Implementing a Scoring Function Based on Interaction Fingerprint for Autogrow4: Protein Kinase CK1δ as a Case Study

A comprehensive study of SARS-CoV-2 main protease (Mpro) inhibitor-resistant mutants selected in a VSV-based system

Nirmatrelvir was the first protease inhibitor specifically developed against the SARS-CoV-2 main protease (3CLpro/Mpro) and licensed for clinical use. As SARS-CoV-2 continues to spread, variants resistant to nirmatrelvir and other currently available treatments are likely to arise. This study aimed to identify and characterize mutations that confer resistance to nirmatrelvir. To safely generate Mpro resistance mutations, we passaged a previously developed, chimeric vesicular stomatitis virus (VSV-Mpro) with increasing, yet suboptimal concentrations of nirmatrelvir. Using Wuhan-1 and Omicron Mpro variants, we selected a large set of mutants. Some mutations are frequently present in GISAID, suggesting their relevance in SARS-CoV-2. The resistance phenotype of a subset of mutations was characterized against clinically available protease inhibitors (nirmatrelvir and ensitrelvir) with cell-based, biochemical and SARS-CoV-2 replicon assays. Moreover, we showed the putative molecular mechanism of resistance based on in silico molecular modelling. These findings have implications on the development of future generation Mpro inhibitors, will help to understand SARS-CoV-2 protease inhibitor resistance mechanisms and show the relevance of specific mutations, thereby informing treatment decisions.

Lipid Trolling to Optimize A3 Adenosine Receptor-Positive Allosteric Modulators (PAMs)

A3 adenosine receptor (A3AR) positive allosteric modulators (PAMs) (2,4-disubstituted-1H-imidazo[4,5-c]quinolin-4-amines) allosterically increase the Emax of A3AR agonists, but not potency, due to concurrent orthosteric antagonism. Following mutagenesis/homology modeling of the proposed lipid-exposed allosteric binding site on the cytosolic side, we functionalized the scaffold, including heteroatom substitutions and exocyclic phenylamine extensions, to increase allosteric binding. Strategically appended linear alkyl-alkynyl chains with terminal amino/guanidino groups improved allosteric effects at both human and mouse A3ARs. The chain length, functionality, and attachment position were varied to modulate A3AR PAM activity. For example, 26 (MRS8247, p-alkyne-linked 8 methylenes) and homologues increased agonist Cl-IB-MECA's Emax and potency ([35S]GTPγS binding). The putative mechanism involves a flexible, terminally cationic chain penetrating the lipid environment for stable electrostatic anchoring to cytosolic phospholipid head groups, suggesting "lipid trolling", supported by molecular dynamic simulation of the active-state model. Thus, we have improved A3AR PAM activity through rational design based on an extrahelical, lipidic binding site.

2-Substituted (N)-Methanocarba A3 Adenosine Receptor Agonists: In Silico, In Vitro, and In Vivo Characterization

2-Arylethynyl (N)-methanocarba adenosine 5'-methylamides are selective A3 adenosine receptor (AR) agonists containing a preestablished receptor-preferred pseudoribose conformation. Here, we compare analogues having bulky 2-substitution, either containing or lacking an ethynyl spacer between adenine and a cyclic group. 2-Aryl compounds 9-11, 13, 14, 19, 22, 23, 27, 29, 31, and 34, lacking a spacer, had human (h) A3AR K i values of 2-30 nM, and others displayed lower affinity. Mouse (m) A3AR affinity varied, with 2-arylethynyl having a higher affinity than 2-aryl analogues (7, 8 > 3c, 3d > 3b). However, 2-aryl-4'-truncated derivatives had greatly reduced hA3AR affinity, even containing affinity-enhancing N 6-dopamine-derived substituents. Molecular modeling, including molecular dynamics simulation, predicted stable poses in the canonical A3AR agonist binding site, but 2-aryl (ECL2 interactions) and 2-arylethynyl (TM2 interactions) substituents have different conformations and environments. In a hA3AR miniGαi recruitment assay, 31 (MRS8062) was (slightly) more potent compared to a β-arrestin2 recruitment assay, both in engineered HEK293T cells, and its maximal efficacy (E max) was much higher (165%) than reference agonist NECA's. Thus, in the 2-aryl series, A3AR affinity and selectivity were variable and generally reduced compared to the 2-arylethynyl series, with a greater dependence on the specific aryl group present. Selected compounds were studied in vivo in an ischemic model of peripheral artery disease (PAD). Rigidified 2-arylethynyl analogues 3a-3c were protective in this model of skeletal muscle ischemia-reperfusion injury/claudication, as previously shown only for moderately A3AR-selective ribosides or (N)-methanocarba derivatives. Thus, we have expanded the A3AR agonist SAR for (N)-methanocarba adenosines.

