Exploring hierarchical refinement techniques for induced fit docking with protein and ligand flexibility

Tuning RXR Modulators for PGC1α Recruitment

The molecular activation mechanism of the nuclear retinoid X receptors (RXRs) crucially involves ligand-induced corepressor release and coactivator recruitment which mediate transcriptional repression or activation. The ability of RXR to bind diverse coactivators suggests that a coregulator-selective modulation by ligands may open an avenue to tissue- or gene-selective RXR activation. Here, we identified strong induction of peroxisome proliferator-activated receptor γ coactivator 1α (PGC1α) binding to RXR by a synthetic agonist but not by the endogenous ligand 9-cis retinoic acid. Structure-guided diversification of this lead resulted in a set of three structurally related RXR agonists with different ability to promote PGC1α recruitment in cell-free and cellular context. These results demonstrate that selective modulation of coregulator recruitment to RXR can be achieved with molecular glues and potentially open new therapeutic opportunities by targeting the ligand-induced RXR-PGC1α interaction.

Selective Bias Virtual Screening for Discovery of Promising Antimalarial Candidates targeting Plasmodium N-Myristoyltransferase

Malaria remains a significant public health challenge, with Plasmodium vivax being the species responsible for the most prevalent form of the disease. Given the limited therapeutic options available, the search for new antimalarials against P. vivax is urgent. This study aims to identify new inhibitors for P. vivax N-myristoyltransferase (PvNMT), an essential drug target against malaria. Through a validated virtual screening campaign, we prioritized 23 candidates for further testing. In the yeast NMT system, seven compounds exhibit a potential inhibitor phenotype. In vitro antimalarial phenotypic assays confirmed the activity of four candidates while demonstrating an absence of cytotoxicity. Enzymatic assays reveal LabMol-394 as the most promising inhibitor, displaying selectivity against the parasite and a strong correlation within the yeast system. Furthermore, molecular dynamics simulations shed some light into its binding mode. This study constitutes a substantial contribution to the exploration of a selective quinoline scaffold and provides valuable insights into the development of new antimalarial candidates.

Inhibition of ATP synthase reverse activity restores energy homeostasis in mitochondrial pathologies

The maintenance of cellular function relies on the close regulation of adenosine triphosphate (ATP) synthesis and hydrolysis. ATP hydrolysis by mitochondrial ATP Synthase (CV) is induced by loss of proton motive force and inhibited by the mitochondrial protein ATPase inhibitor (ATPIF1). The extent of CV hydrolytic activity and its impact on cellular energetics remains unknown due to the lack of selective hydrolysis inhibitors of CV. We find that CV hydrolytic activity takes place in coupled intact mitochondria and is increased by respiratory chain defects. We identified (+)-Epicatechin as a selective inhibitor of ATP hydrolysis that binds CV while preventing the binding of ATPIF1. In cells with Complex-III deficiency, we show that inhibition of CV hydrolytic activity by (+)-Epichatechin is sufficient to restore ATP content without restoring respiratory function. Inhibition of CV-ATP hydrolysis in a mouse model of Duchenne Muscular Dystrophy is sufficient to improve muscle force without any increase in mitochondrial content. We conclude that the impact of compromised mitochondrial respiration can be lessened using hydrolysis-selective inhibitors of CV.

Are Deep Learning Structural Models Sufficiently Accurate for Virtual Screening? Application of Docking Algorithms to AlphaFold2 Predicted Structures

Machine learning-based protein structure prediction algorithms, such as RosettaFold and AlphaFold2, have greatly impacted the structural biology field, arousing a fair amount of discussion around their potential role in drug discovery. While there are few preliminary studies addressing the usage of these models in virtual screening, none of them focus on the prospect of hit-finding in a real-world virtual screen with a model based on low prior structural information. In order to address this, we have developed an AlphaFold2 version where we exclude all structural templates with more than 30% sequence identity from the model-building process. In a previous study, we used those models in conjunction with state-of-the-art free energy perturbation methods and demonstrated that it is possible to obtain quantitatively accurate results. In this work, we focus on using these structures in rigid receptor-ligand docking studies. Our results indicate that using out-of-the-box Alphafold2 models is not an ideal scenario for virtual screening campaigns; in fact, we strongly recommend to include some post-processing modeling to drive the binding site into a more realistic holo model.

