Content uploaded by Masha Y Niv
Author content
All content in this area was uploaded by Masha Y Niv
Content may be subject to copyright.
A preview of the PDF is not available
Protonation States in Molecular Dynamics Simulations of Peptide Folding and Binding
Abstract and Figures
Peptides are important signaling modules, acting both as individual hormones and as parts of larger molecules, mediating their protein-protein interactions. Many peptidic and peptidomimetic drugs have reached the marketplace and opportunities for peptide-based drug discovery are on the rise. pH-dependent behavior of peptides is well documented in the context of misfolding diseases and peptide translocation. Changes in the protonation states of peptide residues often have a crucial effect on a peptide's structure, dynamics and function, which may be exploited for biotechnological applications. The current review surveys the increasing levels of sophistication in the treatment of protonation states in computational studies involving peptides. Specifically we describe I) the common practice of assigning a single protonation state and using it throughout the dynamic simulation, II) approaches that consider multiple protonation states and compare computed observables to experimental ones, III) constant pH molecular dynamics methods that couple changes in protonation states with conformational dynamics "on the fly". Applications of conformational dynamics treatment of peptides in the context of binding, folding and interactions with the membrane are presented, illustrating the growing body of work in this field and highlighting the importance of careful handling of protonation states of peptidic residues.
Figures - uploaded by Masha Y Niv
Author content
All figure content in this area was uploaded by Masha Y Niv
Content may be subject to copyright.
... In certain protein families, such as proteases, determination of the protonation state of the amino 35 acids which comprise the active site of the enzyme, especially those which participate in its catalytic function, has become a problem of scientific interest because of its consequences for the modeling the enzymatic [1][2][3][4][5][6][7][8][9][10][11][12][13] and inhibitory [10,[14][15][16][17][18][19][20][21][22][23][24][25][26] mechanisms of the protein, affecting the quality of virtual screening of libraries of compounds [27][28][29][30][31][32][33][34]. It has been reported that the assignment of protonation states is influenced by the chemical nature of the species entering the active site of the 40 macromolecule [14,30,[35][36][37]. ...
... At the resolutions commonly employed, crystallographic techniques are incapable of determining the location of protons in the system, such that computational approaches are most often used and have been amply reported in the literature to address this problem. Among them are techniques which use docking simulations [16,[38][39][40], QM calculations [15,17,21,41,42], homodesmic reactions [22,[43][44][45], and molecular dynamics (MD) [12,23,[46][47][48][49][50][51]. 45 ...
... Nevertheless, in discussing MD results, it is common practice to use only a few snapshots which are supposed to be representative of the system over time. Consequentially, some part of the system under study is ignored [12,16,38,42,43]. To resolve 50 this, we previously invented the concept of population density [24,26], a powerful quantitative tool which permits the identification of the most stable system configuration among all possible configurations by measuring the number of occurrences of simultaneous events obtained from a set of MD snapshots over time. ...
The population density concept has emerged as a proposal for the analysis of molecular dynamics results, the key characteristic of population density is the evaluation of the simultaneous occurrence of a set of relevant parameters for a system. However, despite its statistical strength, selection of the tolerance level for the comparison of different models may appear as arbitrary. This work introduces the G-score, a function which summarizes and categorizes the results of population density analysis. Additionally, it incorporates parameters based on rmsd and dihedral angles, besides the protein-protein and protein-ligand interatomic distances conventionally used, which complement each other to provide a better description of the behavior of the system. These newly-proposed tools were applied to determine the most probable protonation state of the aspartic dyad of BACE1, Asp93 and Asp289, in the presence of three types of transition state inhibitors namely: reduced amides, tertiary carbinamines and hydroxyethylamines. The results show a full agreement between G-score values and population density charts, with the advantage of allowing a quick and direct comparison among all the considered models. We anticipate that the simplicity of calculating the parameters employed in this study will permit the extensive use of population density and the G-score for other molecular systems.
... Proton transfer is among the most common mechanisms in biomolecules (Ben-Shimon et al., 2013 andAlexov, 2013). It is a step that proves very important in several organisms (Ben-Shimon et al., 2013 andAlexov, 2013). ...
