No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Tools for building a comprehensive modeling system for virtual screening under real biological conditions: The Computational Titration algorithm
Abstract
Computational tools utilizing a unique empirical modeling system based on the hydrophobic effect and the measurement of logP(o/w) (the partition coefficient for solvent transfer between 1-octanol and water) are described. The associated force field, Hydropathic INTeractions (HINT), contains much rich information about non-covalent interactions in the biological environment because of its basis in an experiment that measures interactions in solution. HINT is shown to be the core of an evolving virtual screening system that is capable of taking into account a number of factors often ignored such as entropy, effects of solvent molecules at the active site, and the ionization states of acidic and basic residues and ligand functional groups. The outline of a comprehensive modeling system for virtual screening that incorporates these features is described. In addition, a detailed description of the Computational Titration algorithm is provided. As an example, three complexes of dihydrofolate reductase (DHFR) are analyzed with our system and these results are compared with the experimental free energies of binding.
... HINT Computational Titration (CT) [29] [148] [149] [150] ...
... The HINT model views all biomolecular interactions empirically and we postulate that the water and 1-octanol solvents, respectively, are representative of polar and hydrophobic regions in biomacromolecules. A long list of studies using HINT have shown that the HINT score correlates very well with experimental free energy of binding and/or give valuable insight into the biomolecular system [24] [29] [148] [149] [150] [235] [236] [237] [238] [239] [240] [241] [242]. ...
... [29] [148] [149] [150] [257] ...
... The advent of fast computed tomography (CT) scanners in the 1990's, together with the development of sophisticated post-processing software, has made PCT a powerful tool for investigating pathophysiological processes in the human body. 25 In vivo evaluation and quantitative analysis of brain perfusion by means of PCT has had considerable impact on patient care in the settings of severe head trauma, acute stroke, and cerebral tumors. [26][27][28][29][30] The determination of tissue perfusion by PCT involves the intravenous injection of tracer and subsequential imaging to monitor the concentration of tracer in the tissue and a feeding artery as functions of time. ...
... 27 One important advantage of CT is that the enhancement is linearly proportional to the concentration of tracer in the tissue. 25 Serial CT scans start before the contrast agent arrives to determine the baseline and repeated scans are acquired until the tracer leaves the tissues. Subtraction of the baseline from each of the serial CT scans after the arrival of the contrast agent at the tissue gives the time-density curve (TDC) of the tissue. ...
... Subtraction of the baseline from each of the serial CT scans after the arrival of the contrast agent at the tissue gives the time-density curve (TDC) of the tissue. 25,31 All the physiological information is obtained by mathematical analysis of the tissue TDC. These analyses are based on proposed 'tracer kinetics' models that describe the distribution of contrast in blood vessels and extravascular space of the tissue. ...
Background.
The blood-brain barrier represents the selective diffusion barrier at the level of the cerebral microvascular endothelium. Other functions of blood-brain barrier include transport, signaling and osmoregulation. Endothelial cells interact with surrounding astrocytes, pericytes and neurons. These interactions are crucial to the development, structural integrity and function of the cerebral microvascular endothelium. Dysfunctional blood-brain barrier has been associated with pathologies such as acute stroke, tumors, inflammatory and neurodegenerative diseases.
Conclusions.
Blood-brain barrier permeability can be evaluated in vivo by perfusion computed tomography - an efficient diagnostic method that involves the sequential acquisition of tomographic images during the intravenous administration of iodinated contrast material. The major clinical applications of perfusion computed tomography are in acute stroke and in brain tumor imaging.
... On a published set of 698 monoprotic compounds that were divided into fifteen different classes, this method, when compared to linear Gaussian regression models, performed best in 85% of all the experiments [147]. [29,148149150 is a protocol developed by Kellogg, Cozzini, Mozzarelli and coworkers that exhaustively explores the manifold of ionization state combinations available to a protein-ligand complex by building the full set of models, optimizes the rotatable hydrogen bonds (i.e., in –OH, -NH 2 , etc.) and scores each model. The concept of isocrystallographic was introduced [29] to represent the notion that many molecular models including hydrogens can be fit into the electron density envelope provided by an X-ray crystallographic experiment, which, as noted above, does not generally observe the hydrogens. ...
... The HINT model views all biomolecular interactions empirically and we postulate that the water and 1-octanol solvents, respectively, are representative of polar and hydrophobic regions in biomacromolecules. A long list of studies using HINT have shown that the HINT score correlates very well with experimental free energy of binding and/or give valuable insight into the biomolecular system [24, 29,148149150235236237238239240241242. We suspected, also, that this information is also potentially valuable in macromolecular X-ray crystallography, especially in cases where the data resolution is poor (e.g., > 3.0 Å). ...
... With our CT algorithm [29,148149150 257], we have illustrated the multiplicity of states due to ionizations, have incorporated the energetic contributions of Relevant water molecules, and suggested that a Boltzmann-like treatment of the ensemble may lead to more realistic interaction energy estimates. This algorithm enumerates the potential combinations of ionization states for a complex or interface, but is subject to combinatorial explosion. ...
The unique physicochemical properties of water make it the most important molecule for life. Water molecules have many roles, direct and indirect, related to both biological structure and function. This paper: 1) reviews tools for the prediction of water conservation in and around protein active sites, by empirical (knowledge-based) algorithms and by methods based on thermodynamics principles; 2) reviews principles and approaches to predict pKa for both protein residue ensembles and for ligands; and 3) discusses the HINT biomolecular interaction model and forcefield based on experimental measurements of LogPo/w, the 1-octanol/water partition coefficient, which implicitly incorporates all solution phenomena like these, and others like tautomerism and entropy. Lastly, it must be considered that the "real" biological environment is a continuum of nano-states and it may not be possible to represent it as a single discrete all-atom model.
... Computational titration is an application based on HINT that optimizes the placement and orientation of protons so as to have an idealized interaction between the "ligand" and the "receptor." This idealized interaction may include protonation or deprotonation of basic and acidic residues, and/or ligand functional groups, respectively [55][56][57][58]. For computational titration the protein model was separated into two parts: tyrosine as the pseudo-ligand and the rest of the protein as the pseudo-receptor. ...
... Having a degree of ionization state flexibility, i.e., acidic and basic residues, in the region surrounding the tyrosine would seem to provide alternative mechanisms for the tyrosine to redistribute charge and/or for radical species to return to their ground state. The descriptor we have used to describe this environment is called multiplicity_of_states and is the number of protonation state ensembles considered by computational titration [55][56][57][58][59] for the tyrosine and its neighborhood. The higher this number, the more acids, bases, and/or alternate ionization state schemes are available for the region. ...
... The higher this number, the more acids, bases, and/or alternate ionization state schemes are available for the region. For counting ionization states, tyrosine/tyrosinate counts as 2; an ammonium/amine, e.g., lysine, counts as 2; a carboxylate/carboxylic acid, e.g., aspartic and glutamic acid, counts as 3 (ionized, or protonated at either oxygen); etc. [55][56][57][58][59]. The total number of available states is the product of these. ...
Models for exploring tyrosine nitration in proteins have been created based on 3D structural features of 20 proteins for which high-resolution X-ray crystallographic or NMR data are available and for which nitration of 35 total tyrosines has been experimentally proven under oxidative stress. Factors suggested in previous work to enhance nitration were examined with quantitative structural descriptors. The role of neighboring acidic and basic residues is complex: for the majority of tyrosines that are nitrated the distance to the heteroatom of the closest charged side chain corresponds to the distance needed for suspected nitrating species to form hydrogen bond bridges between the tyrosine and that charged amino acid. This suggests that such bridges play a very important role in tyrosine nitration. Nitration is generally hindered for tyrosines that are buried and for those tyrosines for which there is insufficient space for the nitro group. For in vitro nitration, closed environments with nearby heteroatoms or unsaturated centers that can stabilize radicals are somewhat favored. Four quantitative structure-based models, depending on the conditions of nitration, have been developed for predicting site-specific tyrosine nitration. The best model, relevant for both in vitro and in vivo cases, predicts 30 of 35 tyrosine nitrations (positive predictive value) and has a sensitivity of 60/71 (11 false positives).