Are Deep Learning Structural Models Sufficiently Accurate for Free-Energy Calculations? Application of FEP+ to AlphaFold2-Predicted Structures

The availability of AlphaFold2 has led to great excitement in the scientific community─particularly among drug hunters─due to the ability of the algorithm to predict protein structures with high accuracy. However, beyond globally accurate protein structure prediction, it remains to be determined whether ligand binding sites are predicted with sufficient accuracy in these structures to be useful in supporting computationally driven drug discovery programs. We explored this question by performing free-energy perturbation (FEP) calculations on a set of well-studied protein-ligand complexes, where AlphaFold2 predictions were performed by removing all templates with >30% identity to the target protein from the training set. We observed that in most cases, the ΔΔG values for ligand transformations calculated with FEP, using these prospective AlphaFold2 structures, were comparable in accuracy to the corresponding calculations previously carried out using crystal structures. We conclude that under the right circumstances, AlphaFold2-modeled structures are accurate enough to be used by physics-based methods such as FEP in typical lead optimization stages of a drug discovery program.

Chemical genetics strategy to profile kinase target engagement reveals role of FES in neutrophil phagocytosis

Chemical tools to monitor drug-target engagement of endogenously expressed protein kinases are highly desirable for preclinical target validation in drug discovery. Here, we describe a chemical genetics strategy to selectively study target engagement of endogenous kinases. By substituting a serine residue into cysteine at the DFG-1 position in the ATP-binding pocket, we sensitize the non-receptor tyrosine kinase FES towards covalent labeling by a complementary fluorescent chemical probe. This mutation is introduced in the endogenous FES gene of HL-60 cells using CRISPR/Cas9 gene editing. Leveraging the temporal and acute control offered by our strategy, we show that FES activity is dispensable for differentiation of HL-60 cells towards macrophages. Instead, FES plays a key role in neutrophil phagocytosis via SYK kinase activation. This chemical genetics strategy holds promise as a target validation method for kinases.

Cryo-Electron Microscopy Structure of the αIIbβ3-Abciximab Complex

OBJECTIVE

The αIIbβ3 antagonist antiplatelet drug abciximab is the chimeric antigen-binding fragment comprising the variable regions of murine monoclonal antibody 7E3 and the constant domains of human IgG1 and light chain κ. Previous mutagenesis studies suggested that abciximab binds to the β3 C177-C184 specificity-determining loop (SDL) and Trp129 on the adjacent β1-α1 helix. These studies could not, however, assess whether 7E3 or abciximab prevents fibrinogen binding by steric interference, disruption of either the αIIbβ3-binding pocket for fibrinogen or the β3 SDL (which is not part of the binding pocket but affects fibrinogen binding), or some combination of these effects. To address this gap, we used cryo-electron microscopy to determine the structure of the αIIbβ3-abciximab complex at 2.8 Å resolution. Approach and Results: The interacting surface of abciximab is comprised of residues from all 3 complementarity-determining regions of both the light and heavy chains, with high representation of aromatic residues. Binding is primarily to the β3 SDL and neighboring residues, the β1-α1 helix, and β3 residues Ser211, Val212 and Met335. Unexpectedly, the structure also indicated several interactions with αIIb. As judged by the cryo-electron microscopy model, molecular-dynamics simulations, and mutagenesis, the binding of abciximab does not appear to rely on the interaction with the αIIb residues and does not result in disruption of the fibrinogen-binding pocket; it does, however, compress and reduce the flexibility of the SDL.

CONCLUSIONS

We deduce that abciximab prevents ligand binding by steric interference, with a potential contribution via displacement of the SDL and limitation of the flexibility of the SDL residues.

