Logo: to the web site of Uppsala University

Small molecules are essential for investigating the pharmacology of membrane proteins and remain the most common approach for therapeutically targeting them. However, most experimental small molecule screening methods require ligands containing radiolabels or fluorescent labels and often involve isolating proteins from their cellular environment. Additionally, most conventional screening methods are suited for identifying compounds with moderate to higher affinities (K D < 1 mu M) and are less effective at detecting lower affinity compounds, such as weakly binding molecular fragments. To address these limitations, we demonstrated a proof-of-concept application of high-resolution magic angle spinning nuclear magnetic resonance (HRMAS NMR) spectroscopy with small molecules that bind the human A2A adenosine receptor (A2AAR), a class A G protein-coupled receptor. Our approach leverages a streamlined workflow to prepare NMR samples with only milligrams of unpurified cell membranes containing similar to 1 mu M of A2AAR. Utilizing saturation transfer difference NMR, we identified bound small molecules from spectra recorded within minutes and further derived information on ligand binding poses without the need for detailed structure determination. After establishing optimal criteria for which the HRMAS approach is most sensitive, we leveraged our HRMAS approach to identify and characterize molecular fragments not previously known to be ligands of A2AAR. In molecular docking and simulations, we observed novel binding poses for these fragments, which revealed the potential to grow them into more complex ligands and confirmed HRMAS NMR as a valuable tool for lead compound identification in the context of fragment-based drug discovery.

Fragment-based screening can catalyze drug discovery by identifying novel scaffolds, but this approach is limited by the small chemical libraries studied by biophysical experiments and the challenging hit optimization step. In efforts to identify DNA repair inhibitors, we explored the use of structure-based virtual screening to access ultralarge fragment libraries that cover four orders of magnitude larger fractions of chemical space than traditional techniques. A set of 14 million fragments were docked to 8-oxoguanine DNA glycosylase (OGG1), a challenging drug target involved in cancer and inflammation. Of the 29 top-ranked fragments that were experimentally evaluated, four compounds were shown to bind to OGG1 and X-ray crystallography confirmed the predicted binding modes. Docking of readily synthesizable elaborations guided fragment optimization, leading to the discovery of submicromolar OGG1 inhibitors with anti-inflammatory and anti-cancer effects in cell models. Our results demonstrate that fragment-based virtual screening enables efficient exploration of vast chemical libraries.

A set of 2-aryl-9-H or methyl-6-morpholinopurine derivatives were synthesized and assayed through radioligand binding tests at human A1, A2A, A2B, and A3 adenosine receptor subtypes. Eleven purines showed potent antagonism at A1, A3, dual A1/A2A, A1/A2B, or A1/A3 adenosine receptors. Additionally, three compounds showed high affinity without selectivity for any specific adenosine receptor. The structure-activity relationships were made for this group of new compounds. The 9-methylpurine derivatives were generally less potent but more selective, and the 9H-purine derivatives were more potent but less selective. These compounds can be an important source of new biochemical tools and/or pharmacological drugs.

Artificial intelligence is revolutionizing protein structure prediction, providing unprecedented opportunities for drug design. To assess the potential impact on ligand discovery, we compared virtual screens using protein structures generated by the AlphaFold machine learning method and traditional homology modeling. More than 16 million compounds were docked to models of the trace amine-associated receptor 1 (TAAR1), a G protein-coupled receptor of unknown structure and target for treating neuropsychiatric disorders. Sets of 30 and 32 highly ranked compounds from the AlphaFold and homology model screens, respectively, were experimentally evaluated. Of these, 25 were TAAR1 agonists with potencies ranging from 12 to 0.03 mu M. The AlphaFold screen yielded a more than twofold higher hit rate (60%) than the homology model and discovered the most potent agonists. A TAAR1 agonist with a promising selectivity profile and drug-like properties showed physiological and antipsychotic-like effects in wild-type but not in TAAR1 knockout mice. These results demonstrate that AlphaFold structures can accelerate drug discovery.

Structure-based virtual screening aims to find molecules forming favorable interactions with a biological macromolecule using computational models of complexes. The recent surge of commercially available chemical space provides the opportunity to search for ligands of therapeutic targets among billions of compounds. This review offers a compact overview of structure-based virtual screens of vast chemical spaces, highlighting successful applications in early drug discovery for therapeutically important targets such as G protein-coupled receptors and viral enzymes. Emphasis is placed on strategies to explore ultra-large chemical libraries and synergies with emerging machine learning techniques. The current opportunities and future challenges of virtual screening are discussed, indicating that this approach will play an important role in the next-generation drug discovery pipeline.