... For example, in a typical protein-ligand interaction, there will likely be only a few relevant water molecules, but at a protein-protein interaction surface, there can be one or two dozen. Using the aforementioned computational titration algorithm (Fornabaio et al., 2003;Spyrakis et al., 2004;Kellogg et al., 2006) would typically involve 2-3 ionizable residues or functional groups in a protein-ligand complex, but again, at a protein-protein surface, this could be a much larger number. ...
A long-lasting goal of computational biochemists, medicinal chemists, and structural biologists has been the development of tools capable of deciphering the molecule–molecule interaction code that produces a rich variety of complex biomolecular assemblies comprised of the many different simple and biological molecules of life: water, small metabolites, cofactors, substrates, proteins, DNAs, and RNAs. Software applications that can mimic the interactions amongst all of these species, taking account of the laws of thermodynamics, would help gain information for understanding qualitatively and quantitatively key determinants contributing to the energetics of the bimolecular recognition process. This, in turn, would allow the design of novel compounds that might bind at the intermolecular interface by either preventing or reinforcing the recognition. HINT, hydropathic interaction, was a model and software code developed from a deceptively simple idea of Donald Abraham with the close collaboration with Glen Kellogg at Virginia Commonwealth University. HINT is based on a function that scores atom–atom interaction using LogP, the partition coefficient of any molecule between two phases; here, the solvents are water that mimics the cytoplasm milieu and octanol that mimics the protein internal hydropathic environment. This review summarizes the results of the extensive and successful collaboration between Abraham and Kellogg at VCU and the group at the University of Parma for testing HINT in a variety of different biomolecular interactions, from proteins with ligands to proteins with DNA.
... In normal cerebral vasculature, PS is negligible for all contrast agents presently in use. The relative magnitude of PS and F also determines E. [27,28] In summary, E is the ratio of contrast leaving a voxel, as opposed to measures such as CBF or CBV which measure contrast inflow. ...
Objective
We sought to examine the diagnostic utility of existing predictors of any haemorrhagic transformation (HT) and compare them to novel perfusion imaging permeability measures in ischemic stroke patients receiving alteplase only.
Methods
A pixel‐based analysis of pre‐treatment CT perfusion (CTP) was undertaken to define the optimum CTP permeability thresholds to predict the likelihood of HT. We then compared previously proposed predictors of HT using regression analyses and receiver operator characteristic curve analysis to produce an Area Under the Cure (AUC), and compared AUCs using Chi Square analysis.
Results
From 5 centres, 1407 patients were included in this study, 282 had HT. The cohort was split into a derivation (1025, 70% patients) and validation cohort (382 patients or 30%). The E permeability map at a threshold of 30% relative to contralateral had the highest AUC at predicting any HT (derivation AUC 0.85, 95% CI, 0.79‐0.91, validation AUC 0.84, 95% CI, 0.77‐0.91). The AUC improved when permeability was assessed within the acute perfusion lesion for the E maps at a threshold of 30% (derivation AUC 0.91, 95% CI, 0.86‐0.95, validation AUC 0.89, 95% CI, 0.86‐0.95). Previously proposed associations with HT and PH showed lower AUC values than the permeability measure.
Interpretation
In this large multi‐centre study, we have validated a highly accurate measure of HT prediction. This measure may be useful in clinical practice to predict haemorrhagic transformation in ischemic stroke patients before receiving alteplase alone.
This article is protected by copyright. All rights reserved.
... There is recent progress on this front, however. New algorithms such as the computational titration protocol implemented in Hydropathic Interaction (HINT) seek to identify and optimize all possible protonation states so that rational models with atomic details can be constructed and applied to model ligand-binding energetic [26,30,31]. By modeling all ionizable residues in the binding pocket, and calculating all the possible protonation states of residues and functional groups within the active site, the computational-titration methodology realistically samples the dynamic behavior of labile H-atoms in the active site microenvironment. ...
Accurate modeling of protein ligand binding is an important step in structure-based drug design, is a useful starting point for finding new lead compounds or drug candidates. The 'Lock and Key' concept of protein-ligand binding has dominated descriptions of these interactions, and has been effectively translated to computational molecular docking approaches. In turn, molecular docking can reveal key elements in protein-ligand interactions-thereby enabling design of potent small molecule inhibitors directed against specific targets. However, accurate predictions of binding pose and energetic remain challenging problems. The last decade has witnessed more sophisticated molecular docking approaches to modeling protein-ligand binding and energetics. However, the complexities that confront accurate modeling of binding phenomena remain formidable. Subtle recognition and discrimination patterns governed by three-dimensional features and microenvironments of the active site play vital roles in consolidating the key intermolecular interactions that mediates ligand binding. Herein, we briefly review contemporary approaches and suggest that future approaches treat protein-ligand docking problems in the context of a 'combination lock' system.
... Second, the water/polar network is responsive to changes in ionization state: any neighboring water molecules need to be optimized for each of the models conceived during a CT run. Third, while the protocol we described here and elsewhere [42,[61][62][63] is based on the HINT force field, the concepts and strategy could be simply translated to another paradigm. ...
The importance of water molecules in biological interactions is not debatable, but the various diverse and specific roles that water can play are not as well understood on a molecular scale. In this methods report, the theoretical basis for a computational framework that focuses on water is described. The framework is HINT (for Hydropathic INTeractions) and is a related series of algorithms and methods for probing and modeling the hydrophobic effect, solvation and desolvation, ionization of acids and bases, and tautomerism. HINT is derived from the experimental measurement of the partition coefficient for solute transfer between water and 1-octanol, stylized as log P
o/w, which is a free energy term. Discussion of computational approaches to quantitating the hydrophobic effect, scoring biological associations with a free energy force field, evaluating the conservation of water molecules in complexes, modeling ionization state ensembles in complex environments, enumerating putative small-molecule tautomers in complexes, and predicting the location of important bridging waters are provided. These factors are summarized in terms of their potential effects on drug discovery projects.
... The present paper is in a series of articles describing our work in developing computational tools for drug design3637. The development of the VICE cavity detection algorithm was initially motivated by our need for a tool that could be tightly integrated with other algorithms in the HINT toolkit suite [38]. ...
Systematic investigation of a protein and its binding site characteristics are crucial for designing small molecules that modulate protein functions. However, fundamental uncertainties in binding site interactions and insufficient knowledge of the properties of even well-defined binding pockets can make it difficult to design optimal drugs. Herein, we report the development and implementation of a cavity detection algorithm built with HINT toolkit functions that we are naming Vectorial Identification of Cavity Extents (VICE). This very efficient algorithm is based on geometric criteria applied to simple integer grid maps. In testing, we carried out a systematic investigation on a very diverse data set of proteins and protein-protein/protein-polynucleotide complexes for locating and characterizing the indentations, cavities, pockets, grooves, channels, and surface regions. Additionally, we evaluated a curated data set of unbound proteins for which a ligand-bound protein structures are also known; here the VICE algorithm located the actual ligand in the largest cavity in 83% of the cases and in one of the three largest in 90% of the cases. An interactive front-end provides a quick and simple procedure for locating, displaying and manipulating cavities in these structures. Information describing the cavity, including its volume and surface area metrics, and lists of atoms, residues, and/or chains lining the binding pocket, can be easily obtained and analyzed. For example, the relative cross-sectional surface area (to total surface area) of cavity openings in well-enclosed cavities is 0.06 +/- 0.04 and in surface clefts or crevices is 0.25 +/- 0.09. Proteins 2010. (c) 2009 Wiley-Liss, Inc.
