- Email: [email protected]
Fast Calculation of Protein-Protein Binding Free Energies using Umbrella Sampling with a Coarse-Grained Model
196a
Monday, February 13, 2017
SH3 domain-spectrin repeat interface. We identify specific mutations that promote this unfolding pathway, explaining the evolutionary adaptation of the cytosolic compared to the desmosomal mechano-sensing molecule. Our results suggest a direct role of unique SH3-insertion in spectrin repeat proteins in cellular mechanotransduction, and will be validated by direct comparison to single-molecule force spectroscopy experiments carried out by our collaborators. 968-Pos Board B36 Complex Dynamics in Single Molecule Force Spectroscopy from Simple Simulation Models David De Sancho1, Robert B. Best2. 1 CIC nanoGUNE, Donostia, Spain, 2NIDDK, NIH, Bethesda, MD, USA. Single molecule force spectroscopy studies of protein folding dynamics have yielded unprecedented insight into the mechanisms of the folding reaction. Still, they usually allow for a very limited description of the process of interest (e.g. based on the reaction coordinate that is probed directly). Molecular simulations –even using very simplified models– can give a much enriched view of the underlying molecular events. In a recent example, we explain the observations on a prototypical two-state folding protein, the cold shock protein, that in the AFM turns out to exhibit multiple intermediate states. Using a very simple topology based model for the protein we show that in fact this should not come as a surprise. Indeed, force can selectively stabilize different protein intermediates and act in a very different way compared to that of chemical denaturants. 969-Pos Board B37 Dynamic Estimation of FRET Correction Factors to Study Redox Protein Interactions Halil Bayraktar, Selen Manioglu. Koc University, Istanbul, Turkey. Fluorescence resonance energy transfer (FRET) method is a key tool to determine molecular proximity and conformation changes of biomolecules in living cells. As the distance change between donor and acceptor fluorescent proteins attached to target proteins occurs in the range of 1-10 nm, an appreciable energy transfer can be detected at spatial and temporal resolution. Therefore it provides the most sensitive approach to study intercellular protein dynamics. Despite the earlier methods for characterization of FRET efficiency that accounts for donor and acceptor expression levels in cells; spectral bleedthrough due to broad emission of fluorescent proteins results in inadequate computation of FRET, therefore efficiency and distance change cannot be determined accurately. Normalized FRET (nFRET) by using correction factors, were used to extract after FRET pair establishes an equilibrium state. However, photobleaching, varying expression of FRET pair, uncorrelated dipole orientations of fluorescent proteins, heterogeneous distance changes of linkers between donor and acceptor result in incomplete computation of nFRET since correction factors were not dynamically determined during data acquisition. Thus, the use of single bleed through coefficient for computation of netFRET values may also results in overestimation of efficiency and distance. We demonstrated that donor and acceptor bleed-through values changes during the data acquisition. Therefore, netFRET cannot be computed by a single donor-acceptor correction factors. Alternatively, we presented here dynamic coefficient estimation method to determine correction factors by using the ratio of emissions from FRET channel and donor/acceptor channels for each frame after fitting pixel intensity values and later used them to compute nFRET efficiencies. The method yields a 0.38 nFRET values for Cerulean-Venus (Cer17Ven) FRET pair separated by a short linker, however the use of static coefficients of average or minimum factors yields 0.34 nFRET value. Single coefficient incorrectly reports distance change when the measurements were taken before an equilibrium state. Furthermore, method was used to measure heme induced intercellular conformation changes of Cytochrome c labeled with Cer and Ven fluorescent pairs. The distance after Cyt c folded by Heme Lyase was accurately computed. As a result, a new approach was demonstrated to determine the nFRET efficiency and its direct use to compute distance changes. Taking all values during the data acquisition eliminated the coefficient dependent variation. 970-Pos Board B38 Fast Calculation of Protein-Protein Binding Free Energies using Umbrella Sampling with a Coarse-Grained Model Jagdish Suresh Patel1, F. Marty Ytreberg2. 1 Center for Modeling Complex Interactions, University of Idaho, Moscow, ID, USA, 2Department of Physics, University of Idaho, Moscow, ID, USA. Binding free energy between two proteins determines how strongly they will interact under physiological conditions. Empirical and semi-empirical methods
