Large scale free energy calculations for blind predictions of protein–ligand binding: the D3R Grand Challenge 2015

We describe binding free energy calculations in the D3R Grand Challenge 2015 for blind prediction of the binding affinities of 180 ligands to Hsp90. The present D3R challenge was built around experimental datasets involving Heat shock protein (Hsp) 90, an ATP-dependent molecular chaperone which is an important anticancer drug target. The Hsp90 ATP binding site is known to be a challenging target for accurate calculations of ligand binding affinities because of the ligand-dependent conformational changes in the binding site, the presence of ordered waters and the broad chemical diversity of ligands that can bind at this site. Our primary focus here is to distinguish binders from nonbinders. Large scale absolute binding free energy calculations that cover over 3000 protein–ligand complexes were performed using the BEDAM method starting from docked structures generated by Glide docking. Although the ligand dataset in this study resembles an intermediate to late stage lead optimization project while the BEDAM method is mainly developed for early stage virtual screening of hit molecules, the BEDAM binding free energy scoring has resulted in a moderate enrichment of ligand screening against this challenging drug target. Results show that, using a statistical mechanics based free energy method like BEDAM starting from docked poses offers better enrichment than classical docking scoring functions and rescoring methods like Prime MM-GBSA for the Hsp90 data set in this blind challenge. Importantly, among the three methods tested here, only the mean value of the BEDAM binding free energy scores is able to separate the large group of binders from the small group of nonbinders with a gap of 2.4 kcal/mol. None of the three methods that we have tested provided accurate ranking of the affinities of the 147 active compounds. We discuss the possible sources of errors in the binding free energy calculations. The study suggests that BEDAM can be used strategically to discriminate binders from nonbinders in virtual screening and to more accurately predict the ligand binding modes prior to the more computationally expensive FEP calculations of binding affinity.     

Relationship among ligand conformations in solution, in the solid state, and at the Hsp90 binding site: geldanamycin and radicicol

The unknown effects of a receptor's environment on a ligand's conformation presents a difficult challenge in predicting feasible bioactive conformations, particularly if the receptor is ill-defined. The primary hypothesis of this work is that a structure's conformational ensemble in solution presents viable candidates for protein binding. The experimental solution profile can be achieved with the NAMFIS (NMR analysis of molecular flexibility in solution) method, which deconvolutes the average NMR spectrum of small flexible molecules into individual contributing conformations with varying populations. Geldanamycin and radicicol are structurally different macrocycles determined by X-ray crystallography to bind to a common site on the cellular chaperone heat shock protein 90 (Hsp90). Without benefit of a receptor structure, NAMFIS has identified the bioactive conformers of geldanamycin and radicicol in CDCl3 solution with populations of 4% and 21%, respectively. Conversely, docking the set of NAMFIS conformers into the unliganded proteins with GLIDE followed by MM-GBSA scoring reproduces the experimental crystallographic binding poses.     

Rational design,synthesis, and biological evaluation of novel C6-modified geldanamycin derivatives as potent Hsp90 inhibitors and anti-tumor agents

Heat shock protein 90 (Hsp90) is an appealing anticancer drug target that provoked a tremendous wave of investigations. Geldanamycin (GA) is the first identified Hsp90 inhibitor that exhibited potent anti-cancer activity, but the off-target toxicity associated with the benzoquinone moiety hampered its clinical application. Until now, structure optimization of GA is still in need to fully exploit the therapeutic value of Hsp90. Due to the structural complexity and synthetic challenge of this compound family, conventional optimization is bound to be costly but high efficiency is expected to be reachable by combining the art of rational design and total synthesis. Described in this paper is our first attempt at this approach aiming at rational modification of the C6-position of GA. The binding affinities towards Hsp90 of compound 1 (C6-ethyl) and 2 (C6-methyl) were designed and predicted by using Discovery Studio. These compounds were synthesized and further subjected to a thorough in vitro biological evaluation. We found that compounds 1 and 2 bind to Hsp90 protein with the IC₅₀ of 34.26 nmol/L and 163.7 nmol/L, respectively. Both compounds showed broad-spectrum antitumor effects. Replacing by ethyl, compound 1 exhibited more potent bioactivity than positive control GA, such as in G2/M cell cycle arrest, cell apoptosis and client proteins degradations. The results firstly indicated that the docking study is able to provide a precise prediction of Hsp90 affinities of GA analogues, and the C6 substituent of GA is not erasable without affecting its biological activity.     