... This process encompasses the computational titration algorithm for which some aspects have been described previously. 60,61,62,63 In this report, we are describing an online version of the HINT-based computational titration method, i.e., a free web service for studying protonation states and Gibbs free energies of binding for protein-ligand complexes. This user-friendly web service can be used to solve quite a few potential problems in protein-ligand structural models; namely: questionable group rotations, optimized rotations for H-donor protons, and multiple interacting protonation states. ...
A public web server performing computational titration at the active site in a protein-ligand complex has been implemented. This calculation is based on the Hydropathic interaction noncovalent force field. From 3D coordinate data for the protein, ligand and bridging waters (if available), the server predicts the best combination of protonation states for each ionizable residue and/or ligand functional group as well as the Gibbs free energy of binding for the ionization-optimized protein-ligand complex. The 3D structure for the modified molecules is available as output. In addition, a graph depicting how this energy changes with acidity, i.e., as a function of added protons, can be obtained. This data may prove to be of use in preparing models for virtual screening and molecular docking. A few illustrative examples are presented. In beta secretase (2va7) computational titration flipped the amide groups of Gln12 and Asn37 and protonated a ligand amine yielding an improvement of 6.37 kcal mol(-1) in the protein-ligand binding score. Protonation of Glu139 in mutant HIV-1 reverse transcriptase (2opq) allows a water bridge between the protein and inhibitor that increases the protein-ligand interaction score by 0.16 kcal mol(-1). In human sialidase NEU2 complexed with an isobutyl ether mimetic inhibitor (2f11) computational titration suggested that protonating Glu218, deprotonating Arg237, flipping the amide bond on Tyr334, and optimizing the positions of several other polar protons would increase the protein-ligand interaction score by 0.71 kcal mol(-1).
Since its introduction about four decades ago, docking and scoring are now the very heart of structure‐based drug design. The past 10–20 years have seen a plethora of docking and scoring tools successfully integrated into drug discovery pipelines. Interestingly, while artificial intelligence is now receiving significant attention in the computational modeling arena, docking and scoring have long utilized these techniques. This comprehensive summary highlights some of the most significant achievements of docking and scoring, reviews the current status, provides descriptions of many of the tools available, comments on some of the outstanding challenges facing the paradigm, and offers perspectives and advice on best practices for users. While significant development is still needed in docking of flexible molecules and accurate Gibb's free energy predictions, docking and scoring are very useful when handled by experienced practitioners, but less so if treated as a “black box.” Lastly, we present a hypothetical case so beginners may appreciate the nuances of setting up a docking study. Focus in docking is now shifting toward parallel applications, i.e. protein–protein, protein–oligosaccharide, protein–DNA, or protein–RNA docking and polypharmacology. In summary, this article is intended to elucidate the nuances of the subject, while providing guidelines for practical implementation of effective workflows in drug discovery and structural biology.
Characterizing the nature of interaction between proteins that have not been experimentally co-crystallized requires a computational docking approach that can successfully predict the spatial conformation adopted in the complex. In this work, the Hydropathic INTeractions (HINT) force field model was used for scoring docked models in a data set of 30 high-resolution crystallographically characterized "dry" protein-protein complexes, and was shown to reliably identify native-like models. However, most current protein-protein docking algorithms fail to explicitly account for water molecules involved in bridging interactions that mediate and stabilize the association of the protein partners, so we used HINT to illuminate the physical and chemical properties of bridging waters and account for their energetic stabilizing contributions. The HINT water Relevance metric identified the 'truly' bridging waters at the 30 protein-protein interfaces and we utilized them in "solvated" docking by manually inserting them into the input files for the rigid body ZDOCK program. By accounting for these interfacial waters, a statistically significant improvement of ~24% in the average hit-count within the top-10 predictions the protein-protein dataset was seen, compared to standard "dry" docking. The results also show scoring improvement, with medium and high accuracy models ranking much better than incorrect ones. These improvements can be attributed to the physical presence of water molecules that alter surface properties and better represent native shape and hydropathic complementarity between interacting partners, with concomitantly more accurate native-like structure predictions. © Proteins 2013;. © 2013 Wiley Periodicals, Inc.
The docking and scoring paradigm can be considered as the combination of two separate problems. The first aspect is a geometric, or more broadly an informatics problem: how can we place a solid object (ligand) within a “cavity” of another solid (protein) or close to another molecule in a well-defined Cartesian space? The second one is a more intriguing chemical problem: how can we properly predict the free energy of binding considering all the possible contributions involved in biological interactions? There is a wide range of algorithms and approaches used to produce docking poses and, consequently, a wide range of associated scoring functions used to judge the possible poses. In several cases the scoring functions are deeply entwined with the search method and can not be considered separately. In other cases, more than one scoring function is provided in docking programs, each showing different strengths and limitations. Consensus scoring approaches, combining multiple methods into a single metric, have been created to overcome the weaknesses characterizing the different docking algorithms and the associated scoring functions. Correctly predicting not just the binding mode, but also the binding energy, is a primary exigency in all docking simulations and, in particular, in virtual screening applications. Accurate estimation of binding free energy would allow, not only good discrimination between active and inactive molecules, but also among closely related analogs, this latter case being particularly important for drug design. In this chapter we discuss problems related to docking/scoring techniques for in silico screening and we review the most common scoring methods.
The rational development of new lead compounds requires good understanding of the relationship between all the actors involved in a binding event (protein, ligand, water, metal ions, cofactors, etc.). Computational methods attempt to reproduce and predict the behavior of nature even though this can be very difficult. The docking/scoring paradigm is probably the most widespread and potentially useful computer-aided technique used in the discovery of new drugs. This paradigm can be analyzed as the sum of a “geometric” problem, that is, the implementation of algorithms to find the possible positions of a ligand in a receptor cavity, and a “chemistry” problem, that is, the evaluation of the solution list using good and realistic energy functions. In this chapter, we deal with the panorama of docking and scoring approaches and the related software packages. After a general introduction, some basic principles about the goodness and limits of experimental data used for computational simulations are described. Then an exhaustive examination of the most common docking methods and packages is carried out followed by an analysis of scoring functions developed to date, including the evolving consensus scoring approach. Next, several problems with the paradigm and their state-of-the-art partial solutions are discussed, including active site water, ionization states, tautomerization, flexibility, and the probability of more than one “correct” solution. Particular attention is paid for the upside and the downside of the problem in a short user guide for the docking and scoring beginner, followed by a conclusions and outlook.
Keywords:
computational chemistry;
free energy;
ligand docking;
modeling;
scoring functions
The Estrogen Receptor (ER) is a ligand activated transcription factor involved in numerous fundamental biological processes as in many important diseases and malfunctions. Since 1998, when the first structure of the ER ligand binding domain complexed with 17 beta-estradiol (E2) was released, the number of ER alpha and ER beta crystallographic structures constantly increased. Nevertheless, little is still known about several fundamental events that govern the regular biological activity, or that modulate the transcription response following the interaction of the receptor with xenobiotic compounds. Moreover, the peculiar flexibility of the receptor characterized by two levels of conformational changes, i.e slight adjustments of binding pocket residues side chains, and more significant displacement of helix 12, moving from a close/agonist-like to an open/antagonist-like position, makes experimental approaches unable to properly describe and predict the receptor conformational equilibrium. Which is the most probable structure of the unbound receptor? How do biological ligands enter the receptor? How does the tissue-related pull of coactivators and corepressors affect the puzzling conformational equilibrium of the receptor? Since most of these questions still do not have an answer. A proper description of the structure-activity relationship and of the pharmacophoric properties of the binding pocket would be of paramount importance in order to design new agonist and antagonist molecules, and to understand how diverse xenobiotic compounds can alter the conformational equilibrium of the receptor, inducing estrogenic or anti-estrogenic effects. In this review we report the most relevant computational approaches, both theoretical and applicative, and the latest proposed models.