for modeling and predicting protein-protein interactions are computationally inexpensive but cannot quantitatively estimate binding free energies. Full atomistic molecular simulation methods on the other hand do have this potential, but are completely unfeasible for large-scale applications due to their high computational cost. In this scenario, there is a need for new approaches, which have good balance between speed and accuracy for calculating free energy of interaction. Here, we investigate whether combining enhanced sampling molecular dynamics method i.e. umbrella sampling with coarse-grained molecular dynamics simulation is a viable approach that has good trade-off between speed and accuracy. We chose three test systems of protein-protein complexes with empirical relative binding free energy difference values (DDG) of their observed mutations reported in the literature. We selected eight different mutations occurring at the different sites with varying empirical DDG values for each test complex. We then calculated DDG values for each mutations using hybrid approach of umbrella sampling/coarse-grained simulations and compared the results with empirical DDG values and faster modeling approaches. We obtained significant improvement in the correlation compared to faster approaches with very little loss of computational speed. This hybrid approach combining speed and accuracy will allow one to carry out highthroughput study of several mutations occurring at the interface of the protein pairs. 971-Pos Board B39 Ultrafast Protein Folding through Polar Interactions in MembraneMimetic Environments Georg Krainer1,2, Andreas Hartmann1, Abhinaya Anandamurugan1, Pablo Gracia1, Sandro Keller2, Michael Schlierf1. 1 B CUBE - Center for Molecular Bioengineering, Technische Universit€at Dresden, Dresden, Germany, 2Molecular Biophysics, University of Kaiserslautern, Kaiserslautern, Germany. Proteins fold on timescales from hours down to microseconds. While first principles based on protein size, topology and differences in protein sequence have been identified as determinants of the folding speed, an important but often neglected factor in sculpting the folding landscape is the surrounding environment. This is particularly relevant for membrane-interacting proteins that require the highly complex, anisotropic environment of a lipid membrane or a membrane mimetic to fold. The chemically heterogeneous hydrophilic/ hydrophobic water-headgroup-hydrocarbon-core interface provided by such a lipidic environment presents a variety of different polar and non-polar molecular interactions. Thus, changes in the chemical composition of this surrounding may serve to modulate the folding landscape and therefore become a determining factor tuning folding barriers and stability. Here we have addressed this question and explored how properties of a membrane-mimetic environment modulate the stability and folding kinetics of a membraneinteracting protein, also in respect to folding in the absence of a hydrophobic phase in aqueous solution. By employing a single-molecule FRET approach to quantify folding and unfolding rates in detergent micelles from equilibrium measurements, we were able to analyze how contributions from polar and nonpolar interactions modulate the folding landscape of Mistic. Our results demonstrate that, although both hydrophobic and polar interactions contribute to the thermodynamic stability of Mistic, they exert opposing effects on the folding free-energy barrier: hydrophobic burial in the aliphatic core stabilizes only the folded state but not the transition state, thus slowing down unfolding without affecting folding; by contrast, polar headgroup interactions further stabilize the folded state and, additionally, lower the free-energy barrier to speed up the folding reaction to achieve ultrafast folding on timescales down to 30 ms. 972-Pos Board B40 Monte Carlo Simulations of the Effect of a Turn Sequence in an AlanineBased Peptide Shiloh M. Nold. University of Montana, Missoula, MT, USA. Alanine-based peptides are known to form single alpha helical chains. Furthermore, PSDP sequences form turns in the two-helix bundle needle proteins from type III secretion systems. To determine if the PSDP sequence would bias an alanine-based peptide toward a two helix bundle, we have carried out Monte Carlo simulations of an alanine-based peptide with and without a PSDP sequence in the middle of the peptide. Because needle proteins are 40 to 45 residues in length, we used 42 residue peptides in our simulations. The alaninebased peptide consisted of a (AAAAAK)7 sequence which we call AK42, whereas the AK42_PSDP peptide inserted a PSDP sequence centrally by replacing the fourth repeat of (AAAAAK)7 with APSDPK. We carried out Replica Exchange Monte Carlo simulations on the two peptides using the CAMPARI simulation package. A total of 16 temperatures ranging from