... Proton transfer is among the most common mechanisms in biomolecules (Ben-Shimon et al., 2013 andAlexov, 2013). It is a step that proves very important in several organisms (Ben-Shimon et al., 2013 andAlexov, 2013). However, modeling and simulating this phenomenon 'in silico' is extremely complex. ...
The active site of HIV protease (HIV-PR) is covered by two flaps. These flaps are known to be essential for the catalytic activity of the HIV-PR, but their exact conformations at the different stages of the enzymatic pathway remain subject to debate. Understanding the correct functional dynamics of the flaps might aid the development of new HIV-PR inhibitors. It is known that, the HIV-PR catalytic efficiency is pH-dependent, likely due to the influence of processes such as charge transfer and protonation/deprotonation of ionizable residues. Several Molecular Dynamics (MD) simulations have reported information about the HIV-PR flaps. However, in MD simulations the protonation of a residue is fixed and thus it is not possible to study the correlation between conformation and protonation state. To address this shortcoming, this work attempts to capture, through Constant pH Molecular Dynamics (CpHMD), the conformations of the apo, substrate-bound and inhibitor-bound HIV-PR, which differ drastically in their flap arrangements. The results show that the HIV-PR flaps conformations are defined by the protonation of the catalytic residues Asp25/Asp25' and that these residues are sensitive to pH changes. This study suggests that the catalytic aspartates can modulate the opening of the active site and substrate binding.
... The explicit representation of all absent hydrogen atoms makes the processed protein structures suitable for simulation with all-atom force fields, while the explicit representation of polar hydrogen atoms alone makes the proteins suitable for simulations with united-atom force fields. The selection of appropriate protonation states is pertinent to performing simulations that accurately reproduce protein-lipid interactions given that the protonation states of peptide residues can impact their structure, dynamics and function [73]. ...
Membrane proteins are amphipathic macromolecules whose exposed hydrophobic surfaces promote interactions with lipid membranes. Membrane proteins are remarkably diverse in terms of chemical composition and correspondingly, their biological functions and general biophysical behavior. Conventional experimental techniques provide an approach to study specific properties of membrane proteins e.g. their surface features, the nature and abundance of stabilizing intramolecular forces, preferred bilayer orientation, and the characteristics of their annular lipid shells. Molecular modeling software-and in particular, the suite of molecular dynamics algorithms-enables a more comprehensive exploration of dynamic membrane protein behavior. Molecular dynamics methods enable users to produce stepwise trajectories of proteins on arbitrary spatiotemporal scales that enable the easy identification of dynamic interactions that are beyond the scope of conventional analytical techniques. This article explains the molecular dynamics theoretical framework and popular step-by-step approaches for simulating membrane proteins in planar, and to a lesser extent, nonplanar lipid geometries. We detail popular procedures and computational tools that produce well-packed configurations of lipids and proteins and additionally, the efficient molecular dynamics simulation algorithms that reproduce their dynamic interactions.
... Another challenge is that until recently only limited data were available, especially considering the need to analyze enough high-resolution and diverse protein−ligand complexes. Yet, a better characterization of the interacting environment of ionizable groups would be of key interest in molecular docking simulations, 20 where such a knowledge would help to better position the bridging structural water molecules, select or optimize relevant ionization states, improve the initial placing of the ligand, and design more efficient and accurate scoring functions. 21−24 The aim of this study is to make a quantitative and qualitative assessment of the protein molecular environments for the ligand and protein ionizable groups in the PDB. ...
We conduct a statistical analysis of the molecular environment of common ionizable functional groups in both protein−ligand complexes and inside proteins from the Protein Data Bank (PDB). In particular, we characterize the frequency, type, and density of the interacting atoms as well as the presence of a potential counterion. We found that for ligands, most guanidinium groups, half of primary and secondary amines, and one-fourth of imidazole neighbor a carboxylate group. Tertiary amines bind more rarely near carboxylate groups, which may be explained by a crowded neighborhood and hydrophobic character. In comparison to the environment seen by the ligands, inside proteins, an environment enriched in main-chain atoms is found, and the prevalence of direct charge neutralization by carboxylate groups is different. When the ionizable character of water molecules and phenolic or hydroxyl groups is accounted, considering a high-resolution dataset (less than 1.5 Å), charge neutralization could occur for well above 80% of the ligand functional groups considered, but for tertiary amines.