Hendra virus (HeV) is a recently identified paramyxovirus that is fatal in humans and could be used as an agent of bioterrorism.
The HeV receptor-binding protein (G) is required in order for the fusion protein (F) to mediate fusion, and analysis of the
triggering/activation of HeV F by G should lead to strategies for interfering with this key step in viral entry. HeV F, once
triggered by the receptor-bound G, by analogy with other paramyxovirus F proteins, undergoes multistep conformational changes
leading to a six-helix bundle (6HB) structure that accomplishes fusion of the viral and cellular membranes. The ectodomain
of paramyxovirus F proteins contains two conserved heptad repeat regions (HRN and HRC) near the fusion peptide and the transmembrane
domains, respectively. Peptides derived from the HRN and HRC regions of F are proposed to inhibit fusion by preventing F,
after the initial triggering step, from forming the 6HB structure that is required for fusion. HeV peptides have previously
been found to be effective at inhibiting HeV fusion. However, we found that a human parainfluenza virus 3 F-peptide is more
effective at inhibiting HeV fusion than the comparable HeV-derived peptide.
The NF-kappaB family of transcription factors is an important component of stress-activated cytoprotective signal transduction pathways. Previous studies demonstrated that some activation mechanisms require phosphorylation, ubiquitination, and degradation of the inhibitor protein, IkappaBalpha. Herein, it is demonstrated that ionizing radiation in the therapeutic dose range stimulates NF-kappaB activity by a mechanism in which IkappaBalpha tyrosine 181 is nitrated as a consequence of constitutive NO* synthase activation, leading to dissociation of intact IkappaBalpha from NF-kappaB. This mechanism does not appear to require IkappaBalpha kinase-dependent phosphorylation or proteolytic degradation of IkappaBalpha. Tyrosine 181 is involved in several noncovalent interactions with the p50 subunit of NF-kappaB stabilizing the IkappaBalpha-NF-kappaB complex. Evaluation of hydropathic interactions of the IkappaBalpha-p50 complex on the basis of the crystal structure of the complex is consistent with nitration disrupting these interactions and dissociating the IkappaBalpha-NF-kappaB complex. Tyrosine nitration is not commonly studied in the context of signal transduction. However, these results indicate that tyrosine nitration is an important post-translational regulatory modification for NF-kappaB activation and possibly for other signaling molecules modulated by mild and transient oxidative and nitrosative stresses.
Erythrocyte pyruvate kinase (PK) is an important glycolytic enzyme, and manipulation of its regulatory behavior by allosteric modifiers is of interest for medicinal purposes. Human-erythrocyte PK was expressed in Rosetta cells and purified on an Ni-NTA column. A search of the small-molecules database of the National Cancer Institute (NCI), using the UNITY software, led to the identification of several compounds with similar pharmacophores as fructose-1,6-bisphosphate (FBP), the natural allosteric activator of the human kinases. The compounds were subsequently docked into the FBP binding site using the programs FlexX and GOLD, and their interactions with the protein were analyzed with the energy-scoring function of HINT. Seven promising candidates, compounds 1-7, were obtained from the NCI, and subjected to kinetics analysis, which revealed both activators and inhibitors of the R-isozyme of PK (R-PK). The allosteric effectors discovered in this study could prove to be lead compounds for developing medications for the treatment of hemolytic anemia, sickle-cell anemia, hypoxia-related diseases, and other disorders arising from erythrocyte PK malfunction.
The similarity in structure of potassium (K(+)) channels from different families has been revealed by only recently available crystallographic 3D structural data. The hydropathic analysis presented in this work illuminates whether homologous residues perform the same functions in channels that use different gating mechanisms. We calculated and compared the hydropathic profiles of two K(+) channels, KcsA and Kv1.2 (the latter a member of the Shaker family), at their pore-forming domain. Quantitative information describing important interactions stabilizing the protein beyond obvious secondary-structure elements was extracted from the analysis and applied as a template for subsequent molecular-dynamics (MD) analyses. For example, two key groups of interactions, defining the turns that connect the transmembrane helices and responsible for the orientation of the pore helix, were identified. Our results also indicate that Asp(80) and Asp(379) play a similar role in stabilizing the P-loop of KcsA and Kv1.2, respectively, but to significantly different extents.
The computational-titration (CT) algorithm based on the 'natural' Hydropathic INTeractions (HINT) force field is described. The HINT software model is an empirical, non-Newtonian force field derived from experimentally measured partition coefficients for solvent transfer between octanol and H(2)O (log P(o/w)). The CT algorithm allows the identification, modeling, and optimization of multiple protonation states of residues and ligand functional groups at the protein-ligand active site. The importance of taking into account pH and ionization states of residues, which strongly affect the process of ligand binding, for correctly predicting binding free energies is discussed. The application of the CT protocol to a set of six cyclic inhibitors in their complexes with HIV-1 protease is presented, and the advance of HINT as a virtual-screening tool is outlined.
A general method has been developed to characterize the hydrophobicity or hydrophilicity of the microenvironment (MENV), in which a given amino acid side chain is immersed, by calculating a quantitative property descriptor (QPD) based on the relative (to water) hydrophobicity of the MENV. Values of the QPD were calculated for a test set of 733 proteins to analyze the modulating effects on amino acid residue properties by the MENV in which they are imbedded. The QPD values and solvent accessibility were used to derive a partitioning of residues based on the MENV hydrophobicities. From this partitioning, a new hydrophobicity scale was developed, entirely in the context of protein structure, where amino acid residues are immersed in one or more "MENVpockets." Thus, the partitioning is based on the residues "sampling" a large number of "solvents" (MENVs) that represent a very large range of hydrophobicity values. It was found that the hydrophobicity of around 80% of amino acid side chains and their MENV are complementary to each other, but for about 20%, the MENV and their imbedded residue can be considered as mismatched. Many of these mismatches could be rationalized in terms of the structural stability of the protein and/or the involvement of the imbedded residue in function. The analysis also indicated a remarkable conservation of local environments around highly conserved active site residues that have similar functions across protein families, but where members have relatively low sequence homology. Thus, quantitative evaluation of this QPD is suggested, here, as a tool for structure-function prediction, analysis, and parameter development for the calculation of properties in proteins.
The design of an extension to the HINT (Hydropathic INTeractions) program for locating and automatically placing in relevant orientations bridging water molecules is described in detail. This application is used to analyze the structures of HIV-1 protease-inhibitor complexes for five key bridging water molecules. The tool locates the water molecules with an overall accuracy of 1.28 ± 0.55 Å, and orients them (relative to potentially erroneous molecular mechanics optimized water molecules) to 41 ± 23 degrees. This automated placement is in contrast to other programs that calculate 3D contour maps of energetically-likely binding locations for water molecules, which must be followed by manual placement and energy minimization of these water positions and orientations to create the model. Also, the new object-oriented toolkit design for the HINT program (http://www.edusoft-lc.com/toolkits/) is described. The toolkit includes molecule, atom and monomer objects to represent chemical structure. A second class of objects describes 3D maps and includes tools for their manipulation. The hint objects are designed around the primary goals of HINT: partitioning (calculating LogPo/w for) molecules, calculating interaction scores between molecules and calculating 3D maps of molecular and intermolecular hydropathy.
X-ray data have been extended to 1.7 A for a binary complex of Escherichia coli dihydrofolate reductase with methotrexate and a ternary complex of Lactobacillus casei dihydrofolate reductase with methotrexate and NADPH. Models for both structures have been refined to R factors of 0.15 and include parameters for fixed and liquid solvent. The two species of dihydrofolate reductase resemble one another even more closely than was thought to be the case prior to refinement. Several new structural features have also been discovered. Among them are a cis peptide linking Gly-97 and Gly-98 (L. Casei numbering) in both species, an alpha helix involving residues 43 through 50 in the E. coli enzyme, and the existence of what may be a specific hydration site on exposed alpha helices. Refinement has led to a revised description of the details of methotrexate binding. We now see that a fixed water molecule mediates the interaction between methotrexate's 2-amino group and Thr-116 (L. casei numbering) and that the inhibitor's 4-amino group makes two hydrogen bonds with the enzyme (instead of one). Other revisions are also discussed. A hypothetical model for substrate binding is proposed in which the pteridine ring is turned upside down while all protein and solvent atoms remain fixed. Asp-26 in this model is hydrogen bonded to the substrate's 2-amino group and to N3.