Free-energy barriers in MbCO rebinding

The rebinding of CO to myoglobin (Mb) from locations around the active site is studied using a combination of molecular dynamics and stochastic simulations for native and L29F mutant Mb. The interaction between the dissociated ligand and the protein environment is described by the recently developed fluctuating three-point charge model for the CO molecule. Umbrella sampling along trajectories, previously found to sample the binding site (B) and the Xe4 pocket, is used to construct free-energy profiles for the ligand escape. On the basis of the Smoluchowski equation, the relaxation of different initial population distributions is followed in space and time. For native Mb at room temperature, the calculated rebinding times are in good agreement with experimental values and give an inner barrier of 4.3 kcal/mol between the docking site B (Mb...CO) and the A state (bound MbCO), compared to an effective barrier, Heff, of 4.5 kcal/mol and barriers into the majority conformation A1 and the minority conformation A3 of 2.4 and 4.3 kcal/mol, respectively. In the case of the L29F mutant, the free-energy surface is flatter and the dynamics is much more rapid. As was found in experiment, escape to the Xe4 pocket is facile for L29F whereas, for native Mb, the barriers to this site are larger. At lower temperatures, the rebinding dynamics is delayed by orders of magnitude also due to increased barriers between the docking sites.     

Activation of Hsp90 Enzymatic Activity and Conformational Dynamics through Rationally Designed Allosteric Ligands

Hsp90 is a molecular chaperone of pivotal importance for multiple cell pathways. ATP‐regulated internal dynamics are critical for its function and current pharmacological approaches block the chaperone with ATP‐competitive inhibitors. Herein, a general approach to perturb Hsp90 through design of new allosteric ligands aimed at modulating its functional dynamics is proposed. Based on the characterization of a first set of 2‐phenylbenzofurans showing stimulatory effects on Hsp90 ATPase and conformational dynamics, new ligands were developed that activate Hsp90 by targeting an allosteric site, located 65 Å from the active site. Specifically, analysis of protein responses to first‐generation activators was exploited to guide the design of novel derivatives with improved ability to stimulate ATP hydrolysis. The molecules’ effects on Hsp90 enzymatic, conformational, co‐chaperone and client‐binding properties were characterized through biochemical, biophysical and cellular approaches. These designed probes act as allosteric activators of the chaperone and affect the viability of cancer cell lines for which proper functioning of Hsp90 is necessary.     

A scoring function for docking ligands to low-resolution protein structures

We present a docking method that uses a scoring function for protein-ligand docking that is designed to maximize the docking success rate for low-resolution protein structures. We find that the resulting scoring function parameters are very different depending on whether they were optimized for high- or low-resolution protein structures. We show that this docking method can be successfully applied to predict the ligand-binding site of low-resolution structures. For a set of 25 protein-ligand complexes, in 76% of the cases, more than 50% of ligand-contacting residues are correctly predicted (using receptor crystal structures where the binding site is unspecified). Using decoys of the receptor structures having a 4 A RMSD from the native structure, for the same set of complexes, in 72% of the cases, we obtain at least one correctly predicted ligand-contacting residue. Furthermore, using an 81-protein-ligand set described by Jain, in 76 (93.8%) cases, the algorithm correctly predicts more than 50% of the ligand-contacting residues when native protein structures are used. Using 3 A RMSD from native decoys, in all but two cases (97.5%), the algorithm predicts at least one ligand-binding residue correctly. Finally, compared to the previously published Dolores method, for 298 protein-ligand pairs, the number of cases in which at least half of the specific contacts are correctly predicted is more than four times greater.     