Another string to your bow: machine learning prediction of the pharmacokinetic properties of small molecules

INTRODUCTION

Prediction of pharmacokinetic (PK) properties is crucial for drug discovery and development. Machine-learning (ML) models, which use statistical pattern recognition to learn correlations between input features (such as chemical structures) and target variables (such as PK parameters), are being increasingly used for this purpose. To embed ML models for PK prediction into workflows and to guide future development, a solid understanding of their applicability, advantages, limitations, and synergies with other approaches is necessary.

AREAS COVERED

This narrative review discusses the design and application of ML models to predict PK parameters of small molecules, especially in light of established approaches including in vitro-in vivo extrapolation (IVIVE) and physiologically based pharmacokinetic (PBPK) models. The authors illustrate scenarios in which the three approaches are used and emphasize how they enhance and complement each other. In particular, they highlight achievements, the state of the art and potentials of applying machine learning for PK prediction through a comphrehensive literature review.

EXPERT OPINION

ML models, when carefully crafted, regularly updated, and appropriately used, empower users to prioritize molecules with favorable PK properties. Informed practitioners can leverage these models to improve the efficiency of drug discovery and development process.

Structural Investigations on 2-Amidobenzimidazole Derivatives as New Inhibitors of Protein Kinase CK1 Delta

Protein kinase CK1δ (CK1δ) is a serine-threonine/kinase that modulates different physiological processes, including the cell cycle, DNA repair, and apoptosis. CK1δ overexpression, and the consequent hyperphosphorylation of specific proteins, can lead to sleep disorders, cancer, and neurodegenerative diseases. CK1δ inhibitors showed anticancer properties as well as neuroprotective effects in cellular and animal models of Parkinson's and Alzheimer's diseases and amyotrophic lateral sclerosis. To obtain new ATP-competitive CK1δ inhibitors, three sets of benzimidazole-2-amino derivatives were synthesized (1-32), bearing different substituents on the fused benzo ring (R) and diverse pyrazole-containing acyl moieties on the 2-amino group. The best-performing derivatives were those featuring the (1H-pyrazol-3-yl)-acetyl moiety on the benzimidazol-2-amino scaffold (13-32), which showed CK1δ inhibitor activity in the low micromolar range. Among the R substituents, 5-cyano was the most advantageous, leading to a compound endowed with nanomolar potency (23, IC50 = 98.6 nM). Molecular docking and dynamics studies were performed to point out the inhibitor-kinase interactions.

Extrahelical Binding Site for a 1H-Imidazo[4,5-c]quinolin-4-amine A3 Adenosine Receptor Positive Allosteric Modulator on Helix 8 and Distal Portions of Transmembrane Domains 1 and 7

