Eelke B. Lenselink

Polypharmacology modelling using proteochemometrics (PCM): recent methodological developments, applications to target families, and future prospects

Proteochemometric (PCM) modelling is a computational method to model the bioactivity of multiple ligands against multiple related protein targets simultaneously. Hence it has been found to be particularly useful when exploring the selectivity and promiscuity of ligands on different proteins. In this review, we will firstly provide a brief introduction to the main concepts of PCM for readers new to the field. The next part focuses on recent technical advances, including the application of support vector machines (SVMs) using different kernel functions, random forests, Gaussian processes and collaborative filtering. The subsequent section will then describe some novel practical applications of PCM in the medicinal chemistry field, including studies on GPCRs, kinases, viral proteins (e.g. from HIV) and epigenetic targets such as histone deacetylases. Finally, we will conclude by summarizing novel developments in PCM, which we expect to gain further importance in the future. These developments include adding three-dimensional protein target information, application of PCM to the prediction of binding energies, and application of the concept in the fields of pharmacogenomics and toxicogenomics. This review is an update to a related publication in 2011 and it mainly focuses on developments in the field since then.

Selecting an optimal number of binding site waters to improve virtual screening enrichments against the adenosine A2A receptor.

A major challenge in structure-based virtual screening (VS) involves the treatment of explicit water molecules during docking in order to improve the enrichment of active compounds over decoys. Here we have investigated this in the context of the adenosine A2A receptor, where water molecules have previously been shown to be important for achieving high enrichment rates with docking, and where the positions of some binding site waters are known from a high-resolution crystal structure. The effect of these waters (both their presence and orientations) on VS enrichment was assessed using a carefully curated set of 299 high affinity A2A antagonists and 17,337 decoys. We show that including certain crystal waters greatly improves VS enrichment and that optimization of water hydrogen positions is needed in order to achieve the best results. We also show that waters derived from a molecular dynamics simulation - without any knowledge of crystallographic waters - can improve enrichments to a similar degree as the crystallographic waters, which makes this strategy applicable to structures without experimental knowledge of water positions. Finally, we used decision trees to select an ensemble of structures with different water molecule positions and orientations that outperforms any single structure with water molecules. The approach presented here is validated against independent test sets of A2A receptor antagonists and decoys from the literature. In general, this water optimization strategy could be applied to any target with waters-mediated protein-ligand interactions.

Predicting Binding Affinities for GPCR Ligands Using Free-Energy Perturbation

The rapid growth of structural information for G-protein-coupled receptors (GPCRs) has led to a greater understanding of their structure, function, selectivity, and ligand binding. Although novel ligands have been identified using methods such as virtual screening, computationally driven lead optimization has been possible only in isolated cases because of challenges associated with predicting binding free energies for related compounds. Here, we provide a systematic characterization of the performance of free-energy perturbation (FEP) calculations to predict relative binding free energies of congeneric ligands binding to GPCR targets using a consistent protocol and no adjustable parameters. Using the FEP+ package, first we validated the protocol, which includes a full lipid bilayer and explicit solvent, by predicting the binding affinity for a total of 45 different ligands across four different GPCRs (adenosine A2AAR, β1 adrenergic, CXCR4 chemokine, and δ opioid receptors). Comparison with experimental binding affinity measurements revealed a highly predictive ranking correlation (average spearman ρ = 0.55) and low root-mean-square error (0.80 kcal/mol). Next, we applied FEP+ in a prospective project, where we predicted the affinity of novel, potent adenosine A2A receptor (A2AR) antagonists. Four novel compounds were synthesized and tested in a radioligand displacement assay, yielding affinity values in the nanomolar range. The affinity of two out of the four novel ligands (plus three previously reported compounds) was correctly predicted (within 1 kcal/mol), including one compound with approximately a tenfold increase in affinity compared to the starting compound. Detailed analyses of the simulations underlying the predictions provided insights into the structural basis for the two cases where the affinity was overpredicted. Taken together, these results establish a protocol for systematically applying FEP+ to GPCRs and provide guidelines for identifying potent molecules in drug discovery lead optimization projects.