Docking-undocking combination applied to the D3R Grand Challenge 2015

Novel methods for drug discovery are constantly under development and independent exercises to test and validate them for different goals are extremely useful. The drug discovery data resource (D3R) Grand Challenge 2015 offers an excellent opportunity as an external assessment and validation experiment for Computer-Aided Drug Discovery methods. The challenge comprises two protein targets and prediction tests: binding mode and ligand ranking. We have faced both of them with the same strategy: pharmacophore-guided docking followed by dynamic undocking (a new method tested experimentally here) and, where possible, critical assessment of the results based on pre-existing information. In spite of using methods that are qualitative in nature, our results for binding mode and ligand ranking were amongst the best on Hsp90. Results for MAP4K4 were less positive and we track the different performance across systems to the level of previous knowledge about accessible conformational states. We conclude that docking is quite effective if supplemented by dynamic undocking and empirical information (e.g. binding hot spots, productive protein conformations). This setup is well suited for virtual screening, a frequent application that was not explicitly tested in this edition of the D3R Grand Challenge 2015. Protein flexibility remains as the main cause for hard failures.     

Protein-protein binding sites prediction by 3D structural similarities

Identifying the location of binding sites on proteins is of fundamental importance for a wide range of applications including molecular docking, de novo drug design, structure identification, and comparison of functional sites. In this paper, we develop an efficient approach for finding binding sites between proteins. Our approach consists of four steps: local sequence alignment, protein surface detection, 3D structure comparison, and candidate binding site selection. A comparison of our method with the LSA algorithm shows that the binding sites predicted by our method are somewhat closer to the actual binding sites in the protein-protein complexes. The software package is available at http://sites.google.com/site/guofeics/pro-bs for noncommercial use.     

Artificial Accelerators of the Molecular Chaperone Hsp90 Facilitate Rate‐Limiting Conformational Transitions

The molecular chaperone Hsp90 undergoes an ATP‐driven cycle of conformational changes in which large structural rearrangements precede ATP hydrolysis. Well‐established small‐molecule inhibitors of Hsp90 compete with ATP‐binding. We wondered whether compounds exist that can accelerate the conformational cycle. In a FRET‐based screen reporting on conformational rearrangements in Hsp90 we identified compounds. We elucidated their mode of action and showed that they can overcome the intrinsic inhibition in Hsp90 which prevents these rearrangements. The mode of action is similar to that of the co‐chaperone Aha1 which accelerates the Hsp90 ATPase. However, while the two identified compounds influence conformational changes, they target different aspects of the structural transitions. Also, the binding site determined by NMR spectroscopy is distinct. This study demonstrates that small molecules are capable of triggering specific rate‐limiting transitions in Hsp90 by mechanisms similar to those in protein cofactors.     

Hierarchical dynamics in allostery following ATP hydrolysis monitored by single molecule FRET measurements and MD simulations

We report on a study that combines advanced fluorescence methods with molecular dynamics (MD) simulations to cover timescales from nanoseconds to milliseconds for a large protein. This allows us to delineate how ATP hydrolysis in a protein causes allosteric changes at a distant protein binding site, using the chaperone Hsp90 as test system. The allosteric process occurs via hierarchical dynamics involving timescales from nano- to milliseconds and length scales from Ångstroms to several nanometers. We find that hydrolysis of one ATP is coupled to a conformational change of Arg380, which in turn passes structural information via the large M-domain α-helix to the whole protein. The resulting structural asymmetry in Hsp90 leads to the collapse of a central folding substrate binding site, causing the formation of a novel collapsed state (closed state B) that we characterise structurally. We presume that similar hierarchical mechanisms are fundamental for information transfer induced by ATP hydrolysis through many other proteins.

We report on a study that combines advanced fluorescence methods with molecular dynamics simulations to cover timescales from nanoseconds to milliseconds for a large protein, the chaperone Hsp90.     

A QXP-based multistep docking procedure for accurate prediction of protein-ligand complexes

The two great challenges of the docking process are the prediction of ligand poses in a protein binding site and the scoring of the docked poses. Ligands that are composed of extended chains in their molecular structure display the most difficulties, predominantly because of the torsional flexibility. On the basis of the molecular docking program QXP-Flo+0802, we have developed a procedure particularly for ligands with a high degree of rotational freedom that allows the accurate prediction of the orientation and conformation of ligands in protein binding sites. Starting from an initial full Monte Carlo docking experiment, this was achieved by performing a series of successive multistep docking runs using a local Monte Carlo search with a restricted rotational angle, by which the conformational search space is limited. The method was established by using a highly flexible acetylcholinesterase inhibitor and has been applied to a number of challenging protein-ligand complexes known from the literature.     