Hydrogen bonds are a major feature of protein structure. By a generally accepted definition, they occur whenever a proton is shared by two electronegative atoms. Hence, only hydrogens bonded to nitrogen and oxygen atoms are usually considered in analyses of protein hydrogen bond networks. However, X-ray and neutron diffraction studies have shown that crystals of various organic compounds exhibit close C-H...X contacts (where X is an electronegative atom, in most cases oxygen) which show all the stereochemical hallmarks of hydrogen bonds. In this work, we describe an analysis of short C-H...O interactions in a sample of known protein structures representing different categories of tertiary folds and refined at a resolution of at least 2 A. Although our analysis is based on the calculated coordinates of hydrogen atoms, its results are statistically significant: we find strong evidence that a large percentage of short C...O contacts constitute cohesive interactions. Moreover, the stereochemical study of C-H...O = C contacts, in which the orientation of free electron orbitals on the acceptor oxygen atom can be predicted, reveals that these interactions exhibit stereochemical features typical of hydrogen bonds. Among the hydrogen atoms involved in these contacts, the most common are those bonded to alpha carbon. This is consistent with the fact that these hydrogens are more acidic than others. We describe four different categories of C-H...O = C bonds. Those found between C alpha-H groups and main chain oxygens in adjacent strands of beta sheets are the most ubiquitous. Our results call for a revision of crystallographic restrained refinement programs which treat close carbon-oxygen contacts as purely repulsive; they may also have implications for the understanding of some enzymatic reaction mechanisms.
A congeneric series of paullones were characterized using a 3-D QSAR with cyclin-dependent kinase 1 (CDK1) inhibition data. A homology model of CDK1-cyclin B was developed from the crystal structure of CDK2-cyclin A, which subsequently served as the basis for the structure-based design. Paullones were docked into the ATP binding site of the CDK1-cylin B models and were optimized with molecular mechanics. Hydropathic analyses of the paullone-CDK1 complexes were performed after the atom types were assigned based on each ligand's electronic properties calculated from quantum mechanics. Hydropathic descriptors formed a significant multiple regression equation that predicts paullone IC50 data. The results indicate that the combination of hydropathic descriptors with molecular mechanics geometries are sufficient to design overt steric and chemical complementarity of the ligands. However, the electronic properties derived from quantum mechanics helped direct synthetic chemistry efforts to produce ligands that promote better charge transfer and strengthen hydrogen bonding as facilitated by resonance stabilization. Compounds with low affinity for CDK1 were poor charge acceptors and made less than ideal hydrogen bonding arrangements with the receptor. These considerations led to the prediction that structures such as 9-cyanopaullone would be considerably more potent than the parent compound, a finding supported by enzyme inhibition data. Also, 9-nitropaullone emerged as a paullone which also had similar potency in enzyme inhibition as well as a favorable anti-proliferative activity profile in living cells.
For the first time, a statistical potential has been developed to quantitatively describe the CH···O hydrogen bonding interaction
at the protein-protein interface. The calculated energies of the CH···O pair interaction show a favorable valley at ∼3.3 Å,
exhibiting a feature typical of an H-bond and similar to the ab initio quantum calculation result (Scheiner, S., Kar, T., and Gu, Y. (2001) J. Biol. Chem. 276, 9832–9837). The potentials have been applied to a set of 469 protein-protein complexes to calculate the contribution
of different types of interactions to each protein complex: the average energy contribution of a conventional H-bond is ∼30%;
that of a CH···O H-bond is 17%; and that of a hydrophobic interaction is 50%. In some protein-protein complexes, the contribution
of the CH···O H-bond can reach as high as ∼40–50%, indicating the importance of the CH···O H-bond at the protein interface.
At the interfaces of these complexes, CαH···O H-bonds frequently occur between adjacent strands in both parallel and antiparallel orientations, having the obvious
structural motif of bifurcated H-bonds. Our study suggests that the weak CH···O H-bond makes an important contribution to
the association and stability of protein complexes and needs more attention in protein-protein interaction studies.
New empirical scoring functions have been developed to estimate the binding affinity of a given protein-ligand complex with known three-dimensional structure. These scoring functions include terms accounting for van der Waals interaction, hydrogen bonding, deformation penalty, and hydrophobic effect. A special feature is that three different algorithms have been implemented to calculate the hydrophobic effect term, which results in three parallel scoring functions. All three scoring functions are calibrated through multivariate regression analysis of a set of 200 protein-ligand complexes and they reproduce the binding free energies of the entire training set with standard deviations of 2.2 kcal/mol, 2.1 kcal/mol, and 2.0 kcal/mol, respectively. These three scoring functions are further combined into a consensus scoring function, X-CSCORE. When tested on an independent set of 30 protein-ligand complexes. X-CSCORE is able to predict their binding free energies with a standard deviation of 2.2 kcal/mol. The potential application of X-CSCORE to molecular docking is also investigated. Our results show that this consensus scoring function improves the docking accuracy considerably when compared to the conventional force field computation used for molecular docking.
A novel and robust automated docking method that predicts the bound conformations of flexible ligands to macromolecular targets has been developed and tested, in combination with a new scoring function that estimates the free energy change upon binding. Interestingly, this method applies a Lamarckian model of genetics, in which environmental adaptations of an individual's phenotype are reverse transcribed into its genotype and become heritable traits (sic). We consider three search methods, Monte Carlo simulated annealing, a traditional genetic algorithm, and the Lamarckian genetic algorithm, and compare their performance in dockings of seven protein–ligand test systems having known three-dimensional structure. We show that both the traditional and Lamarckian genetic algorithms can handle ligands with more degrees of freedom than the simulated annealing method used in earlier versions of AUTODOCK, and that the Lamarckian genetic algorithm is the most efficient, reliable, and successful of the three. The empirical free energy function was calibrated using a set of 30 structurally known protein–ligand complexes with experimentally determined binding constants. Linear regression analysis of the observed binding constants in terms of a wide variety of structure-derived molecular properties was performed. The final model had a residual standard error of 9.11 kJ mol⁻¹ (2.177 kcal mol⁻¹) and was chosen as the new energy function. The new search methods and empirical free energy function are available in AUTODOCK, version 3.0. © 1998 John Wiley & Sons, Inc. J Comput Chem 19: 1639–1662, 1998
A novel and robust automated docking method that predicts the bound conformations of flexible ligands to macromolecular targets has been developed and tested, in combination with a new scoring function that estimates the free energy change upon binding. Interestingly, this method applies a Lamarckian model of genetics, in which environmental adaptations of an individual's phenotype are reverse transcribed into its genotype and become . heritable traits sic . We consider three search methods, Monte Carlo simulated annealing, a traditional genetic algorithm, and the Lamarckian genetic algorithm, and compare their performance in dockings of seven protein)ligand test systems having known three-dimensional structure. We show that both the traditional and Lamarckian genetic algorithms can handle ligands with more degrees of freedom than the simulated annealing method used in earlier versions of AUTODOCK, and that the Lamarckian genetic algorithm is the most efficient, reliable, and successful of the three. The empirical free energy function was calibrated using a set of 30 structurally known protein)ligand complexes with experimentally determined binding constants. Linear regression analysis of the observed binding constants in terms of a wide variety of structure-derived molecular properties was performed. The final model had a residual standard y1 y1 .