G-protein-coupled receptors (GPCRs) play important roles in physiological processes and are modulated by drugs that either activate or block signaling. Rational design of the pharmacological efficacy profiles of GPCR ligands could enable the development of more efficient drugs, but is challenging even if high-resolution receptor structures are available. We performed molecular dynamics simulations of the β₂ adrenergic receptor in active and inactive conformations to assess if binding free energy calculations can predict differences in ligand efficacy for closely related compounds. Previously identified ligands were successfully classified into groups with comparable efficacy profiles based on the calculated shift in ligand affinity upon activation. A series of ligands were then predicted and synthesized, leading to the discovery of partial agonists with nanomolar potencies and novel scaffolds. Our results demonstrate that free energy simulations enable design of ligand efficacy and the same approach can be applied to other GPCR drug targets.

The dopamine D-2 receptor, which belongs to the family of G protein-coupled receptors (GPCR), is an important and well-validated drug target in the field of medicinal chemistry due to its wide distribution, particularly in the central nervous system, and involvement in the pathomechanism of many disorders thereof. Schizophrenia is one of the most frequent diseases associated with disorders in dopaminergic neurotransmission, and in which the D-2 receptor is the main target for the drugs used. In this work, we aimed at discovering new selective D-2 receptor antagonists with potential antipsychotic activity. Twenty-three compounds were synthesized, based on the scaffold represented by the D2AAK2 compound, which was discovered by our group. This compound is an interesting example of a D-2 receptor ligand because of its non-classical binding to this target. Radioligand binding assays and SAR analysis indicated structural modifications of D2AAK2 that are possible to maintain its activity. These findings were further rationalized using molecular modeling. Three active derivatives were identified as D-2 receptor antagonists in cAMP signaling assays, and the selected most active compound 17 was subjected to X-ray studies to investigate its stable conformation in the solid state. Finally, effects of 17 assessed in animal models confirmed its antipsychotic activity in vivo.

Transport assays using purified glucose transporters (GLUTs) have proven to be difficult to implement, hampering deeper mechanistic insights. Here the authors have optimized a transport assay in liposomes that will provide insight to study other membrane transport proteins. Glucose transporters (GLUTs) are essential for organism-wide glucose homeostasis in mammals, and their dysfunction is associated with numerous diseases, such as diabetes and cancer. Despite structural advances, transport assays using purified GLUTs have proven to be difficult to implement, hampering deeper mechanistic insights. Here, we have optimized a transport assay in liposomes for the fructose-specific isoform GLUT5. By combining lipidomic analysis with native MS and thermal-shift assays, we replicate the GLUT5 transport activities seen in crude lipids using a small number of synthetic lipids. We conclude that GLUT5 is only active under a specific range of membrane fluidity, and that human GLUT1-4 prefers a similar lipid composition to GLUT5. Although GLUT3 is designated as the high-affinity glucose transporter, in vitro D-glucose kinetics demonstrates that GLUT1 and GLUT3 actually have a similar K-M,K- but GLUT3 has a higher turnover. Interestingly, GLUT4 has a high K-M for D-glucose and yet a very slow turnover, which may have evolved to ensure uptake regulation by insulin-dependent trafficking. Overall, we outline a much-needed transport assay for measuring GLUT kinetics and our analysis implies that high-levels of free fatty acid in membranes, as found in those suffering from metabolic disorders, could directly impair glucose uptake.

Development of subtype-selective leads is essential in drug discovery campaigns targeting G protein-coupled receptors (GPCRs). Herein, a structure-based virtual screening approach to rationally design subtype-selective ligands was applied to the A1 and A2A adenosine receptors (A1R and A2AR). Crystal structures of these closely related subtypes revealed a non-conserved subpocket in the binding sites that could be exploited to identify A1R selective ligands. A library of 4.6 million compounds was screened computationally against both receptors using molecular docking and 20 A1R selective ligands were predicted. Of these, seven antagonized the A1R with micromolar activities and several compounds displayed slight selectivity for this subtype. Twenty-seven analogs of two discovered scaffolds were designed, resulting in antagonists with nanomolar potency and up to 76-fold A1R-selectivity. Our results show the potential of structure-based virtual screening to guide discovery and optimization of subtype-selective ligands, which could facilitate the development of safer drugs.

Upregulation of the transcription factor Nrf2 by inhibition of the interaction with its negative regulator Keap1 constitutes an opportunity for the treatment of disease caused by oxidative stress. We report a structurally unique series of nanomolar Keap1 inhibitors obtained from a natural product-derived macrocyclic lead. Initial exploration of the structure-derived macrocyclic lead. Initial exploration of the structure-activity relationship of the lead, followed by structure-guided optimization, resulted in a 100-fold improvement in inhibitory potency. The macrocyclic core of the nanomolar inhibitors positions three pharmacophore units for productive interactions with key residues of Keap1, including R415, R483, and Y572. Ligand optimization resulted in the displacement of a coordinated water molecule from the Keap1 binding site and a significantly altered thermodynamic profile. In addition, minor reorganizations of R415 and R483 were accompanied by major differences in affinity between ligands. This study therefore indicates the importance of accounting both for the hydration and flexibility of the Keap1 binding site when designing high-affinity ligands.