Monday, February 13, 2017
SH3 domain-spectrin repeat interface. We identify specific mutations that promote this unfolding pathway, explaining the evolutionary adaptation of the cytosolic compared to the desmosomal mechano-sensing molecule. Our results suggest a direct role of unique SH3-insertion in spectrin repeat proteins in cellular mechanotransduction, and will be validated by direct comparison to single-molecule force spectroscopy experiments carried out by our collaborators. 968-Pos Board B36 Complex Dynamics in Single Molecule Force Spectroscopy from Simple Simulation Models David De Sancho1, Robert B. Best2. 1 CIC nanoGUNE, Donostia, Spain, 2NIDDK, NIH, Bethesda, MD, USA. Single molecule force spectroscopy studies of protein folding dynamics have yielded unprecedented insight into the mechanisms of the folding reaction. Still, they usually allow for a very limited description of the process of interest (e.g. based on the reaction coordinate that is probed directly). Molecular simulations –even using very simplified models– can give a much enriched view of the underlying molecular events. In a recent example, we explain the observations on a prototypical two-state folding protein, the cold shock protein, that in the AFM turns out to exhibit multiple intermediate states. Using a very simple topology based model for the protein we show that in fact this should not come as a surprise. Indeed, force can selectively stabilize different protein intermediates and act in a very different way compared to that of chemical denaturants. 969-Pos Board B37 Dynamic Estimation of FRET Correction Factors to Study Redox Protein Interactions Halil Bayraktar, Selen Manioglu. Koc University, Istanbul, Turkey. Fluorescence resonance energy transfer (FRET) method is a key tool to determine molecular proximity and conformation changes of biomolecules in living cells. As the distance change between donor and acceptor fluorescent proteins attached to target proteins occurs in the range of 1-10 nm, an appreciable energy transfer can be detected at spatial and temporal resolution. Therefore it provides the most sensitive approach to study intercellular protein dynamics. Despite the earlier methods for characterization of FRET efficiency that accounts for donor and acceptor expression levels in cells; spectral bleedthrough due to broad emission of fluorescent proteins results in inadequate computation of FRET, therefore efficiency and distance change cannot be determined accurately. Normalized FRET (nFRET) by using correction factors, were used to extract after FRET pair establishes an equilibrium state. However, photobleaching, varying expression of FRET pair, uncorrelated dipole orientations of fluorescent proteins, heterogeneous distance changes of linkers between donor and acceptor result in incomplete computation of nFRET since correction factors were not dynamically determined during data acquisition. Thus, the use of single bleed through coefficient for computation of netFRET values may also results in overestimation of efficiency and distance. We demonstrated that donor and acceptor bleed-through values changes during the data acquisition. Therefore, netFRET cannot be computed by a single donor-acceptor correction factors. Alternatively, we presented here dynamic coefficient estimation method to determine correction factors by using the ratio of emissions from FRET channel and donor/acceptor channels for each frame after fitting pixel intensity values and later used them to compute nFRET efficiencies. The method yields a 0.38 nFRET values for Cerulean-Venus (Cer17Ven) FRET pair separated by a short linker, however the use of static coefficients of average or minimum factors yields 0.34 nFRET value. Single coefficient incorrectly reports distance change when the measurements were taken before an equilibrium state. Furthermore, method was used to measure heme induced intercellular conformation changes of Cytochrome c labeled with Cer and Ven fluorescent pairs. The distance after Cyt c folded by Heme Lyase was accurately computed. As a result, a new approach was demonstrated to determine the nFRET efficiency and its direct use to compute distance changes. Taking all values during the data acquisition eliminated the coefficient dependent variation. 970-Pos Board B38 Fast Calculation of Protein-Protein Binding Free Energies using Umbrella Sampling with a Coarse-Grained Model Jagdish Suresh Patel1, F. Marty Ytreberg2. 1 Center for Modeling Complex Interactions, University of Idaho, Moscow, ID, USA, 2Department of Physics, University of Idaho, Moscow, ID, USA. Binding free energy between two proteins determines how strongly they will interact under physiological conditions. Empirical and semi-empirical methods