This study describes the localization and computational prediction of a binding site for the A3 adenosine receptor (A3AR) positive allosteric modulator 2-cyclohexyl-1H-imidazo[4,5-c]quinolin-4-(3,4-dichlorophenyl)amine (LUF6000). The work reveals an extrahelical lipid-facing binding pocket disparate from the orthosteric binding site that encompasses transmembrane domain (TMD) 1, TMD7, and Helix (H) 8, which was predicted by molecular modeling and validated by mutagenesis. According to the model, the nearly planar 1H-imidazo[4,5-c]quinolinamine ring system lies parallel to the transmembrane segments, inserted into an aromatic cage formed by π-π stacking interactions with the side chains of Y2847.55 in TMD7 and Y2938.54 in H8 and by π-NH bonding between Y2847.55 and the exocyclic amine. The 2-cyclohexyl group is positioned "upward" within a small hydrophobic subpocket created by residues in TMDs 1 and 7, while the 3,4-dichlorophenyl group extends toward the lipid interface. An H-bond between the N-1 amine of the heterocycle and the carbonyl of G291.49 further stabilizes the interaction. Molecular dynamics simulations predicted two metastable intermediates, one resembling a pose determined by molecular docking and a second involving transient interactions with Y2938.54; in simulations, each of these intermediates converges into the final bound state. Structure-activity-relationships for replacement of either of the identified exocyclic or endocyclic amines with heteroatoms lacking H-bond donating ability were consistent with the hypothetical pose. Thus, we characterized an allosteric pocket for 1H-imidazo[4,5-c]quinolin-4-amines that is consistent with data generated by orthogonal methods, which will aid in the rational design of improved A3AR positive allosteric modulators. SIGNIFICANCE STATEMENT: Orthosteric A3AR agonists have advanced in clinical trials for inflammatory conditions, liver diseases, and cancer. Thus, the clinical appeal of selective receptor activation could extend to allosteric enhancers, which would induce site- and time-specific activation in the affected tissue. By identifying the allosteric site for known positive allosteric modulators, structure-based drug discovery modalities can be enabled to enhance the pharmacological properties of the 1H-imidazo[4,5-c]quinolin-4-amine class of A3AR positive allosteric modulators.

Thermal titration molecular dynamics (TTMD): shedding light on the stability of RNA-small molecule complexes

Ribonucleic acids are gradually becoming relevant players among putative drug targets, thanks to the increasing amount of structural data exploitable for the rational design of selective and potent binders that can modulate their activity. Mainly, this information allows employing different computational techniques for predicting how well would a ribonucleic-targeting agent fit within the active site of its target macromolecule. Due to some intrinsic peculiarities of complexes involving nucleic acids, such as structural plasticity, surface charge distribution, and solvent-mediated interactions, the application of routinely adopted methodologies like molecular docking is challenged by scoring inaccuracies, while more physically rigorous methods such as molecular dynamics require long simulation times which hamper their conformational sampling capabilities. In the present work, we present the first application of Thermal Titration Molecular Dynamics (TTMD), a recently developed method for the qualitative estimation of unbinding kinetics, to characterize RNA-ligand complexes. In this article, we explored its applicability as a post-docking refinement tool on RNA in complex with small molecules, highlighting the capability of this method to identify the native binding mode among a set of decoys across various pharmaceutically relevant test cases.

Federated Learning in Computational Toxicology: An Industrial Perspective on the Effiris Hackathon

In silico approaches have acquired a towering role in pharmaceutical research and development, allowing laboratories all around the world to design, create, and optimize novel molecular entities with unprecedented efficiency. From a toxicological perspective, computational methods have guided the choices of medicinal chemists toward compounds displaying improved safety profiles. Even if the recent advances in the field are significant, many challenges remain active in the on-target and off-target prediction fields. Machine learning methods have shown their ability to identify molecules with safety concerns. However, they strongly depend on the abundance and diversity of data used for their training. Sharing such information among pharmaceutical companies remains extremely limited due to confidentiality reasons, but in this scenario, a recent concept named "federated learning" can help overcome such concerns. Within this framework, it is possible for companies to contribute to the training of common machine learning algorithms, using, but not sharing, their proprietary data. Very recently, Lhasa Limited organized a hackathon involving several industrial partners in order to assess the performance of their federated learning platform, called "Effiris". In this paper, we share our experience as Roche in participating in such an event, evaluating the performance of the federated algorithms and comparing them with those coming from our in-house-only machine learning models. Our aim is to highlight the advantages of federated learning and its intrinsic limitations and also suggest some points for potential improvements in the method.

