Protein-ligand X-ray Structures Research Articles

The Good, the Bad, and the Twisted Revisited: An Analysis of Ligand Geometry in Highly Resolved Protein-Ligand X-ray Structures.

An analysis of the rotatable bond geometry of drug-like ligand models is reported for high-resolution (<1.1 Å) crystallographic protein-ligand complexes. In cases where the ligand fit to the electron density is very good, unusual torsional geometry is rare and, most often, though not exclusively, associated with strong polar, metal, or covalent ligand-protein interactions. It is rarely associated with a torsional strain of greater than 2 kcal mol-1 by calculation. An unusual torsional geometry is more prevalent where the fit to electron density is not perfect. Multiple low-strain conformer bindings were observed in 21% of the set and, it is suggested, may also lie behind many of the 35% of single-occupancy cases, where a poor fit to the e-density was found. It is concluded that multiple conformer ligand binding is an under-recognized phenomenon in structure-based drug design and that there is a need for more robust crystallographic refinement methods to better handle such cases.

Hidden bias in the DUD-E dataset leads to misleading performance of deep learning in structure-based virtual screening.

Recently much effort has been invested in using convolutional neural network (CNN) models trained on 3D structural images of protein-ligand complexes to distinguish binding from non-binding ligands for virtual screening. However, the dearth of reliable protein-ligand x-ray structures and binding affinity data has required the use of constructed datasets for the training and evaluation of CNN molecular recognition models. Here, we outline various sources of bias in one such widely-used dataset, the Directory of Useful Decoys: Enhanced (DUD-E). We have constructed and performed tests to investigate whether CNN models developed using DUD-E are properly learning the underlying physics of molecular recognition, as intended, or are instead learning biases inherent in the dataset itself. We find that superior enrichment efficiency in CNN models can be attributed to the analogue and decoy bias hidden in the DUD-E dataset rather than successful generalization of the pattern of protein-ligand interactions. Comparing additional deep learning models trained on PDBbind datasets, we found that their enrichment performances using DUD-E are not superior to the performance of the docking program AutoDock Vina. Together, these results suggest that biases that could be present in constructed datasets should be thoroughly evaluated before applying them to machine learning based methodology development.

KRAS Binders Hidden in Nature.

Natural products have proven to be a rich source of molecular architectures for drugs. Here, an integrated approach to natural product screening is proposed, which uncovered eight new natural product scaffolds for KRAS—the most frequently mutated oncogenic driver in human cancers, which has remained thus far undrugged. The approach combines aspects of virtual screening, fragment‐based screening, structure‐activity relationships (SAR) by NMR, and structure‐based drug discovery to overcome the limitations in traditional natural product approaches. By using our approach, a new “snugness of fit” scoring function and the first crystal‐soaking system of the active form of KRASG12D, the protein–ligand X‐ray structures of a tricyclic indolopyrrole fungal alkaloid and an indoloisoquinolinone have been successfully elucidated. The natural product KRAS hits discovered provide fruitful ground for the optimization of highly potent natural‐product‐based inhibitors of the active form of oncogenic RAS. This integrated approach for screening natural products also holds promise for other “undruggable” targets.

Thermodynamic Characterization of Hydration Sites from Integral Equation-Derived Free Energy Densities: Application to Protein Binding Sites and Ligand Series.

Water molecules play an essential role for mediating interactions between ligands and protein binding sites. Displacement of specific water molecules can favorably modulate the free energy of binding of protein-ligand complexes. Here, the nature of water interactions in protein binding sites is investigated by 3D RISM (three-dimensional reference interaction site model) integral equation theory to understand and exploit local thermodynamic features of water molecules by ranking their possible displacement in structure-based design. Unlike molecular dynamics-based approaches, 3D RISM theory allows for fast and noise-free calculations using the same detailed level of solute-solvent interaction description. Here we correlate molecular water entities instead of mere site density maxima with local contributions to the solvation free energy using novel algorithms. Distinct water molecules and hydration sites are investigated in multiple protein-ligand X-ray structures, namely streptavidin, factor Xa, and factor VIIa, based on 3D RISM-derived free energy density fields. Our approach allows the semiquantitative assessment of whether a given structural water molecule can potentially be targeted for replacement in structure-based design. Finally, PLS-based regression models from free energy density fields used within a 3D-QSAR approach (CARMa - comparative analysis of 3D RISM Maps) are shown to be able to extract relevant information for the interpretation of structure-activity relationship (SAR) trends, as demonstrated for a series of serine protease inhibitors.