... for structure and topology representations, respectively. Protonation state of peptide histidines can be defined at this stage [37]. For case 2 peptide, the δ-nitrogen protonation (HID) was used. ...
Due to increasing interest in peptides as signaling modulators and drug candidates, several methods for peptide docking to their target proteins are under active development. The “blind” docking problem, where the peptide-binding site on the protein surface is unknown, presents one of the current challenges in the field. AnchorDock protocol was developed by Ben-Shimon and Niv to address this challenge.
This protocol narrows the docking search to the most relevant parts of the conformational space. This is achieved by pre-folding the free peptide and by computationally detecting anchoring spots on the surface of the unbound protein. Multiple flexible simulated annealing molecular dynamics (SAMD) simulations are subsequently carried out, starting from pre-folded peptide conformations, constrained to the various precomputed anchoring spots.
Here, AnchorDock is demonstrated using two known protein-peptide complexes. A PDZ-peptide complex provides a relatively easy case due to the relatively small size of the protein, and a typical peptide conformation and binding region; a more challenging example is a complex between USP7N-term and a p53-derived peptide, where the protein is larger, and the peptide conformation and a binding site are generally assumed to be unknown. AnchorDock returned native-like solutions ranked first and third for the PDZ and USP7 complexes, respectively. We describe the procedure step by step and discuss possible modifications where applicable.
Enzyme engineering aims to improve or install a new function in biocatalysts for applications ranging from chemical synthesis to biomedicine. For decades, computational techniques have been developed to predict the effect of protein changes and design new enzymes. However, these techniques may have been optimized to deal with proteins composed of the standard amino acid alphabet, while the function of many enzymes relies on non‐proteogenic parts like cofactors, nucleic acids, and post‐translational modifications. Enzyme systems containing such molecules might be handled or modeled improperly by computational tools, and thus be unsuitable, or require additional tweaking, parameterization, or preparation. In this review, we give an overview of common and recent tools and workflows available to computational enzyme engineers. We highlight the various pitfalls that come with including non‐proteogenic compounds in computations and outline potential ways to address common issues. Finally, we showcase successful examples from the literature that computationally engineered such enzymes.
In most instances, the usual fastness of protein unfolding events hinders determining changes in secondary structures associated with this process because these determinations rely on the recording of high-resolution circular dichroism (CD) spectra. In this work, far-UV CD spectra, recorded at ten-minute intervals, were used to evaluate the time course followed by four classes of secondary structures in the slow temperature-induced unfolding of yeast triosephosphate isomerase (yTIM) under distinct pH conditions. CONTIN-LL and SELCON3 algorithms were used for the deconvolution of spectra. Both algorithms furnished helix and unordered structure contents that changed according to first-order kinetics, agreeing with the behavior shown by CD data at specific wavelengths. Analyses of unfolded yTIM spectra, using a dataset that includes spectra of unfolded proteins and either one of the two algorithms, clearly showed a more unordered protein structure at high pH; this finding was corroborated with analysis of the difference spectra. Molecular dynamics (MD) simulations performed with AMBER and OPLS force fields resulted in more extensive loss of helices and gain in coils at high pH, in agreement with spectroscopic results. However, structural differences between low- and high-pH unfolded yTIM were relatively small. Comparison of results from CD and MD thus point to the need of fine-tuning of MD procedures.
Diseases are often caused by mutant proteins. Many drugs have limited effectiveness and/or toxic side effects because of a failure to selectively target the disease-causing mutant variant, rather than the functional wild type protein. Otherwise, the drugs may even target different proteins with similar structural features. Designing drugs that successfully target mutant proteins selectively represents a major challenge. Decades of cancer research have led to an abundance of potential therapeutic targets, often touted to be “master regulators”. For many of these proteins, there are no FDA-approved drugs available; for others, off-target effects result in dose-limiting toxicity. Cancer-related proteins are an excellent medium to carry the story of mutant-specific targeting, as the disease is both initiated by and sustained by mutant proteins; furthermore, current chemotherapies generally fail at adequate selective distinction. This review discusses some of the challenges associated with selective targeting from a structural biology perspective, as well as some of the developments in algorithm approach and computational workflow that can be applied to address those issues. One of the most widely researched proteins in cancer biology is p53, a tumor suppressor. Here, p53 is discussed as a specific example of a challenging target, with contemporary drugs and methodologies used as examples of burgeoning successes. The oncogene KRAS, which has been described as “undruggable”, is another extensively investigated protein in cancer biology. This review also examines KRAS to exemplify progress made towards selective targeting of disease-causing mutant proteins. Finally, possible future directions relevant to the topic are discussed.
