 monomer conformations using CHARMM36m and its innate TIP3P water model, which was shown to perform optimally based on comparison to available spectroscopic data on short unfolded peptides when compared to other modern force fields. 31,33–35,67…”

“…We here examine the effect of salt and DMPC lipids on fully atomistic Ab42 monomer conformations using CHARMM36m and its innate TIP3P water model, which was shown to perform optimally based on comparison to available spectroscopic data on short unfolded peptides when compared to other modern force fields. 31,[33][34][35]67 We acquired six sets of MD trajectories, following the protocol described in Methods. Sets (i)-(iii) correspond Ab42 monomer in pure water, water with 12 DMPC lipid molecules (27 mM), and water with 48 DMPC lipid molecules (109 mM), respectively.…”

“…27 The strategy of using a large number of replicas in MD simulations to improve the phase space sampling was used by some of us in earlier studies of Ab40 and Ab42 monomers, dimers, trimers, tetrames, and pentamers using OPLS-AA/L with TIP3P water. 28,29 In this paper, we employ CHARMM36m 30 which was shown to be one of the most reliable MD force fields for modeling short unfolded peptides, [31][32][33][34] although lacking amino acid specificity in reproducing experimentally-derived Ramachandran distributions of residues in water. 35 In vivo, Ab42 exerts its activity in the presence of physiological levels of salt and other biomolecules.…”

Comparing the influence of explicit and implicit solvation models on site‐specific thermodynamic stability of proteins

“…25,26 The proteins of interest were representative β-sheet and α-helix proteins, the WW domain and HP-36. These short proteins have often been the subject of protein folding simulations that use explicit 24,[27][28][29][30][31][32][33][34] and implicit solvents. 7,[35][36][37][38] Subsequently, we identified the critical residues that lead to the differences in solvation models using a recently modified sitedirected thermodynamic analysis method that resolves the solvation free energy G solv and the gas-phase potential energy E u into contributions from the main and side chains.…”