Efficient conformational sampling and weak scoring in docking programs? Strategy of the wisdom of crowds

BackgroundIn drug design, an efficient structure-based optimization of a ligand needs the precise knowledge of the protein–ligand interactions. In the absence of experimental information, docking programs are necessary for ligand positioning, and the choice of a reliable program is essential for the success of such an optimization. The performances of four popular docking programs, Gold, Glide, Surflex and FlexX, were investigated using 100 crystal structures of complexes taken from the Directory of Useful Decoys-Enhanced database.ResultsThe ligand conformational sampling was rather efficient, with a correct pose found for a maximum of 84 complexes, obtained by Surflex. However, the ranking of the correct poses was not as efficient, with a maximum of 68 top-rank or 75 top-4 rank correct poses given by Glidescore. No relationship was found between either the sampling or the scoring performance of the four programs and the properties of either the targets or the small molecules, except for the number of ligand rotatable bonds. As well, no exploitable relationship was found between each program performance in docking and in virtual screening; a wrong top-rank pose may obtain a good score that allows it to be ranked among the most active compounds and vice versa. Also, to improve the results of docking, the strengths of the programs were combined either by using a rescoring procedure or the United Subset Consensus (USC). Oddly, positioning with Surflex and rescoring with Glidescore did not improve the results. However, USC based on docking allowed us to obtain a correct pose in the top-4 rank for 87 complexes. Finally, nine complexes were scrutinized, because a correct pose was found by at least one program but poorly ranked by all four programs. Contrarily to what was expected, except for one case, this was not due to weaknesses of the scoring functions.ConclusionsWe conclude that the scoring functions should be improved to detect the correct poses, but sometimes their failure may be due to other varied considerations. To increase the chances of success, we recommend to use several programs and combine their results.Graphical abstractSummary of the results obtained by semi-rigid docking of crystallographic ligands. The docking was done on 100 protein-ligand X-ray structures, taken from the DUD-E database, and using four programs, Glide, Gold, Surflex and FlexX. Based on the docking results, we applied our United Subset Consensus method (USC), for which only the top4-rank poses are relevant. The number of complexes for which the best pose is correct, is represented by the gray boxes, the blue and red boxes correspond to the number of complexes with a correct pose ranked as the top 1 or within the top 4. A pose is considered correct when its root-mean-square deviation from the crystal structure is less than 2 Å

Fluorine multipolar interaction: Toward elucidating its energetics in binding recognition

Multipolar interactions have gained attention due to their roles in molecular recognition events of chemical and biological systems. Structural evidence suggests that the fluorine multipolar interaction is a favorable molecular interaction in the paradigm of protein-ligand recognition. With an aim to understand its propensity and energetics in ligand-protein assembly, we mined the “Pfizer protein-ligand X-ray structure database” with three-dimensional constraints for the presence of such interactions. A set of transformation rules were applied to generate their corresponding matched molecular pairs (MMPs) that bear hydrogen(s) in place of the interacting fluorine(s). Biological activities were retrieved from our internal data warehouse for the MMPs with and without the ability to form the multipolar interaction. On the basis of the observed potency differences we determined that the free energy gain associated with the fluorine multipolar interaction is relatively modest, ∼0.3–0.6kcal/mol, and the net impact on lipophilic efficiency (LipE) is essentially neutral. A general guideline for medicinal chemists has emerged from this study, enabling a more rational employment of this type of interaction in molecular design.

Structure-Based Drug Design of Mineralocorticoid Receptor Antagonists to Explore Oxosteroid Receptor Selectivity.