Atomistic simulations shed new light on the activation mechanisms of RORγ and classify it as Type III nuclear hormone receptor regarding ligand-binding paths

The molecular recognition of the RORγ nuclear hormone receptor (NHR) ligand-binding domain (LBD) has been extensively studied with numerous X-ray crystal structures. However, the picture afforded by these complexes is static and does not fully explain the functional behavior of the LBD. In particular, the apo structure of the LBD seems to be in a fully active state, with no obvious differences to the agonist-bound structure. Further, several atypical in vivo inverse agonists have surprisingly been found to co-crystallize with the LBD in agonist mode (with co-activator), leading to a disconnection between molecular recognition and functional activity. Moreover, the experimental structures give no clues on how RORγ LBD binders access the interior of the LBD. To address all these points, we probe here, with a variety of simulation techniques, the fine structural balance of the RORγ LBD in its apo vs. holo form, the differences in flexibility and stability of the LBD in complex with agonists vs. inverse agonists and how binders diffuse in and out of the LBD in unbiased simulations. Our data conclusively point to the stability afforded by the so-called “agonist lock” between H479 and Y502 and the precise location of Helix 12 (H12) for the competence of the LBD to bind co-activator proteins. We observe the “water trapping” mechanism suggested previously for the atypical inverse agonists and discover a different behavior for the latter when co-activator is present or absent, which might help explain their conflicting data. Additionally, we unveil the same entry/exit path for agonists and inverse agonist into and out of the LBD for RORγ, suggesting it belongs to the type III NHR sub-family.

Determinants and Prediction of Esterase Substrate Promiscuity Patterns

Esterases receive special attention because of their wide distribution in biological systems and environments and their importance for physiology and chemical synthesis. The prediction of esterases' substrate promiscuity level from sequence data and the molecular reasons why certain such enzymes are more promiscuous than others remain to be elucidated. This limits the surveillance of the sequence space for esterases potentially leading to new versatile biocatalysts and new insights into their role in cellular function. Here, we performed an extensive analysis of the substrate spectra of 145 phylogenetically and environmentally diverse microbial esterases, when tested with 96 diverse esters. We determined the primary factors shaping their substrate range by analyzing substrate range patterns in combination with structural analysis and protein-ligand simulations. We found a structural parameter that helps rank (classify) the promiscuity level of esterases from sequence data at 94% accuracy. This parameter, the active site effective volume, exemplifies the topology of the catalytic environment by measuring the active site cavity volume corrected by the relative solvent accessible surface area (SASA) of the catalytic triad. Sequences encoding esterases with active site effective volumes (cavity volume/SASA) above a threshold show greater substrate spectra, which can be further extended in combination with phylogenetic data. This measure provides also a valuable tool for interrogating substrates capable of being converted. This measure, found to be transferred to phosphatases of the haloalkanoic acid dehalogenase superfamily and possibly other enzymatic systems, represents a powerful tool for low-cost bioprospecting for esterases with broad substrate ranges, in large scale sequence data sets.

NMR Signal Quenching from Bound Biradical Affinity Reagents in DNP Samples

We characterize the effect of specifically bound biradicals on the NMR spectra of dihydrofolate reductase from E. coli. Dynamic nuclear polarization methods enhance the signal-to-noise of solid state NMR experiments by transferring polarization from unpaired electrons of biradicals to nuclei. There has been recent interest in colocalizing the paramagnetic polarizing agents with the analyte of interest through covalent or noncovalent specific interactions. This experimental approach broadens the scope of dynamic nuclear polarization methods by offering the possibility of selective signal enhancements and the potential to work in a broad range of environments. Paramagnetic compounds can have other effects on the NMR spectroscopy of nearby nuclei, including broadening of nuclear resonances due to the proximity of the paramagnetic agent. Understanding the distance dependence of these interactions is important for the success of the technique. Here we explore paramagnetic signal quenching due to a bound biradical, specifically a biradical-derivatized trimethoprim ligand of E. coli dihydrofolate reductase. Biradical-derivatized trimethoprim has nanomolar affinity for its target, and affords strong and selective signal enhancements in dynamic nuclear polarization experiments. In this work, we show that, although the trimethoprim fragment is well ordered, the biradical (TOTAPOL) moiety is disordered when bound to the protein. The distance dependence in bleaching of NMR signal intensity allows us to detect numerous NMR signals in the protein. We present the possibility that static disorder and electron spin diffusion play roles in this observation, among other contributions. The fact that the majority of signals are observed strengthens the case for the use of high affinity or covalent radicals in dynamic nuclear polarization solid state NMR enhancement.

