Limits...
Analysis of factors influencing hydration site prediction based on molecular dynamics simulations.

Yang Y, Hu B, Lill MA - J Chem Inf Model (2014)

Bottom Line: Water contributes significantly to the binding of small molecules to proteins in biochemical systems.However, questions associated with the influence of the simulation protocol on hydration site analysis remain.Here, we provide a first quantification of this effect and further indicate that similar conformations of binding site residues (RMSD < 0.5 Å) are required to obtain consistent hydration site predictions.

View Article: PubMed Central - PubMed

Affiliation: Department of Medicinal Chemistry and Molecular Pharmacology, College of Pharmacy, Purdue University , 575 Stadium Mall Drive, West Lafayette, Indiana 47907, United States.

ABSTRACT
Water contributes significantly to the binding of small molecules to proteins in biochemical systems. Molecular dynamics (MD) simulation based programs such as WaterMap and WATsite have been used to probe the locations and thermodynamic properties of hydration sites at the surface or in the binding site of proteins generating important information for structure-based drug design. However, questions associated with the influence of the simulation protocol on hydration site analysis remain. In this study, we use WATsite to investigate the influence of factors such as simulation length and variations in initial protein conformations on hydration site prediction. We find that 4 ns MD simulation is appropriate to obtain a reliable prediction of the locations and thermodynamic properties of hydration sites. In addition, hydration site prediction can be largely affected by the initial protein conformations used for MD simulations. Here, we provide a first quantification of this effect and further indicate that similar conformations of binding site residues (RMSD < 0.5 Å) are required to obtain consistent hydration site predictions.

Show MeSH
Correlationof energy values of the paired hydration sites betweenthose obtained from the 20 ns MD simulations and those obtained fromshorter simulation lengths: (A) 1, (B) 2.5, (C) 4 ns. The correlationcoefficients (R2) is calculated to theregression line with the slope =1 and zero point = 0, i.e. y = x. (left) Desolvation free energy ΔG (kcal/mol), (middle) enthalpy ΔH (kcal/mol), and (right) entropy −TΔS (kcal/mol). The distances between paired hydration sitesare color coded according to the color bar.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4210176&req=5

fig1: Correlationof energy values of the paired hydration sites betweenthose obtained from the 20 ns MD simulations and those obtained fromshorter simulation lengths: (A) 1, (B) 2.5, (C) 4 ns. The correlationcoefficients (R2) is calculated to theregression line with the slope =1 and zero point = 0, i.e. y = x. (left) Desolvation free energy ΔG (kcal/mol), (middle) enthalpy ΔH (kcal/mol), and (right) entropy −TΔS (kcal/mol). The distances between paired hydration sitesare color coded according to the color bar.

Mentions: The correlation between the energy values of different time pointsof simulations (1, 1.5, 2, 2.5, 3, 4, 5, and 10 ns), and the energyvalues of the entire 20 ns simulation were calculated. As describedin the Materials and Methods section, thepaired hydration sites between two simulations were first determined,and estimated energy values of the same hydration site were pairwisecompared. In order to study the convergence of the energy values,the Pearson correlation coefficients R2 to the regression line with slope = 1 and zero point = 0, i.e. y = x were then calculated as shown inFigure 1. The geometric distances between pairedhydration sites were color coded, ranging from red (identical position)to blue (1 Å distance). For protein PLP (PDB: 1XXO and 2AQ6), using 24 processorsthe required computation time for the three experiments (1, 2.5, and4 ns) in Figure 1 was about 12, 30, and 52h, respectively.


Analysis of factors influencing hydration site prediction based on molecular dynamics simulations.

Yang Y, Hu B, Lill MA - J Chem Inf Model (2014)

Correlationof energy values of the paired hydration sites betweenthose obtained from the 20 ns MD simulations and those obtained fromshorter simulation lengths: (A) 1, (B) 2.5, (C) 4 ns. The correlationcoefficients (R2) is calculated to theregression line with the slope =1 and zero point = 0, i.e. y = x. (left) Desolvation free energy ΔG (kcal/mol), (middle) enthalpy ΔH (kcal/mol), and (right) entropy −TΔS (kcal/mol). The distances between paired hydration sitesare color coded according to the color bar.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4210176&req=5

fig1: Correlationof energy values of the paired hydration sites betweenthose obtained from the 20 ns MD simulations and those obtained fromshorter simulation lengths: (A) 1, (B) 2.5, (C) 4 ns. The correlationcoefficients (R2) is calculated to theregression line with the slope =1 and zero point = 0, i.e. y = x. (left) Desolvation free energy ΔG (kcal/mol), (middle) enthalpy ΔH (kcal/mol), and (right) entropy −TΔS (kcal/mol). The distances between paired hydration sitesare color coded according to the color bar.
Mentions: The correlation between the energy values of different time pointsof simulations (1, 1.5, 2, 2.5, 3, 4, 5, and 10 ns), and the energyvalues of the entire 20 ns simulation were calculated. As describedin the Materials and Methods section, thepaired hydration sites between two simulations were first determined,and estimated energy values of the same hydration site were pairwisecompared. In order to study the convergence of the energy values,the Pearson correlation coefficients R2 to the regression line with slope = 1 and zero point = 0, i.e. y = x were then calculated as shown inFigure 1. The geometric distances between pairedhydration sites were color coded, ranging from red (identical position)to blue (1 Å distance). For protein PLP (PDB: 1XXO and 2AQ6), using 24 processorsthe required computation time for the three experiments (1, 2.5, and4 ns) in Figure 1 was about 12, 30, and 52h, respectively.

Bottom Line: Water contributes significantly to the binding of small molecules to proteins in biochemical systems.However, questions associated with the influence of the simulation protocol on hydration site analysis remain.Here, we provide a first quantification of this effect and further indicate that similar conformations of binding site residues (RMSD < 0.5 Å) are required to obtain consistent hydration site predictions.

View Article: PubMed Central - PubMed

Affiliation: Department of Medicinal Chemistry and Molecular Pharmacology, College of Pharmacy, Purdue University , 575 Stadium Mall Drive, West Lafayette, Indiana 47907, United States.

ABSTRACT
Water contributes significantly to the binding of small molecules to proteins in biochemical systems. Molecular dynamics (MD) simulation based programs such as WaterMap and WATsite have been used to probe the locations and thermodynamic properties of hydration sites at the surface or in the binding site of proteins generating important information for structure-based drug design. However, questions associated with the influence of the simulation protocol on hydration site analysis remain. In this study, we use WATsite to investigate the influence of factors such as simulation length and variations in initial protein conformations on hydration site prediction. We find that 4 ns MD simulation is appropriate to obtain a reliable prediction of the locations and thermodynamic properties of hydration sites. In addition, hydration site prediction can be largely affected by the initial protein conformations used for MD simulations. Here, we provide a first quantification of this effect and further indicate that similar conformations of binding site residues (RMSD < 0.5 Å) are required to obtain consistent hydration site predictions.

Show MeSH