The mineralocorticoid receptor (MR) is a nuclear hormone receptor involved in the regulation of body fluid and electrolyte homeostasis. In this study we explore selectivity triggers for a series of nonsteroidal MR antagonists to improve selectivity over other members of the oxosteroid receptor family. A biaryl sulfonamide compound was identified in a high-throughput screening (HTS) campaign. The compound bound to MR with pKi =6.6, but displayed poor selectivity over the glucocorticoid receptor (GR) and the progesterone receptor (PR). Following X-ray crystallography of MR in complex with the HTS hit, a compound library was designed that explored an induced-fit hypothesis that required movement of the Met852 side chain. An improvement in MR selectivity of 11- to 79-fold over PR and 23- to 234-fold over GR was obtained. Given the U-shaped binding conformation, macrocyclizations were explored, yielding a macrocycle that bound to MR with pKi =7.3. Two protein-ligand X-ray structures were determined, confirming the hypothesized binding mode for the designed compounds.

Docking pose selection by interaction pattern graph similarity: application to the D3R grand challenge 2015.

High affinity ligands for a given target tend to share key molecular interactions with important anchoring amino acids and therefore often present quite conserved interaction patterns. This simple concept was formalized in a topological knowledge-based scoring function (GRIM) for selecting the most appropriate docking poses from previously X-rayed interaction patterns. GRIM first converts protein-ligand atomic coordinates (docking poses) into a simple 3D graph describing the corresponding interaction pattern. In a second step, proposed graphs are compared to that found from template structures in the Protein Data Bank. Last, all docking poses are rescored according to an empirical score (GRIMscore) accounting for overlap of maximum common subgraphs. Taking the opportunity of the public D3R Grand Challenge 2015, GRIM was used to rescore docking poses for 36 ligands (6 HSP90α inhibitors, 30 MAP4K4 inhibitors) prior to the release of the corresponding protein-ligand X-ray structures. When applied to the HSP90α dataset, for which many protein-ligand X-ray structures are already available, GRIM provided very high quality solutions (mean rmsd=1.06Å, n=6) as top-ranked poses, and significantly outperformed a state-of-the-art scoring function. In the case of MAP4K4 inhibitors, for which preexisting 3D knowledge is scarce and chemical diversity is much larger, the accuracy of GRIM poses decays (mean rmsd=3.18 Å, n=30) although GRIM still outperforms an energy-based scoring function. GRIM rescoring appears to be quite robust with comparison to the other approaches competing for the same challenge (42 submissions for the HSP90 dataset, 27 for the MAP4K4 dataset) as it ranked 3rd and 2nd respectively, for the two investigated datasets. The rescoring method is quite simple to implement, independent on a docking engine, and applicable to any target for which at least one holo X-ray structure is available.

MTH1 Substrate Recognition--An Example of Specific Promiscuity.

MTH1 (NUDT1) is an oncologic target involved in the prevention of DNA damage. We investigate the way MTH1 recognises its substrates and present substrate-bound structures of MTH1 for 8-oxo-dGTP and 8-oxo-rATP as examples of novel strong and weak binding substrate motifs. Investigation of a small set of purine-like fragments using 2D NMR resulted in identification of a fragment with weak potency. The protein-ligand X-Ray structure of this fragment provides insight into the role of water molecules in substrate selectivity. Wider fragment screening by NMR resulted in three new protein structures exhibiting alternative binding configurations to the key Asp-Asp recognition element of the protein. These inhibitor binding modes demonstrate that MTH1 employs an intricate yet promiscuous mechanism of substrate anchoring through its Asp-Asp pharmacophore. The structures suggest that water-mediated interactions convey selectivity towards oxidized substrates over their non-oxidised counterparts, in particular by stabilization of a water molecule in a hydrophobic environment through hydrogen bonding. These findings may be useful in the design of inhibitors of MTH1.

Beware of machine learning-based scoring functions-on the danger of developing black boxes.

Training machine learning algorithms with protein-ligand descriptors has recently gained considerable attention to predict binding constants from atomic coordinates. Starting from a series of recent reports stating the advantages of this approach over empirical scoring functions, we could indeed reproduce the claimed superiority of Random Forest and Support Vector Machine-based scoring functions to predict experimental binding constants from protein-ligand X-ray structures of the PDBBind dataset. Strikingly, these scoring functions, trained on simple protein-ligand element-element distance counts, were almost unable to enrich virtual screening hit lists in true actives upon docking experiments of 10 reference DUD-E datasets; this is a a feature that, however, has been verified for an a priori less-accurate empirical scoring function (Surflex-Dock). By systematically varying ligand poses from true X-ray coordinates, we show that the Surflex-Dock scoring function is logically sensitive to the quality of docking poses. Conversely, our machine-learning based scoring functions are totally insensitive to docking poses (up to 10 Å root-mean square deviations) and just describe atomic element counts. This report does not disqualify using machine learning algorithms to design scoring functions. Protein-ligand element-element distance counts should however be used with extreme caution and only applied in a meaningful way. To avoid developing novel but meaningless scoring functions, we propose that two additional benchmarking tests must be systematically done when developing novel scoring functions: (i) sensitivity to docking pose accuracy, and (ii) ability to enrich hit lists in true actives upon structure-based (docking, receptor-ligand pharmacophore) virtual screening of reference datasets.