Inhibition of Human Enhancer of Zeste Homolog 2 with Tambjamine Analogs

Combining computational modeling, de novo compound synthesis, and in vitro and cellular assays, we have performed an inhibition study against the enhancer of zeste homolog 2 (EZH2) histone-lysine N-methyltransferase. This enzyme is an important catalytic component of the PRC2 complex whose alterations have been associated with different cancers. We introduce here several tambjamine-inspired derivatives with low micromolar in vitro activity that produce a significant decrease in histone 3 trimethylation levels in cancer cells. We demonstrate binding at the methyl transfer active site, showing, in addition, that the EZH2 isolated crystal structure is capable of being used in molecular screening studies. Altogether, this work provides a successful molecular model that will help in the identification of new specific EZH2 inhibitors and identify a novel class of tambjamine-derived EZH2 inhibitors with promising activities for their use in cancer treatment.

Adaptive simulations, towards interactive protein-ligand modeling

Modeling the dynamic nature of protein-ligand binding with atomistic simulations is one of the main challenges in computational biophysics, with important implications in the drug design process. Although in the past few years hardware and software advances have significantly revamped the use of molecular simulations, we still lack a fast and accurate ab initio description of the binding mechanism in complex systems, available only for up-to-date techniques and requiring several hours or days of heavy computation. Such delay is one of the main limiting factors for a larger penetration of protein dynamics modeling in the pharmaceutical industry. Here we present a game-changing technology, opening up the way for fast reliable simulations of protein dynamics by combining an adaptive reinforcement learning procedure with Monte Carlo sampling in the frame of modern multi-core computational resources. We show remarkable performance in mapping the protein-ligand energy landscape, being able to reproduce the full binding mechanism in less than half an hour, or the active site induced fit in less than 5 minutes. We exemplify our method by studying diverse complex targets, including nuclear hormone receptors and GPCRs, demonstrating the potential of using the new adaptive technique in screening and lead optimization studies.

Ligand binding modes from low resolution GPCR models and mutagenesis: chicken bitter taste receptor as a test-case

Bitter taste is one of the basic taste modalities, warning against consuming potential poisons. Bitter compounds activate members of the bitter taste receptor (Tas2r) subfamily of G protein-coupled receptors (GPCRs). The number of functional Tas2rs is species-dependent. Chickens represent an intriguing minimalistic model, because they detect the bitter taste of structurally different molecules with merely three bitter taste receptor subtypes. We investigated the binding modes of several known agonists of a representative chicken bitter taste receptor, ggTas2r1. Because of low sequence similarity between ggTas2r1 and crystallized GPCRs (~10% identity, ~30% similarity at most), the combination of computational approaches with site-directed mutagenesis was used to characterize the agonist-bound conformation of ggTas2r1 binding site between TMs 3, 5, 6 and 7. We found that the ligand interactions with N93 in TM3 and/or N247 in TM5, combined with hydrophobic contacts, are typically involved in agonist recognition. Next, the ggTas2r1 structural model was successfully used to identify three quinine analogues (epiquinidine, ethylhydrocupreine, quinidine) as new ggTas2r1 agonists. The integrated approach validated here may be applicable to additional cases where the sequence identity of the GPCR of interest and the existing experimental structures is low.

Evaluation of the novel algorithm of flexible ligand docking with moveable target-protein atoms

We present the novel docking algorithm based on the Tensor Train decomposition and the TT-Cross global optimization. The algorithm is applied to the docking problem with flexible ligand and moveable protein atoms. The energy of the protein-ligand complex is calculated in the frame of the MMFF94 force field in vacuum. The grid of precalculated energy potentials of probe ligand atoms in the field of the target protein atoms is not used. The energy of the protein-ligand complex for any given configuration is computed directly with the MMFF94 force field without any fitting parameters. The conformation space of the system coordinates is formed by translations and rotations of the ligand as a whole, by the ligand torsions and also by Cartesian coordinates of the selected target protein atoms. Mobility of protein and ligand atoms is taken into account in the docking process simultaneously and equally. The algorithm is realized in the novel parallel docking SOL-P program and results of its performance for a set of 30 protein-ligand complexes are presented. Dependence of the docking positioning accuracy is investigated as a function of parameters of the docking algorithm and the number of protein moveable atoms. It is shown that mobility of the protein atoms improves docking positioning accuracy. The SOL-P program is able to perform docking of a flexible ligand into the active site of the target protein with several dozens of protein moveable atoms: the native crystallized ligand pose is correctly found as the global energy minimum in the search space with 157 dimensions using 4700 CPU ∗ h at the Lomonosov supercomputer.