Functional concerted motions in the bovine serum retinol-binding protein

The large concerted motions in the apo/holo bovine serum retinol- binding protein were studied using molecular dynamics simulation and 'essential dynamics' analysis. Initially, concerted motions were calculated from conformational differences between various crystal structures. The dynamic behaviour of the protein in the configurational space directions, described by these concerted motions, is analysed. This reveals that the large backbone dynamics of the protein is not influenced by the presence of retinol. Study of free retinol dynamics and retinol in the retinol binding site reveals that the protein binds retinol in a favourable conformation, as opposed to what has been previously described for the bovine cellular retinol-binding protein.     

Evaluation of a knowledge‐based potential of mean force for scoring docked protein–ligand complexes

The Biomolecular Ligand Energy Evaluation Protocol (BLEEP) is a knowledge‐based potential derived from high‐resolution X‐ray structures of protein–ligand complexes. The performance of this potential in ranking the hypothetical structures resulting from a docking study has been evaluated using fifteen protein–ligand complexes from the Protein Data Bank. In the majority of complexes BLEEP was successful in identifying the native (experimental) binding mode or an alternative of low rms deviation (from the native) as the lowest in energy. Overall BLEEP is slightly better than the DOCK energy function in discriminating native‐like modes. Even when alternative binding modes rank lower than the native structure, a reasonable energy is assigned to the latter. Breaking down the BLEEP scores into the atom–atom contributions reveals that this type of potential is grossly dominated by longer range interactions (>5 Å), which makes it relatively insensitive to small local variations in the binding site. However, despite this limitation, the lack, at present, of accurate protein–ligand potentials means that BLEEP is a promising approach to improve the filtering of structures resulting from docking programs. Moreover, BLEEP should improve with the continuously increasing number of complexes available in the PDB. © 2001 John Wiley & Sons, Inc. J Comput Chem 22: 673–688, 2001     

Predicting the accuracy of protein–ligand docking on homology models

Ligand–protein docking is increasingly used in Drug Discovery. The initial limitations imposed by a reduced availability of target protein structures have been overcome by the use of theoretical models, especially those derived by homology modeling techniques. While this greatly extended the use of docking simulations, it also introduced the need for general and robust criteria to estimate the reliability of docking results given the model quality. To this end, a large‐scale experiment was performed on a diverse set including experimental structures and homology models for a group of representative ligand–protein complexes. A wide spectrum of model quality was sampled using templates at different evolutionary distances and different strategies for target–template alignment and modeling. The obtained models were scored by a selection of the most used model quality indices. The binding geometries were generated using AutoDock, one of the most common docking programs. An important result of this study is that indeed quantitative and robust correlations exist between the accuracy of docking results and the model quality, especially in the binding site. Moreover, state‐of‐the‐art indices for model quality assessment are already an effective tool for an a priori prediction of the accuracy of docking experiments in the context of groups of proteins with conserved structural characteristics. © 2010 Wiley Periodicals, Inc. J Comput Chem, 2010     

Can we trust docking results? Evaluation of seven commonly used programs on PDBbind database

Docking is one of the most commonly used techniques in drug design. It is used for both identifying correct poses of a ligand in the binding site of a protein as well as for the estimation of the strength of protein–ligand interaction. Because millions of compounds must be screened, before a suitable target for biological testing can be identified, all calculations should be done in a reasonable time frame. Thus, all programs currently in use exploit empirically based algorithms, avoiding systematic search of the conformational space. Similarly, the scoring is done using simple equations, which makes it possible to speed up the entire process. Therefore, docking results have to be verified by subsequent in vitro studies. The purpose of our work was to evaluate seven popular docking programs (Surflex, LigandFit, Glide, GOLD, FlexX, eHiTS, and AutoDock) on the extensive dataset composed of 1300 protein–ligands complexes from PDBbind 2007 database, where experimentally measured binding affinity values were also available. We compared independently the ability of proper posing [according to Root mean square deviation (or Root mean square distance) of predicted conformations versus the corresponding native one] and scoring (by calculating the correlation between docking score and ligand binding strength). To our knowledge, it is the first large‐scale docking evaluation that covers both aspects of docking programs, that is, predicting ligand conformation and calculating the strength of its binding. More than 1000 protein–ligand pairs cover a wide range of different protein families and inhibitor classes. Our results clearly showed that the ligand binding conformation could be identified in most cases by using the existing software, yet we still observed the lack of universal scoring function for all types of molecules and protein families. © 2010 Wiley Periodicals, Inc. J Comput Chem, 2011     