Le transfert d'un analogue de lysine au travers de membranes de phospholipides a été étudié. Des simulations de dynamique moléculaire et des calculs d'énergie libre ont été conduits afin d'étudier l'évolution du pKa de cette molécule en fonction de sa position à l'intérieur de la membrane. Des grandeurs cinétiques telles que la perméabilité et les constantes de réaction associées au transfert ont montré que ce processus est susceptible de ce produire sur des échelles de temps appartenant à la ms. La comparaison de ces grandeurs dans des membranes constituées d'étherlipides et de lipides avec des chaînes branchées a été faite par la suite. Les études ont montré une diminution de la perméabilité et une augmentation du temps de passage dans des membranes faites d'étherlipides. L'ajout de méthyles le long des chaînes carbonées augmentent également le temps de passage et la perméabilité mais de manière moins importante. Cependant, le pKa indique que la forme chargée ne peut être retrouvée aussi profondément que dans une membrane constituée de chaînes linéaires. Finalement, le transfert du peptide a été étudié en estimant des surfaces d'énergie libre multidimensionnelles. La coordination de l'amine avec les molécules d'eau dans sa première sphère d'hydratation et la projection de la distance par rapport au centre de la membrane a été étudié. De nouvelles possibilités pour le changement d'état de protonation du peptide sont apparues. Contrairement au cas 1D, la forme neutre peut apparaître déjà dans la région des têtes polaires pour une certaine valeur de la coordination. Ces valeurs, inférieures à celles du milieux aqueux, peuvent être atteintes à l'interface d'après les surfaces déterminées
Protein internal ionizable groups can exhibit large shifts in pKa values. Although the environment and interaction changes have been extensively studied both experimentally and computationally, direct calculation of pKa values of these internal ionizable groups in explicit water is challenging due to energy barriers in solvent interaction and in conformational transition. The virtual mixture of multiple states (VMMS) method is a new approach designed to study chemical state equilibrium. This method constructs a virtual mixture of multiple chemical states in order to sample the conformational space of all states simultaneously and to avoid crossing energy barriers related to state transition. By applying VMMS to 25 variants of staphylococcal nuclease with lysine residues at internal positions, we obtained the pKa values of these lysine residues and investigated the physics underlining the pKa shifts. Our calculation results agree reasonably well with experimental measurements, validating the VMMS method for pKa calculation and providing molecular details of the protonation equilibrium for protein internal ionizable groups. Based on our analyses of protein conformation relaxation, lysine side chain flexibility, water penetration, and the microenvironment, we conclude that the hydrophobicity of the microenvironment around the lysine side chain (which affects water penetration differently for different protonation states) plays an important role in the pKa shifts.
A method is developed for performing classical explicit-solvent molecular dynamics (MD) simulations at constant pH, where the protonation state of each ionizable (titratable) group in a simulated compound is allowed to fluctuate in time, depending on the instantaneous system configuration and the imposed pH. In this method, each ionizable group is treated as a mixed state, i.e., the interaction-function parameters for the group are a linear combination of those of the protonated state and those of the deprotonated state. Free protons are not handled explicitly. Instead, the extent of deprotonation of each group is relaxed towards its equilibrium value by weak coupling to a "proton bath." The method relies on precalibrated empirical functions, one for each type of ionizable group present in the simulated compound, which are obtained through multiple MD simulations of monofunctional model compounds. In this study, the method is described in detail and its application illustrated by a series of constant-pH MD simulations of small monofunctional amines. In particular, we investigate the influence of the relaxation time used in the weak-coupling scheme, the choice of appropriate model compounds for the calibration of the required empirical functions, and corrections for finite-size effects linked with the small size of the simulation box. (C) 2001 American Institute of Physics.