Molecular docking is a key problem in structure-based drug design. Main applications are the selection of molecules for biological screening assays as well as the generation of protein-ligand complex structures for the basic understanding of binding principles and for providing guidance to improve lead structures. In this review, we summarize our methodical developments of the software package FLEXX and we demonstrate its application to several test cases.
Introduction Theory Computational Details Free Energy Perturbation Calculations for Small Molecules Free Energy Perturbation Calculations for Macromolecules Guide to Structure-Based Ligand Optimization Optimization of Ligands to HIV-1 Protease: Using the FEP Method Conclusions Brief Guide for Free Energy Calculations and Their Use in Ligand Optimization
The ability to generate feasible binding orientations of a small molecule within a site of known structure is important for ligand design. We present a method that combines a rapid, geometric docking algorithm with the evaluation of molecular mechanics interaction energies. The computational costs of evaluation are minimal because we precalculate the receptor-dependent terms in the potential function at points on a three-dimensional grid. In four test cases where the components of crystallographically determined complexes are redocked, the “force field” score correctly identifies the family of orientations closest to the experimental binding geometry. Scoring functions that consider only steric factors or only electrostatic factors are less successful. The force field function will play an important role in our efforts to search databases for potential lead compounds.
A non-covalent interaction force field model derived from the partition coefficient of 1-octanol/water solubility is described. This model, HINT for Hydropathic INTeractions, is shown to include, in very empirical and approximate terms, all components of biomolecular associations, including hydrogen bonding, Coulombic interactions, hydrophobic interactions, entropy and solvation/desolvation. Particular emphasis is placed on: (1) demonstrating the relationship between the total empirical HINT score and free energy of association, G
interaction; (2) showing that the HINT hydrophobic-polar interaction sub-score represents the energy cost of desolvation upon binding for interacting biomolecules; and (3) a new methodology for treating constrained water molecules as discrete independent small ligands. An example calculation is reported for dihydrofolate reductase (DHFR) bound with methotrexate (MTX). In that case the observed very tight binding, G
interaction–13.6kcal mol–1, is largely due to ten hydrogen bonds between the ligand and enzyme with estimated strength ranging between –0.4 and –2.3kcalmol–1. Four water molecules bridging between DHFR and MTX contribute an additional –1.7kcalmol–1 stability to the complex. The HINT estimate of the cost of desolvation is +13.9kcalmol–1.
A novel and robust automated docking method that predicts the bound conformations of flexible ligands to macromolecular targets has been developed and tested, in combination with a new scoring function that estimates the free energy change upon binding. Interestingly, this method applies a Lamarckian model of genetics, in which environmental adaptations of an individual's phenotype are reverse transcribed into its genotype and become . heritable traits sic . We consider three search methods, Monte Carlo simulated annealing, a traditional genetic algorithm, and the Lamarckian genetic algorithm, and compare their performance in dockings of seven protein)ligand test systems having known three-dimensional structure. We show that both the traditional and Lamarckian genetic algorithms can handle ligands with more degrees of freedom than the simulated annealing method used in earlier versions of AUTODOCK, and that the Lamarckian genetic algorithm is the most efficient, reliable, and successful of the three. The empirical free energy function was calibrated using a set of 30 structurally known protein)ligand complexes with experimentally determined binding constants. Linear regression analysis of the observed binding constants in terms of a wide variety of structure-derived molecular properties was performed. The final model had a residual standard y1 y1. error of 9.11 kJ mol 2.177 kcal mol and was chosen as the new energy
The 2.2-A crystal structure of chicken liver dihydrofolate reductase (EC 1.5.1.3, DHFR) has been solved as a ternary complex with NADP+ and biopterin (a poor substrate). The space group and unit cell are isomorphous with the previously reported structure of chicken liver DHFR complexed with NADPH and phenyltriazine [Volz, K. W., Matthews, D. A., Alden, R. A., Freer, S. T., Hansch, C., Kaufman, B. T., & Kraut, J. (1982) J. Biol. Chem. 257, 2528-2536]. The structure contains an ordered water molecule hydrogen-bonded to both hydroxyls of the biopterin dihydroxypropyl group as well as to O4 and N5 of the biopterin pteridine ring. This water molecule, not observed in previously determined DHFR structures, is positioned to complete a proposed route for proton transfer from the side-chain carboxylate of E30 to N5 of the pteridine ring. Protonation of N5 is believed to occur during the reduction of dihydropteridine substrates. The positions of the NADP+ nicotinamide and biopterin pteridine rings are quite similar to the nicotinamide and pteridine ring positions in the Escherichia coli DHFR.NADP+.folate complex [Bystroff, C., Oatley, S. J., & Kraut, J. (1990) Biochemistry 29, 3263-3277], suggesting that the reduction of biopterin and the reduction of folate occur via similar mechanisms, that the binding geometry of the nicotinamide and pteridine rings is conserved between DHFR species, and that the p-aminobenzoylglutamate moiety of folate is not required for correct positioning of the pteridine ring in ground-state ternary complexes. Instead, binding of the p-aminobenzoylglutamate moiety of folate may induce the side chain of residue 31 (tyrosine or phenylalanine) in vertebrate DHFRs to adopt a conformation in which the opening to the pteridine binding site is too narrow to allow the substrate to diffuse away rapidly. A reverse conformational change of residue 31 is proposed to be required for tetrahydrofolate release.
Finding novel leads from which to design drug molecules has traditionally been a matter of screening and serendipity. We present a method for finding a wide assortment of chemical structures that are complementary to the shape of a macromoleculer receptor site whose X-ray crystallographic structure is known. Each of a set of small molecules from the Cambridge Crystallographic Database (Allen; et al. J. Chem. Doc. 1973, 13, 119) is individually docked to the receptor in a number of geometrically permissible orientations with use of the docking algorithm developed by Kuntz et al. (J. Mol. Biol. 1982, 161, 269). The orientations are evaluated for goodness-of-fit, and the best are kept for further examination using the molecular mechanics program AMBER (Weiner; Kollman J. Comput. Chem. 1981, 106, 765). The shape-search algorithm finds known ligands as well as novel molecules that fit the binding site being studied. The highest scoring orientations of known ligands resemble binding modes generated by interactive modeling or determined crystallographically. We describe the application of this procedure to the binding sites of papain and carbonic anhydrase. While the compounds recovered from the Cambridge Crystallographic Database are not, themselves, likely to be inhibitors or substrates of these enzymes, we expect that the structures from such searches will be useful in the design of active compounds.
The crystal structure of recombinant human dihydrofolate reductase with folate bound in the active site has been determined and the structural model refined at 0.2-nm resolution. Preliminary studies of the binding of the inhibitors methotrexate and trimethoprim to the human apoenzyme have been performed at 0.35-nm resolution. The conformations of the chemically very similar ligands folate and methotrexate, one a substrate the other a potent inhibitor, differ substantially in that their pteridine rings are in inverse orientations relative to their p-aminobenzoyl-L-glutamate moieties. Methotrexate binding is similar to that previously observed in two bacterial enzymes but is quite different from that observed in the enzyme from a mouse lymphoma cell line [Stammers et al. (1987) FEBS Lett. 218, 178-184]. The geometry of the polypeptide chain around the folate binding site in the human enzyme is not consistent with conclusions previously drawn with regard to the species selectivity of the inhibitor trimethoprim [Matthews et al. (1985) J. Biol. Chem. 260, 392-399].
The interaction of a probe group with a protein of known structure is computed at sample positions throughout and around the macromolecule, giving an array of energy values. The probes include water, the methyl group, amine nitrogen, carboxy oxygen, and hydroxyl. Contour surfaces at appropriate energy levels are calculated for each probe and displayed by computer graphics together with the protein structure. Contours at negative energy levels delineate contours also enable other regions of attraction between probe and protein and are found at known ligand binding clefts in particular. The contours also enable other regions of attraction to be identified and facilitate the interpretation of protein-ligand energetics. They may, therefore, be of value for drug design.