Sodium Ion Binding Pocket Mutations and Adenosine A2A Receptor Function

Recently we identified a sodium ion binding pocket in a high-resolution structure of the human adenosine A2A receptor. In the present study we explored this binding site through site-directed mutagenesis and molecular dynamics simulations. Amino acids in the pocket were mutated to alanine, and their influence on agonist and antagonist affinity, allosterism by sodium ions and amilorides, and receptor functionality was explored. Mutation of the polar residues in the Na+ pocket were shown to either abrogate (D52A2.50 and N284A7.49) or reduce (S91A3.39, W246A6.48, and N280A7.45) the negative allosteric effect of sodium ions on agonist binding. Mutations D52A2.50 and N284A7.49 completely abolished receptor signaling, whereas mutations S91A3.39 and N280A7.45 elevated basal activity and mutations S91A3.39, W246A6.48, and N280A7.45 decreased agonist-stimulated receptor signaling. In molecular dynamics simulations D52A2.50 directly affected the mobility of sodium ions, which readily migrated to another pocket formed by Glu131.39 and His2787.43. The D52A2.50 mutation also decreased the potency of amiloride with respect to ligand displacement but did not change orthosteric ligand affinity. In contrast, W246A6.48 increased some of the allosteric effects of sodium ions and amiloride, whereas orthosteric ligand binding was decreased. These new findings suggest that the sodium ion in the allosteric binding pocket not only impacts ligand affinity but also plays a vital role in receptor signaling. Because the sodium ion binding pocket is highly conserved in other class A G protein–coupled receptors, our findings may have a general relevance for these receptors and may guide the design of novel synthetic allosteric modulators or bitopic ligands.

Proteochemometric modeling in a Bayesian framework.

Proteochemometrics (PCM) is an approach for bioactivity predictive modeling which models the relationship between protein and chemical information. Gaussian Processes (GP), based on Bayesian inference, provide the most objective estimation of the uncertainty of the predictions, thus permitting the evaluation of the applicability domain (AD) of the model. Furthermore, the experimental error on bioactivity measurements can be used as input for this probabilistic model.In this study, we apply GP implemented with a panel of kernels on three various (and multispecies) PCM datasets. The first dataset consisted of information from 8 human and rat adenosine receptors with 10,999 small molecule ligands and their binding affinity. The second consisted of the catalytic activity of four dengue virus NS3 proteases on 56 small peptides. Finally, we have gathered bioactivity information of small molecule ligands on 91 aminergic GPCRs from 9 different species, leading to a dataset of 24,593 datapoints with a matrix completeness of only 2.43%.GP models trained on these datasets are statistically sound, at the same level of statistical significance as Support Vector Machines (SVM), with R02 values on the external dataset ranging from 0.68 to 0.92, and RMSEP values close to the experimental error. Furthermore, the best GP models obtained with the normalized polynomial and radial kernels provide intervals of confidence for the predictions in agreement with the cumulative Gaussian distribution. GP models were also interpreted on the basis of individual targets and of ligand descriptors. In the dengue dataset, the model interpretation in terms of the amino-acid positions in the tetra-peptide ligands gave biologically meaningful results.

Removal of Human Ether-à-go-go Related Gene (hERG) K+ Channel Affinity through Rigidity: A Case of Clofilium Analogues

Cardiotoxicity is a side effect that plagues modern drug design and is very often due to the off-target blockade of the human ether-à-go-go related gene (hERG) potassium channel. To better understand the structural determinants of this blockade, we designed and synthesized a series of 40 derivatives of clofilium, a class III antiarrhythmic agent. These were evaluated in radioligand binding and patch-clamp assays to establish structure-affinity relationships (SAR) for this potassium channel. Efforts were especially focused on studying the influence of the structural rigidity and the nature of the linkers composing the clofilium scaffold. It was shown that introducing triple bonds and oxygen atoms in the n-butyl linker of the molecule greatly reduced affinity without significantly modifying the pKa of the essential basic nitrogen. These findings could prove useful in the first stages of drug discovery as a systematic way of reducing the risk of hERG K(+) channel blockade-induced cardiotoxicity.

A yeast screening method to decipher the interaction between the adenosine A2B receptor and the C-terminus of different G protein α-subunits

The expression of human G protein-coupled receptors (GPCRs) in Saccharomyces cerevisiae containing chimeric yeast/mammalian Gα subunits provides a useful tool for the study of GPCR activation. In this study, we used a one-GPCR-one-G protein yeast screening method in combination with molecular modeling and mutagenesis studies to decipher the interaction between GPCRs and the C-terminus of different α-subunits of G proteins. We chose the human adenosine A2B receptor (hA2BR) as a paradigm, a typical class A GPCR that shows promiscuous behavior in G protein coupling in this yeast system. The wild-type hA2BR and five mutant receptors were expressed in 8 yeast strains with different humanized G proteins, covering the four major classes: Gαi, Gαs, Gαq, and Gα12. Our experiments showed that a tyrosine residue (Y) at the C-terminus of the Gα subunit plays an important role in controlling the activation of GPCRs. Receptor residues R1033.50 and I1073.54 are vital too in G protein-coupling and the activation of the hA2BR, whereas L213IL3 is more important in G protein inactivation. Substitution of S2356.36 to alanine provided the most divergent G protein-coupling profile. Finally, L2366.37 substitution decreased receptor activation in all G protein pathways, although to a different extent. In conclusion, our findings shed light on the selectivity of receptor/G protein coupling, which may help in further understanding GPCR signaling.