Predicting Ligand Binding Modes from Neural Networks Trained on Protein–Ligand Interaction Fingerprints

We herewith present a novel approach to predict protein-ligand binding modes from the single two-dimensional structure of the ligand. Known protein-ligand X-ray structures were converted into binary bit strings encoding protein-ligand interactions. An artificial neural network was then set up to first learn and then predict protein-ligand interaction fingerprints from simple ligand descriptors. Specific models were constructed for three targets (CDK2, p38-α, HSP90-α) and 146 ligands for which protein-ligand X-ray structures are available. These models were able to predict protein-ligand interaction fingerprints and to discriminate important features from minor interactions. Predicted interaction fingerprints were successfully used as descriptors to discriminate true ligands from decoys by virtual screening. In some but not all cases, the predicted interaction fingerprints furthermore enable to efficiently rerank cross-docking poses and prioritize the best possible docking solutions.

Cover Picture: Probing Multidrug‐Resistance and Protein–Ligand Interactions with Oxatricyclic Designed Ligands in HIV‐1 Protease Inhibitors (ChemMedChem 11/2010)

The cover picture shows the overlay of the protein–ligand X-ray structures of darunavir and saquinivir-bound HIV-1 protease. This structural architecture provided the inspiration for a new tris-THF ligand that would both maintain the “backbone binding” interactions of darunavir, as well as fill the S3 subsite comparable to saquinavir. GRL-0519A contains a stereochemically defined syn-anti-syn architecture in the P2 ligand. The Purdue University Bell Tower (shown on the right-hand side) also exemplifies a beautiful and functional structural architecture. It has long been a symbol of the will to ever strive for that next greater achievement. For more details, see the Communication by Arun K. Ghosh et al. on p. 1850 ff. Cover art credit to: Dr. Xiaoming Xu and Mr. Steve Scherer.

Novel Protease Inhibitors (PIs) Containing Macrocyclic Components and 3( R ),3a( S ),6a( R )- bis -Tetrahydrofuranylurethane That Are Potent against Multi-PI-Resistant HIV-1 Variants In Vitro

Natural products with macrocyclic structural features often display intriguing biological properties. Molecular design incorporating macrocycles may lead to molecules with unique protein-ligand interactions. We generated novel human immunodeficiency virus type 1 (HIV-1) protease inhibitors (PIs) containing a macrocycle and bis-tetrahydrofuranylurethane. Four such compounds exerted potent activity against HIV-1LAI and had 50% effective concentrations (EC50s) of as low as 0.002 microM with minimal cytotoxicity. GRL-216 and GRL-286 blocked the replication of HIV-1NL4-3 variants selected by up to 5 microM saquinavir, ritonavir, nelfinavir, lopinavir, or atazanavir; they had EC50s of 0.020 to 0.046 microM and potent activities against six multi-PI-resistant clinical HIV-1 (HIVmPIr) variants with EC50s of 0.027 to 0.089 microM. GRL-216 and -286 also blocked HIV-1 protease dimerization as efficiently as darunavir. When HIV-1NL4-3 was selected by GRL-216, it replicated progressively more poorly and failed to replicate in the presence of >0.26 microM GRL-216, suggesting that the emergence of GRL-216-resistant HIV-1 variants is substantially delayed. At passage 50 with GRL-216 (the HIV isolate selected with GRL-216 at up to 0.16 microM [HIV216-0.16 microM]), HIV-1NL4-3 containing the L10I, L24I, M46L, V82I, and I84V mutations remained relatively sensitive to PIs, including darunavir, with the EC50s being 3- to 8-fold-greater than the EC50 of each drug for HIV-1NL4-3. Interestingly, HIV216-0.16 microM had 10-fold increased sensitivity to tipranavir. Analysis of the protein-ligand X-ray structures of GRL-216 revealed that the macrocycle occupied a greater volume of the binding cavity of protease and formed greater van der Waals interactions with V82 and I84 than darunavir. The present data warrant the further development of GRL-216 as a potential antiviral agent for treating individuals harboring wild-type and/or HIVmPIr.