Relative Binding Free Energy Calculations Applied to Protein Homology Models

A significant challenge and potential high-value application of computer-aided drug design is the accurate prediction of protein-ligand binding affinities. Free energy perturbation (FEP) using molecular dynamics (MD) sampling is among the most suitable approaches to achieve accurate binding free energy predictions, due to the rigorous statistical framework of the methodology, correct representation of the energetics, and thorough treatment of the important degrees of freedom in the system (including explicit waters). Recent advances in sampling methods and force fields coupled with vast increases in computational resources have made FEP a viable technology to drive hit-to-lead and lead optimization, allowing for more efficient cycles of medicinal chemistry and the possibility to explore much larger chemical spaces. However, previous FEP applications have focused on systems with high-resolution crystal structures of the target as starting points-something that is not always available in drug discovery projects. As such, the ability to apply FEP on homology models would greatly expand the domain of applicability of FEP in drug discovery. In this work we apply a particular implementation of FEP, called FEP+, on congeneric ligand series binding to four diverse targets: a kinase (Tyk2), an epigenetic bromodomain (BRD4), a transmembrane GPCR (A2A), and a protein-protein interaction interface (BCL-2 family protein MCL-1). We apply FEP+ using both crystal structures and homology models as starting points and find that the performance using homology models is generally on a par with the results when using crystal structures. The robustness of the calculations to structural variations in the input models can likely be attributed to the conformational sampling in the molecular dynamics simulations, which allows the modeled receptor to adapt to the "real" conformation for each ligand in the series. This work exemplifies the advantages of using all-atom simulation methods with full system flexibility and offers promise for the general application of FEP to homology models, although additional validation studies should be performed to further understand the limitations of the method and the scenarios where FEP will work best.

The unravelling of the complex pattern of tyrosinase inhibition

Tyrosinases are responsible for melanin formation in all life domains. Tyrosinase inhibitors are used for the prevention of severe skin diseases, in skin-whitening creams and to avoid fruit browning, however continued use of many such inhibitors is considered unsafe. In this study we provide conclusive evidence of the inhibition mechanism of two well studied tyrosinase inhibitors, KA (kojic acid) and HQ (hydroquinone), which are extensively used in hyperpigmentation treatment. KA is reported in the literature with contradicting inhibition mechanisms, while HQ is described as both a tyrosinase inhibitor and a substrate. By visualization of KA and HQ in the active site of TyrBm crystals, together with molecular modeling, binding constant analysis and kinetic experiments, we have elucidated their mechanisms of inhibition, which was ambiguous for both inhibitors. We confirm that while KA acts as a mixed inhibitor, HQ can act both as a TyrBm substrate and as an inhibitor.

Unveiling prolyl oligopeptidase ligand migration by comprehensive computational techniques

Prolyl oligopeptidase (POP) is a large 80 kDa protease, which cleaves oligopeptides at the C-terminal side of proline residues and constitutes an important pharmaceutical target. Despite the existence of several crystallographic structures, there is an open debate about migration (entrance and exit) pathways for ligands, and their coupling with protein dynamics. Recent studies have shown the capabilities of molecular dynamics and classical force fields in describing spontaneous binding events and nonbiased ligand migration pathways. Due to POP's size and to the buried nature of its active site, an exhaustive sampling by means of conventional long enough molecular dynamics trajectories is still a nearly impossible task. Such a level of sampling, however, is possible with the breakthrough protein energy landscape exploration technique. Here, we present an exhaustive sampling of POP with a known inhibitor, Z-pro-prolinal. In >3000 trajectories Z-pro-prolinal explores all the accessible surface area, showing multiple entrance events into the large internal cavity through the pore in the β-propeller domain. Moreover, we modeled a natural substrate binding and product release by predicting the entrance of an undecapeptide substrate, followed by manual active site cleavage and nonbiased exit of one of the products (a dipeptide). The product exit shows preference from a flexible 18-amino acid residues loop, pointing to an overall mechanism where entrance and exit occur in different sites.