Observations of short CH…O contacts in biological macromolecules, including nucleic acids, proteins and carbohydrates, suggest that these unconventional hydrogen bonds have both a structurally and functionally important role.
Prediction of small molecule binding modes to macromolecules of known three-dimensional structure is a problem of paramount importance in rational drug design (the "docking" problem). We report the development and validation of the program GOLD (Genetic Optimisation for Ligand Docking). GOLD is an automated ligand docking program that uses a genetic algorithm to explore the full range of ligand conformational flexibility with partial flexibility of the protein, and satisfies the fundamental requirement that the ligand must displace loosely bound water on binding. Numerous enhancements and modifications have been applied to the original technique resulting in a substantial increase in the reliability and the applicability of the algorithm. The advanced algorithm has been tested on a dataset of 100 complexes extracted from the Brookhaven Protein DataBank. When used to dock the ligand back into the binding site, GOLD achieved a 71% success rate in identifying the experimental binding mode.
The HIV-1 transframe region (TFR) is between the structural and functional domains of the Gag-Pol polyprotein, flanked by the nucleocapsid and the protease domains at its N and C termini, respectively. Transframe octapeptide (TFP) Phe-Leu-Arg-Glu-Asp-Leu-Ala-Phe, the N terminus of TFR, and its analogues are competitive inhibitors of the action of the mature HIV-1 protease. The smallest, most potent analogues are tripeptides: Glu-Asp-Leu and Glu-Asp-Phe with Ki values of approximately 50 and approximately 20 microM, respectively. Substitution of the acidic amino acids in the TFP by neutral amino acids and d or retro-d configurations of Glu-Asp-Leu results in an >40-fold increase in Ki. Protease inhibition by Glu-Asp-Leu is dependent on a protonated form of a group with a pKa of 3.8; unlike other inhibitors of HIV-1 protease which are highly hydrophobic, Glu-Asp-Leu is extremely soluble in water, and its binding affinity decreases with increasing NaCl concentration. However, Glu-Asp-Leu is a poor inhibitor (Ki approximately 7.5 mM) of the mammalian aspartic acid protease pepsin. X-ray crystallographic studies at pH 4.2 show that the interactions of Glu at P2 and Leu at P1 of Glu-Asp-Leu with residues of the active site of HIV-1 protease are similar to those of other product-enzyme complexes. It was not feasible to understand the interaction of intact TFP with HIV-1 protease under conditions of crystal growth due to its hydrolysis giving rise to two products. The sequence-specific, selective inhibition of the HIV-1 protease by the viral TFP suggests a role for TFP in regulating protease function during HIV-1 replication.
A dataset of 82 protein-ligand complexes of known 3D structure and binding constant Ki was analysed to elucidate the important factors that determine the strength of protein-ligand interactions. The following parameters were investigated: the number and geometry of hydrogen bonds and ionic interactions between the protein and the ligand, the size of the lipophilic contact surface, the flexibility of the ligand, the electrostatic potential in the binding site, water molecules in the binding site, cavities along the protein-ligand interface and specific interactions between aromatic rings. Based on these parameters, a new empirical scoring function is presented that estimates the free energy of binding for a protein-ligand complex of known 3D structure. The function distinguishes between buried and solvent accessible hydrogen bonds. It tolerates deviations in the hydrogen bond geometry of up to 0.25 A in the length and up to 30 degrees in the hydrogen bond angle without penalizing the score. The new energy function reproduces the binding constants (ranging from 3.7 x 10(-2) M to 1 x 10(-14) M, corresponding to binding energies between -8 and -80 kJ/mol) of the dataset with a standard deviation of 7.3 kJ/mol corresponding to 1.3 orders of magnitude in binding affinity. The function can be evaluated very fast and is therefore also suitable for the application in a 3D database search or de novo ligand design program such as LUDI. The physical significance of the individual contributions is discussed.
The empirically calculated parameter LogP(o/w), the log(10) of the coefficient for solvent partitioning between 1-octanol and water, has been used to provide the key data for a unique non-covalent interaction force field called HINT (Hydropathic INTeractions). This experimentally-derived force field encodes entropic as well as enthalpic information and also includes some representation of solvation and desolvation energetics in biomolecular associations. The theoretical basis for the HINT model is discussed. This review includes: 1) discussion of calculational representation of the hydrophobic effect, 2) the rationale for describing the experimental LogP(o/w) based descriptors used in the HINT force field and model as free energy-like, 3) the relationship between hydrophobic fragment constants and partial group electrostatic charge, and 4) the implications of structurally-conserved water molecules on free energy of molecular association. Several recent applications of HINT in structure-based and ligand-based drug discovery are reviewed. Finally, future directions in the HINT model development are proposed.
The prediction of the binding affinity between a protein and ligands is one of the most challenging issues for computational biochemistry and drug discovery. While the enthalpic contribution to binding is routinely available with molecular mechanics methods, the entropic contribution is more difficult to estimate. We describe and apply a relatively simple and intuitive calculation procedure for estimating the free energy of binding for 53 protein-ligand complexes formed by 17 proteins of known three-dimensional structure and characterized by different active site polarity. HINT, a software model based on experimental LogP(o/w) values for small organic molecules, was used to evaluate and score all atom-atom hydropathic interactions between the protein and the ligands. These total scores (H(TOTAL)), which have been previously shown to correlate with DeltaG(interaction) for protein-protein interactions, correlate with DeltaG(binding) for protein-ligand complexes in the present study with a standard error of +/-2.6 kcal mol(-1) from the equation DeltaG(binding) = -0.001 95 H(TOTAL) - 5.543. A more sophisticated model, utilizing categorized (by interaction class) HINT scores, produces a superior standard error of +/-1.8 kcal mol(-1). It is shown that within families of ligands for the same protein binding site, better models can be obtained with standard errors approaching +/-1.0 kcal mol(-1). Standardized methods for preparing crystallographic models for hydropathic analysis are also described. Particular attention is paid to the relationship between the ionization state of the ligands and the pH conditions under which the binding measurements are made. Sources and potential remedies of experimental and modeling errors affecting prediction of DeltaG(binding) are discussed.
Simplified free energy calculations based on force field energy estimates of ligand-receptor interactions and thermal conformational sampling have emerged as a useful tool in structure-based ligand design. Here we give an overview of the linear interaction energy (LIE) method for calculating ligand binding free energies from molecular dynamics simulations. A notable feature is that the binding energetics can be predicted by considering only the intermolecular interactions of the ligand in the associated and dissociated states. The approximations behind this approach are examined, and different parametrizations of the model are discussed. LIE-type methods appear particularly promising for computational "lead optimization". Recent applications to protein-protein interactions and ion channel blocking are also discussed.
A shape-based Gaussian docking function is constructed which uses Gaussian functions to represent the shapes of individual atoms. A set of 20 trypsin ligand-protein complexes are drawn from the Protein Data Bank (PDB), the ligands are separated from the proteins, and then are docked back into the active sites using numerical optimization of this function. It is found that by employing this docking function, quasi-Newton optimization is capable of moving ligands great distances [on average 7 A root mean square distance (RMSD)] to locate the correctly docked structure. It is also found that a ligand drawn from one PDB file can be docked into a trypsin structure drawn from any of the trypsin PDB files. This implies that this scoring function is not limited to more accurate x-ray structures, as is the case for many of the conventional docking methods, but could be extended to homology models.