The acidostat method previously developed for performing explicit-solvent molecular dynamics simulations at constant pH (J. Chem. Phys. 2001, 114, 9706) is applied to polyfunctional compounds, namely 1,4-diaminobutane and a decalysine peptide. The titration behavior of 1,4-diaminobutane is investigated by performing a series of simulations at different pH, using the acidostat method. The method accounts at least to some extent for site-site coupling and reproduces the experimental pK(a) values of the compound within half a pK unit, although the simulations reveal insufficient sampling of the protonation- state variables. In a second step, the ability of the acidostat method to account for correlations between the solution pH and the structure and dynamics of a biomolecule is tested by studying the pH-dependent stability of an a.-helical decalysine peptide. To this end, four 32-ns constant-pH simulations at different pH values are performed. The results are compared to those of standard molecular dynamics simulations of a fully protonated or a fully deprotonated peptide, and to experimental data on (comparatively longer) polylysine peptides. In agreement with experiment, the peptide predominantly remains in an a.-helical conformation under high-pH conditions, but becomes disordered under low-pH conditions. The helix-coil transition pH for the peptide is found to be between 9.5 and 10.3, in good agreement with the experimental value for polylysine (10.3). The constant-pH simulations also evidence a correlation between the protonation of specific lysine side chains and the local loss of backbone hydrogen bonds and partial peptide unfolding, both effects occurring predominantly in the C-terminal region of the peptide.
Protein-protein interactions (PPIs) play a central and crucial role in almost every cellular process. Understanding the structural basis of protein-protein interactions can lead to the development of new drugs for treatment of various diseases. With this purpose, peptide-based drug design (PBDD) has been extensively explored in the last few decades. Peptidomimetics are compounds which mimic the biological activity of peptides while offering the advantages of improving their pharmacokinetics profiles. In this review, we would like to summarize the state of the art of computational methods which have been recently introduced to design novel peptidomimetics involved in a therapeutically relevant protein-protein recognition processes.
The three-dimensional solution structure of residues 1-28 of the amyloid beta-peptide was determined using nuclear magnetic resonance spectroscopy, distance geometry, and molecular dynamic techniques. The nuclear magnetic resonance data used to derive the structure consisted of nuclear Overhauser enhancements, vicinal coupling constants, and temperature coefficients of the amide-NH chemical shifts. The beta-peptide is the major proteinaceous component of amyloid deposits in Alzheimer's disease. In membranelike media, the peptide folds to form a predominately alpha-helical structure with a bend centered at residue 12. The side chains of histidine-13 and lysine-16 are close, residing on the same face of the helix. Their proximity may constitute a binding motif with the heparan sulfate proteoglycans. The molecular details of the structure shown here could facilitate the design of rational treatments to curtail the binding of heparan sulfate proteoglycans or to prevent an alpha-helix-->beta-sheet conversion that may occur during the early stages of amyloid formation in Alzheimer's disease.
Beta-peptide is a major component of amyloid deposits in Alzheimer's disease. We report here a proton nuclear magnetic resonance (NMR) spectroscopic investigation of a synthetic peptide that is homologous to residues 1-28 of beta-peptide [abbreviated as beta-(1-28)]. The beta-(1-28) peptide produces insoluble beta-pleated sheet structures in vitro, similar to the beta-pleated sheet structures of beta-peptide in amyloid deposits in vivo. For peptide solutions in the millimolar range, in aqueous solution at pH 1-4 the beta-(1-28) peptide adopts a monomeric random coil structure, and at pH 4-7 the peptide rapidly precipitates from solution as an oligomeric beta-sheet structure, analogous to amyloid deposition in vivo. The NMR work shown here demonstrates that the beta-(1-28) peptide can adopt a monomeric alpha-helical conformation in aqueous trifluoroethanol solution at pH 1-4. Assignment of the complete proton NMR spectrum and the determination of the secondary structure were arrived at from interpretation of two-dimensional (2D) NMR data, primarily (1) nuclear Overhauser enhancement (NOE), (2) vicinal coupling constants between the amide (NH) and alpha-H protons, and (3) temperature coefficients of the NH chemical shifts. The results show that at pH 1.0 and 10-degrees-C the beta-(1-28) peptide adopts an alpha-helical structure that spans the entire primary sequence. With increasing temperature and pH, the alpha-helix unfolds to produce two alpha-helical segments from Ala2 to Asp7 and Tyr10 to Asn27. Further increases in temperature to 35-degrees-C cause the Ala2-Asp7 section to become random coil, while the His13-Phe20 section stays alpha-helical. A mechanism involving unfavorable interactions between charged groups and the alpha-helix macrodipole is proposed for the alpha-helix --> beta-sheet conversion observed at midrange pH.