Design, synthesis, protein-ligand X-ray structure, and biological evaluation of a series of novel macrocyclic human immunodeficiency virus-1 protease inhibitors to combat drug resistance.

The structure-based design, synthesis, and biological evaluation of a series of nonpeptidic macrocyclic HIV protease inhibitors are described. The inhibitors are designed to effectively fill in the hydrophobic pocket in the S1'-S2' subsites and retain all major hydrogen bonding interactions with the protein backbone similar to darunavir (1) or inhibitor 2. The ring size, the effect of methyl substitution, and unsaturation within the macrocyclic ring structure were assessed. In general, cyclic inhibitors were significantly more potent than their acyclic homologues, saturated rings were less active than their unsaturated analogues and a preference for 10- and 13-membered macrocylic rings was revealed. The addition of methyl substituents resulted in a reduction of potency. Both inhibitors 14b and 14c exhibited marked enzyme inhibitory and antiviral activity, and they exerted potent activity against multidrug-resistant HIV-1 variants. Protein-ligand X-ray structures of inhibitors 2 and 14c provided critical molecular insights into the ligand-binding site interactions.

Potent HIV-1 protease inhibitors incorporating meso-bicyclic urethanes as P2-ligands: structure-based design, synthesis, biological evaluation and protein–ligand X-ray studies

Recently, we designed a series of novel HIV-1 protease inhibitors incorporating a stereochemically defined bicyclic fused cyclopentyl (Cp-THF) urethane as the high affinity P2-ligand. Inhibitor with this P2-ligand has shown very impressive potency against multi-drug-resistant clinical isolates. Based upon the -bound HIV-1 protease X-ray structure, we have now designed and synthesized a number of meso-bicyclic ligands which can conceivably interact similarly to the Cp-THF ligand. The design of meso-ligands is quite attractive as they do not contain any stereocenters. Inhibitors incorporating urethanes of bicyclic-1,3-dioxolane and bicyclic-1,4-dioxane have shown potent enzyme inhibitory and antiviral activities. Inhibitor (K(i) = 0.11 nM; IC(50) = 3.8 nM) displayed very potent antiviral activity in this series. While inhibitor showed comparable enzyme inhibitory activity (K(i) = 0.18 nM) its antiviral activity (IC(50) = 170 nM) was significantly weaker than inhibitor . Inhibitor maintained an antiviral potency against a series of multi-drug resistant clinical isolates comparable to amprenavir. A protein-ligand X-ray structure of -bound HIV-1 protease revealed a number of key hydrogen bonding interactions at the S2-subsite. We have created an active model of inhibitor based upon this X-ray structure.

Potent memapsin 2 (β-secretase) inhibitors: Design, synthesis, protein-ligand X-ray structure, and in vivo evaluation

Structure-based design, synthesis, and biological evaluation of a series of peptidomimetic beta-secretase inhibitors incorporating hydroxyethylamine isosteres are described. We have identified inhibitor 24 which has shown exceedingly potent activity in memapsin 2 enzyme inhibitory (K(i) 1.8 nM) and cellular (IC(50)=1 nM in Chinese hamster ovary cells) assays. Inhibitor 24 has also shown very impressive in vivo properties (up to 65% reduction of plasma A beta) in transgenic mice. The X-ray structure of protein-ligand complex of memapsin 2 revealed critical interactions in the memapsin 2 active site.

Structure-based drug design (SBDD): Every structure tells a story...

In this article we report how protein-ligand X-ray structures can be used to identify and develop new drug candidates. The structure-based drug design approach (SBDD) is described, illustrating the way crystallographic information is utilized in an iterative manner to identify and optimize novel lead compounds, thereby accelerating the overall drug discovery process. The requirements for the effective use of SBDD and its implementation, drawing upon case histories from our Thymidylate Synthase (TS) research program, are described.