for modeling and predicting protein-protein interactions are computationally inexpensive but cannot quantitatively estimate binding free energies. Full atomistic molecular simulation methods on the other hand do have this potential, but are completely unfeasible for large-scale applications due to their high computational cost. In this scenario, there is a need for new approaches, which have good balance between speed and accuracy for calculating free energy of interaction. Here, we investigate whether combining enhanced sampling molecular dynamics method i.e. umbrella sampling with coarse-grained molecular dynamics simulation is a viable approach that has good trade-off between speed and accuracy. We chose three test systems of protein-protein complexes with empirical relative binding free energy difference values (DDG) of their observed mutations reported in the literature. We selected eight different mutations occurring at the different sites with varying empirical DDG values for each test complex. We then calculated DDG values for each mutations using hybrid approach of umbrella sampling/coarse-grained simulations and compared the results with empirical DDG values and faster modeling approaches. We obtained significant improvement in the correlation compared to faster approaches with very little loss of computational speed. This hybrid approach combining speed and accuracy will allow one to carry out highthroughput study of several mutations occurring at the interface of the protein pairs. 971-Pos Board B39 Ultrafast Protein Folding through Polar Interactions in MembraneMimetic Environments Georg Krainer1,2, Andreas Hartmann1, Abhinaya Anandamurugan1, Pablo Gracia1, Sandro Keller2, Michael Schlierf1. 1 B CUBE - Center for Molecular Bioengineering, Technische Universit€at Dresden, Dresden, Germany, 2Molecular Biophysics, University of Kaiserslautern, Kaiserslautern, Germany. Proteins fold on timescales from hours down to microseconds. While first principles based on protein size, topology and differences in protein sequence have been identified as determinants of the folding speed, an important but often neglected factor in sculpting the folding landscape is the surrounding environment. This is particularly relevant for membrane-interacting proteins that require the highly complex, anisotropic environment of a lipid membrane or a membrane mimetic to fold. The chemically heterogeneous hydrophilic/ hydrophobic water-headgroup-hydrocarbon-core interface provided by such a lipidic environment presents a variety of different polar and non-polar molecular interactions. Thus, changes in the chemical composition of this surrounding may serve to modulate the folding landscape and therefore become a determining factor tuning folding barriers and stability. Here we have addressed this question and explored how properties of a membrane-mimetic environment modulate the stability and folding kinetics of a membraneinteracting protein, also in respect to folding in the absence of a hydrophobic phase in aqueous solution. By employing a single-molecule FRET approach to quantify folding and unfolding rates in detergent micelles from equilibrium measurements, we were able to analyze how contributions from polar and nonpolar interactions modulate the folding landscape of Mistic. Our results demonstrate that, although both hydrophobic and polar interactions contribute to the thermodynamic stability of Mistic, they exert opposing effects on the folding free-energy barrier: hydrophobic burial in the aliphatic core stabilizes only the folded state but not the transition state, thus slowing down unfolding without affecting folding; by contrast, polar headgroup interactions further stabilize the folded state and, additionally, lower the free-energy barrier to speed up the folding reaction to achieve ultrafast folding on timescales down to 30 ms. 972-Pos Board B40 Monte Carlo Simulations of the Effect of a Turn Sequence in an AlanineBased Peptide Shiloh M. Nold. University of Montana, Missoula, MT, USA. Alanine-based peptides are known to form single alpha helical chains. Furthermore, PSDP sequences form turns in the two-helix bundle needle proteins from type III secretion systems. To determine if the PSDP sequence would bias an alanine-based peptide toward a two helix bundle, we have carried out Monte Carlo simulations of an alanine-based peptide with and without a PSDP sequence in the middle of the peptide. Because needle proteins are 40 to 45 residues in length, we used 42 residue peptides in our simulations. The alaninebased peptide consisted of a (AAAAAK)7 sequence which we call AK42, whereas the AK42_PSDP peptide inserted a PSDP sequence centrally by replacing the fourth repeat of (AAAAAK)7 with APSDPK. We carried out Replica Exchange Monte Carlo simulations on the two peptides using the CAMPARI simulation package. A total of 16 temperatures ranging from