Methods for the prediction of protein-ligand binding sites for structure-based drug design and virtual ligand screening

Structure Based Drug Design (SBDD) is a computational approach to lead discovery that uses the three-dimensional structure of a protein to fit drug-like molecules into a ligand binding site to modulate function. Identifying the location of the binding site is therefore a vital first step in this process, restricting the search space for SBDD or virtual screening studies. The detection and characterisation of functional sites on proteins has increasingly become an area of interest. Structural genomics projects are increasingly yielding protein structures with unknown functions and binding sites. Binding site prediction was pioneered by pocket detection, since the binding site is often found in the largest pocket. More recent methods involve phylogenetic analysis, identifying structural similarity with proteins of known function and identifying regions on the protein surface with a potential for high binding affinity. Binding site prediction has been used in several SBDD projects and has been incorporated into several docking tools. We discuss different methods of ligand binding site prediction, their strengths and weaknesses, and how they have been used in SBDD.     

Design, synthesis, and structure--activity relationships for chimeric inhibitors of Hsp90

Inhibition of the 90 kDa heat shock protein (Hsp90) family of molecular chaperones represents a promising new chemotherapeutic approach toward the treatment of several cancers. Previous studies have demonstrated that the natural products, radicicol and geldanamycin, are potent inhibitors of the Hsp90 N-terminal ATP binding site. The cocrystal structures of these molecules bound to Hsp90 have been determined, and through molecular modeling and superimposition of these ligands, hybrids of radicicol and geldanamycin have been designed. A series of macrocylic chimeras of radicicol and geldanamycin and the corresponding seco-agents have been prepared and evaluated for both antiproliferative activity and their ability to induce Hsp90-dependent client protein degradation.     

Exploring Weak Ligand–Protein Interactions by Long‐Lived NMR States: Improved Contrast in Fragment‐Based Drug Screening

Ligands that have an affinity for protein targets can be screened very effectively by exploiting favorable properties of long‐lived states (LLS) in NMR spectroscopy. In this work, we describe the use of LLS for competitive binding experiments to measure accurate dissociation constants of fragments that bind weakly to the ATP binding site of the N‐terminal ATPase domain of heat shock protein 90 (Hsp90), a therapeutic target for cancer treatment. The LLS approach allows one to characterize ligands with an exceptionally wide range of affinities, since it can be used for ligand concentrations [L] that are several orders of magnitude smaller than the dissociation constants K_D. This property makes the LLS method particularly attractive for the initial steps of fragment‐based drug screening, where small molecular fragments that bind weakly to a target protein must be identified, which is a difficult task for many other biophysical methods.     

Crystal structure and molecular modeling of 17-DMAG in complex with human Hsp90

Hsp90 is an attractive chemotherapeutic target because it chaperones the folding of proteins found in multiple signal transduction pathways. We describe the 1.75 A resolution crystal structure of human Hsp90 alpha (residues 9-236) complexed with 17-desmethoxy-17-N,N-dimethylaminoethylamino-geldanamycin (17-DMAG). The structure revealed an altered set of interactions between the 17-substituent and the protein compared to geldanamycin and the 17-dimethylaminoethyl moiety pointing into solvent, but otherwise was similar to that reported for the complex with geldanamycin. Targeted molecular dynamics simulations and energetic analysis indicate that geldanamycin undergoes two major conformational changes when it binds Hsp90, with the key step of the conversion being the trans to cis conformational change of the macrocycle amide bond. We speculate that 17-DMAG analogs constrained to a cis-amide in the ground state could provide a significant increase in affinity for Hsp90.