The photodissociation of HCl embedded in argon clusters is studied by semiclassical molecular dynamics, based on a surface-hopping approach for the non-adiabatic transitions. The diatomics-in-molecules (DIM) method is used to construct the 12 electronic potential energy surfaces that are involved, and the non-adiabatic couplings. Caging effects, including recombination and electronic relaxation are investigated for Ar12(HCl) and Ar54(HCl), corresponding respectively to one and two complete solvation layers. The effects of the process on the cluster, e.g. fragmentation and structural deformation, are also studied. The main findings are: (1) non-adiabatic transitions play a major role in the dynamics for both clusters; (2) some recombination occurs in Ar12(HCl), and it is much greater, about 7%, in Ar54(HCl); (3) all 12 electronic states are visited, at least to some extent, in the process, but the distributions remain non-statistical throughout in both systems; (4) rates of spin-forbidden transitions are roughly of similar magnitudes to those of spin-allowed transitions between electronic states; (5) the energy gap law of radiationless relaxation theory does not work well for these systems. Symmetry and shape of the electronic states greatly affect the relaxation rates; (6) the clusters undergo melting and extensive evaporation in the processes.
In astrophysical sources, TeV gamma-rays may originate by means of two well- know mechanisms: via electromagnetic or hadronic processes. In this talk we discuss a third mechanism: the photodisintegration of nuclei at the source, followed by de-excitation of the daughter nuclei. We examine the conditions that need to be satised in the source so that a relevant contribution to the TeV gamma-ray ux may come from this mechanism and we also present the distinctive features of this dynamical framework, as is the absence of low energy counterparts. We also comment on the concomitant associated ux of antineutrinos coming from the -decay of stripped neutrons, which turns out to be smaller than that of gamma-rays.
A new method for performing constant pH molecular dynamics (pHMD) simulations utilizing a generalized Born implicit solvent model and continous titration coordinates will be discussed. Test calculations on calculating pKa's of several proteins will be presented, and simulations of polypeptides and proteins at different pH's will be discussed. The couplings between pH,pKa's and conformations will be emphasized.
Methods for estimation of pKa values of residues in proteins were tested on a set of benchmark proteins with experimentally known pKa values. The benchmark set includes 80 different residues (20 each for Asp, Glu, Lys, and His), half of which consists of significantly variant cases (ΔpKa ≥ 1 pKa unit from the amino acid in solution). The method introduced by Case and co-workers [J. Am. Chem. Soc. 2004, 126, 4167−4180], referred to as the molecular dynamics/generalized-Born/thermodynamic integration (MD/GB/TI) technique, gives a root-mean-square deviation (rmsd) of 1.4 pKa units on the benchmark set. The use of explicit waters in the immediate region surrounding the residue was shown to generally reduce high errors for this method. Longer simulation time was also shown to increase the accuracy of this method. The empirical approach developed by Jensen and co-workers [Proteins 2005, 61, 704−721], PROPKA, also gives an overall rmsd of 1.4 pKa units and is more or less accurate based on residue typethe method does very well for Lys and Glu, but less so for Asp and His. Likewise, the absolute deviation is quite similar for the two methods5.2 for PROPKA and 5.1 for MD/GB/TI. A comparison of these results with several prediction methods from the literature is presented. The error in pKa prediction is analyzed as a function of variation of the pKa from that in water and the solvent accessible surface area (SASA) of the residue. A case study of the catalytic lysine residue in 2-deoxyribose-5-phosphate aldolase (DERA) is also presented.