Eleven popular scoring functions have been tested on 100 protein-ligand complexes to evaluate their abilities to reproduce experimentally determined structures and binding affinities. They include four scoring functions implemented in the LigFit module in Cerius2 (LigScore, PLP, PMF, and LUDI), four scoring functions implemented in the CScore module in SYBYL (F-Score, G-Score, D-Score, and ChemScore), the scoring function implemented in the AutoDock program, and two stand-alone scoring functions (DrugScore and X-Score). These scoring functions are not tested in the context of a particular docking program. Instead, conformational sampling and scoring are separated into two consecutive steps. First, an exhaustive conformational sampling is performed by using the AutoDock program to generate an ensemble of docked conformations for each ligand molecule. This conformational ensemble is required to cover the entire conformational space as much as possible rather than to focus on a few energy minima. Then, each scoring function is applied to score this conformational ensemble to see if it can identify the experimentally observed conformation from all of the other decoys. Among all of the scoring functions under test, six of them, i.e., PLP, F-Score, LigScore, DrugScore, LUDI, and X-Score, yield success rates higher than the AutoDock scoring function. The success rates of these six scoring functions range from 66% to 76% if using root-mean-square deviation < or =2.0 A as the criterion. Combining any two or three of these six scoring functions into a consensus scoring scheme further improves the success rate to nearly 80% or even higher. However, when applied to reproduce the experimentally determined binding affinities of the 100 protein-ligand complexes, only X-Score, PLP, DrugScore, and G-Score are able to give correlation coefficients over 0.50. All of the 11 scoring functions are further inspected by their abilities to construct a descriptive, funnel-shaped energy surface for protein-ligand complexation. The results indicate that X-Score and DrugScore perform better than the other ones at this aspect.
One factor that can strongly influence predicted free energy of binding is the ionization state of functional groups on the ligands and at the binding site at which calculations are performed. This analysis is seldom performed except in very detailed computational simulations. In this work, we address the issues of (i) modeling the complexity resulting from the different ionization states of ligand and protein residues involved in binding, (ii) if, and how, computational methods can evaluate the pH dependence of ligand inhibition constants, and (iii) how to score the protonation-dependent models. We developed a new and fairly rapid protocol called "computational titration" that enables parallel modeling of multiple ionization ensembles for each distinct protonation level. Models for possible protonation combinations for site/ligand ionizable groups are built, and the free energy of interaction for each of them is quantified by the HINT (Hydropathic INTeractions) software. We applied this procedure to the evaluation of the binding affinity of nine inhibitors (six derived from 2,3-didehydro-2-deoxy-N-acetylneuraminic acid, DANA) of influenza virus neuraminidase (NA), a surface glycoprotein essential for virus replication and thus a pharmaceutically relevant target for the design of anti-influenza drugs. The three-dimensional structures of the NA enzyme-inhibitor complexes indicate considerable complexity as the ligand-protein recognition site contains several ionizable moieties. Each computational titration experiment reveals a peak HINT score as a function of added protons. This maximum HINT score indicates the optimum pH (or the optimum protonation state of each inhibitor-protein binding site) for binding. The pH at which inhibition is measured and/or crystals were grown and analyzed can vary from this optimum. A protonation model is proposed for each ligand that reconciles the experimental complex structure with measured inhibition and the free energy of binding. Computational titration methods allow us to analyze the effect of pH in silico and may be helpful in improving ligand binding free energy prediction when protonation or deprotonation of the residues or ligand functional groups at the binding site might be significant.
"Getting it right" refers to the careful modeling of all elements in the living system, i.e. biological macromolecules, ligands and water molecules. In addition, careful attention should be paid to the protonation state of ionizable functional groups on the ligands and residues at the active site. Computational technology based on the empirical HINT program is described to: (1) calculate free energy scores for ligand binding; (2) include the implicit and explicit effects of water in and around the ligand binding site; and (3) incorporate the effects of global and local pH in molecular models. This last point argues for the simultaneous consideration of a number of molecular models, each with different protonation profiles. Data from recent studies of protein-ligand systems (trypsin, thrombin, neuraminidase, HIV-1 protease and others) are used to illustrate the concepts in the paper. Also discussed are experimental factors related to accurate free energy predictions with this and other computational technologies.
Structural water molecules within protein active sites are relevant for ligand-protein recognition because they modify the active site geometry and contribute to binding affinity. In this work an analysis of the interactions between 23 ligands and dimeric HIV-1 protease is reported. The X-ray structures of these complexes show the presence of four types of structural water molecules: water 301 (on the symmetry axis), water 313, water 313bis, and peripheral waters. Except for water 301, these are generally complemented with a symmetry-related set. The GRID program was used both for checking water locations and for placing water molecules that appear to be missing from the complexes due to crystallographic uncertainty. Hydropathic analysis of the energetic contributions using HINT indicates a significant improvement of the correlation between HINT scores and the experimentally determined binding constants when the appropriate bridging water molecules are taken into account. In the absence of water r2 = 0.30 with a standard error of +/- 1.30 kcal mol(-1) and when the energetic contributions of the constrained waters are included r2 = 0.61 with a standard error of +/- 0.98 kcal mol(-1). HINT was shown to be able to map quantitatively the contribution of individual structural waters to binding energy. The order of relevance for the various types of water is water 301 > water 313 > water 313bis > peripheral waters. Thus, to obtain the most reliable free energy predictions, the contributions of structural water molecules should be included. However, care must be taken to include the effects of water molecules that add information value and not just noise.
A new computational method for analyzing the protonation states of protein-ligand complexes with multiple ionizable groups is applied to the structurally characterized complex between the peptide Glu-Asp-Leu and HIV-1 protease. This complex has eight ionizable groups at the active site: four from the ligand and four Asp residues on the protein. Correlation, with an error of ca. 0.6 kcal mol-1, is made between the calculated titration curve and the experimental titration curve. The analysis suggests that between four and five of the eight ionizable groups are protonated at the pH of crystallization.
One of the more challenging issues in medicinal chemistry is the computation of the free energy of ligand binding to macromolecular targets. This allows for the screening of libraries of chemicals for fast and inexpensive identification of lead compounds. Many attempts have been made and several algorithms have been developed for this purpose. Whereas enthalpic contributions are evaluated using methods and equations for which there is a reasonable consensus among researchers, the entropic contribution is evaluated using very different, and, in some cases, very approximate methods, or it is entirely ignored. Entropic contributions are of primary importance in the formation of many ligand-protein complexes, as well as in protein folding. The hydrophobic interaction, associated with the release of water molecules from the protein active site and the ligand, plays a significant role in complex formation, predominantly contributing to the total entropy change and, in some cases, to the total free energy of binding. There are distinct approaches for the evaluation of the contribution of water molecules to the free energy of binding based on Newtonian mechanics force fields, multi-parameter empirical scoring functions and experimental force fields. This review describes these methods -- discussing both their advantages and limitations. Particular emphasis will be placed on HINT (Hydropatic INTeractions), a "natural" force field that takes into account in a unified way enthalpic and entropic contributions of all interacting atoms in protein-ligand complexes, including released and structured water molecules. As a case-study, the contribution of water molecules to the binding free energy of HIV-1 protease inhibitors is evaluated.
Algorithms and protocols are described for the optimization for H-bonding of isolated singular H2O molecules and entire networks of H2O molecules. Unlike other approaches that are prone to being trapped in local energy minima, these methods rely on exhaustive searches of orientation space for the H2O molecules. The results are scored with the HINT hydropathic interaction model, but the algorithms should be general for any energy-scoring computation. Two examples are provided: 1) the tightly-bound H2O molecule 301 of HIV-1 protease is shown to be more reasonably oriented in terms of forming H-bonds with this method than with a molecular mechanics energy minimization method; and 2) the H2O network surrounding carbonmonoxymyoglobin is constructed and analyzed for a 1.80-A neutron-diffraction structure. The H-atom positions calculated with this method show a somewhat better agreement with the experimental results than do the H-atom positions calculated with molecular mechanics, and both are considerably better than random.
Is log P o/w more than the sum of its parts?
- G E Kellogg
- D J Abraham
- Hydrophobicity
G.E. Kellogg, D.J. Abraham, Hydrophobicity. Is log P o/w more than the sum of its parts? Eur. J. Med. Chem. 35 (2000) 651–661.