PELE web server: atomistic study of biomolecular systems at your fingertips

PELE, Protein Energy Landscape Exploration, our novel technology based on protein structure prediction algorithms and a Monte Carlo sampling, is capable of modelling the all-atom protein–ligand dynamical interactions in an efficient and fast manner, with two orders of magnitude reduced computational cost when compared with traditional molecular dynamics techniques. PELE’s heuristic approach generates trial moves based on protein and ligand perturbations followed by side chain sampling and global/local minimization. The collection of accepted steps forms a stochastic trajectory. Furthermore, several processors may be run in parallel towards a collective goal or defining several independent trajectories; the whole procedure has been parallelized using the Message Passing Interface. Here, we introduce the PELE web server, designed to make the whole process of running simulations easier and more practical by minimizing input file demand, providing user-friendly interface and producing abstract outputs (e.g. interactive graphs and tables). The web server has been implemented in C++ using Wt (http://www.webtoolkit.eu) and MySQL (http://www.mysql.com). The PELE web server, accessible at http://pele.bsc.es, is free and open to all users with no login requirement.

Molecular interactions of prodiginines with the BH3 domain of anti-apoptotic Bcl-2 family members

Prodigiosin and obatoclax, members of the prodiginines family, are small molecules with anti-cancer properties that are currently under preclinical and clinical trials. The molecular target(s) of these agents, however, is an open question. Combining experimental and computational techniques we find that prodigiosin binds to the BH3 domain in some BCL-2 protein families, which play an important role in the apoptotic programmed cell death. In particular, our results indicate a large affinity of prodigiosin for MCL-1, an anti-apoptotic member of the BCL-2 family. In melanoma cells, we demonstrate that prodigiosin activates the mitochondrial apoptotic pathway by disrupting MCL-1/BAK complexes. Computer simulations with the PELE software allow the description of the induced fit process, obtaining a detailed atomic view of the molecular interactions. These results provide new data to understand the mechanism of action of these molecules, and assist in the development of more specific inhibitors of anti-apoptotic BCL-2 proteins.

Modulating O2 reactivity in a fungal flavoenzyme: involvement of aryl-alcohol oxidase Phe-501 contiguous to catalytic histidine

Aryl-alcohol oxidase (AAO) is a flavoenzyme responsible for activation of O(2) to H(2)O(2) in fungal degradation of lignin. The AAO crystal structure shows a buried active site connected to the solvent by a hydrophobic funnel-shaped channel, with Phe-501 and two other aromatic residues forming a narrow bottleneck that prevents the direct access of alcohol substrates. However, ligand diffusion simulations show O(2) access to the active site following this channel. Site-directed mutagenesis of Phe-501 yielded a F501A variant with strongly reduced O(2) reactivity. However, a variant with increased reactivity, as shown by kinetic constants and steady-state oxidation degree, was obtained by substitution of Phe-501 with tryptophan. The high oxygen catalytic efficiency of F501W, ∼2-fold that of native AAO and ∼120-fold that of F501A, seems related to a higher O(2) availability because the turnover number was slightly decreased with respect to the native enzyme. Free diffusion simulations of O(2) inside the active-site cavity of AAO (and several in silico Phe-501 variants) yielded >60% O(2) population at 3-4 Å from flavin C4a in F501W compared with 44% in AAO and only 14% in F501A. Paradoxically, the O(2) reactivity of AAO decreased when the access channel was enlarged and increased when it was constricted by introducing a tryptophan residue. This is because the side chain of Phe-501, contiguous to the catalytic histidine (His-502 in AAO), helps to position O(2) at an adequate distance from flavin C4a (and His-502 Nε). Phe-501 substitution with a bulkier tryptophan residue resulted in an increase in the O(2) reactivity of this flavoenzyme.