Beyond the Hype : Deep Neural Networks Outperform Established Methods Using A ChEMBL Bioactivity Benchmark Set

The increase of publicly available bioactivity data in recent years has fueled and catalyzed research in chemogenomics, data mining, and modeling approaches. As a direct result, over the past few years a multitude of different methods have been reported and evaluated, such as target fishing, nearest neighbor similarity-based methods, and Quantitative Structure Activity Relationship (QSAR)-based protocols. However, such studies are typically conducted on different datasets, using different validation strategies, and different metrics. In this study, different methods were compared using one single standardized dataset obtained from ChEMBL, which is made available to the public, using standardized metrics (BEDROC and Matthews Correlation Coefficient). Specifically, the performance of Naïve Bayes, Random Forests, Support Vector Machines, Logistic Regression, and Deep Neural Networks was assessed using QSAR and proteochemometric (PCM) methods. All methods were validated using both a random split validation and a temporal validation, with the latter being a more realistic benchmark of expected prospective execution. Deep Neural Networks are the top performing classifiers, highlighting the added value of Deep Neural Networks over other more conventional methods. Moreover, the best method (‘DNN_PCM’) performed significantly better at almost one standard deviation higher than the mean performance. Furthermore, Multi-task and PCM implementations were shown to improve performance over single task Deep Neural Networks. Conversely, target prediction performed almost two standard deviations under the mean performance. Random Forests, Support Vector Machines, and Logistic Regression performed around mean performance. Finally, using an ensemble of DNNs, alongside additional tuning, enhanced the relative performance by another 27% (compared with unoptimized ‘DNN_PCM’). Here, a standardized set to test and evaluate different machine learning algorithms in the context of multi-task learning is offered by providing the data and the protocols.Graphical Abstract.

Discovery and Mapping of an Intracellular Antagonist Binding Site at the Chemokine Receptor CCR2

The chemokine receptor CCR2 is a G protein-coupled receptor that is involved in many diseases characterized by chronic inflammation, and therefore a large variety of CCR2 small molecule antagonists has been developed. On the basis of their chemical structures these antagonists can roughly be divided into two groups with most likely two topographically distinct binding sites. The aim of the current study was to identify the binding site of one such group of ligands, exemplified by three allosteric antagonists, CCR2-RA-[R], JNJ-27141491, and SD-24. We first used a chimeric CCR2/CCR5 receptor approach to obtain insight into the binding site of the allosteric antagonists and additionally introduced eight single point mutations in CCR2 to further characterize the putative binding pocket. All constructs were studied in radioligand binding and/or functional IP turnover assays, providing evidence for an intracellular binding site for CCR2-RA-[R], JNJ-27141491, and SD-24. For CCR2-RA-[R] the most important residues for binding were found to be the highly conserved tyrosine Y7.53 and phenylalanine F8.50 of the NPxxYx(5,6)F motif, as well as V6.36 at the bottom of TM-VI and K8.49 in helix-VIII. These findings demonstrate for the first time the presence of an allosteric intracellular binding site for CCR2 antagonists. This contributes to an increased understanding of the interactions of diverse ligands at CCR2 and may allow for a more rational design of future allosteric antagonists.

Domains for activation and inactivation in G protein-coupled receptors--a mutational analysis of constitutive activity of the adenosine A2B receptor.

G protein-coupled receptors (GPCRs) are a major drug target and can be activated by a range of stimuli, from photons to proteins. Most, if not all, GPCRs also display a basal level of biological response in the absence of such a stimulus. This level of so-called constitutive activity results from a delicate energy equilibrium that exists between the active and the inactive state of the receptor and is the first determinant in the GPCR activation mechanism. Here we describe new insights in specific regions of the adenosine A2B receptor that are essential in activation and inactivation. We developed a new screening method using the MMY24 S. Cerevisiae strain by which we were able to screen for constitutively inactive mutants receptors (CIMs). We applied this screening method on a mutagenic library of the adenosine A2B receptor, where random mutations were introduced in transmembrane domains four and five (TM4 and TM5) linked by extracellular loop 2 (EL2). The screen resulted in the identification of 22 single and double mutant receptors, all showing a decrease in constitutive activity as well as in agonist potency. By comparing these results with a previous screen of the same mutagenic library for constitutively active mutant receptors (CAMs), we discovered specific regions in this G protein-coupled receptor involved in either inactivation or activation or both. The results suggest the activation mechanism of GPCRs to be much less restricted to sites of high conservation or direct interaction with the ligand or G protein and illustrate how dynamic the activation process of GPCRs is.