Molecular Modeling Unveils the Effective Interaction of B-RAF Inhibitors with Rare B-RAF Insertion Variants

The Food and Drug Administration (FDA) has approved MAPK inhibitors as a treatment for melanoma patients carrying a mutation in codon V600 of the BRAF gene exclusively. However, BRAF mutations outside the V600 codon may occur in a small percentage of melanomas. Although these rare variants may cause B-RAF activation, their predictive response to B-RAF inhibitor treatments is still poorly understood. We exploited an integrated approach for mutation detection, tumor evolution tracking, and assessment of response to treatment in a metastatic melanoma patient carrying the rare p.T599dup B-RAF mutation. He was addressed to Dabrafenib/Trametinib targeted therapy, showing an initial dramatic response. In parallel, in-silico ligand-based homology modeling was set up and performed on this and an additional B-RAF rare variant (p.A598_T599insV) to unveil and justify the success of the B-RAF inhibitory activity of Dabrafenib, showing that it could adeptly bind both these variants in a similar manner to how it binds and inhibits the V600E mutant. These findings open up the possibility of broadening the spectrum of BRAF inhibitor-sensitive mutations beyond mutations at codon V600, suggesting that B-RAF V600 WT melanomas should undergo more specific investigations before ruling out the possibility of targeted therapy.

Past, Present, and Future Perspectives on Computer-Aided Drug Design Methodologies

The application of computational approaches in drug discovery has been consolidated in the last decades. These families of techniques are usually grouped under the common name of "computer-aided drug design" (CADD), and they now constitute one of the pillars in the pharmaceutical discovery pipelines in many academic and industrial environments. Their implementation has been demonstrated to tremendously improve the speed of the early discovery steps, allowing for the proficient and rational choice of proper compounds for a desired therapeutic need among the extreme vastness of the drug-like chemical space. Moreover, the application of CADD approaches allows the rationalization of biochemical and interactive processes of pharmaceutical interest at the molecular level. Because of this, computational tools are now extensively used also in the field of rational 3D design and optimization of chemical entities starting from the structural information of the targets, which can be experimentally resolved or can also be obtained with other computer-based techniques. In this work, we revised the state-of-the-art computer-aided drug design methods, focusing on their application in different scenarios of pharmaceutical and biological interest, not only highlighting their great potential and their benefits, but also discussing their actual limitations and eventual weaknesses. This work can be considered a brief overview of computational methods for drug discovery.

Targeting the I7L Protease: A Rational Design for Anti-Monkeypox Drugs?

The latest monkeypox virus outbreak in 2022 showcased the potential threat of this viral zoonosis to public health. The lack of specific treatments against this infection and the success of viral protease inhibitors-based treatments against HIV, Hepatitis C, and SARS-CoV-2, brought the monkeypox virus I7L protease under the spotlight as a potential target for the development of specific and compelling drugs against this emerging disease. In the present work, the structure of the monkeypox virus I7L protease was modeled and thoroughly characterized through a dedicated computational study. Furthermore, structural information gathered in the first part of the study was exploited to virtually screen the DrugBank database, consisting of drugs approved by the Food and Drug Administration (FDA) and clinical-stage drug candidates, in search for readily repurposable compounds with similar binding features as TTP-6171, the only non-covalent I7L protease inhibitor reported in the literature. The virtual screening resulted in the identification of 14 potential inhibitors of the monkeypox I7L protease. Finally, based on data collected within the present work, some considerations on developing allosteric modulators of the I7L protease are reported.

Lessons Learnt from COVID-19: Computational Strategies for Facing Present and Future Pandemics

Since its outbreak in December 2019, the COVID-19 pandemic has caused the death of more than 6.5 million people around the world. The high transmissibility of its causative agent, the SARS-CoV-2 virus, coupled with its potentially lethal outcome, provoked a profound global economic and social crisis. The urgency of finding suitable pharmacological tools to tame the pandemic shed light on the ever-increasing importance of computer simulations in rationalizing and speeding up the design of new drugs, further stressing the need for developing quick and reliable methods to identify novel active molecules and characterize their mechanism of action. In the present work, we aim at providing the reader with a general overview of the COVID-19 pandemic, discussing the hallmarks in its management, from the initial attempts at drug repurposing to the commercialization of Paxlovid, the first orally available COVID-19 drug. Furthermore, we analyze and discuss the role of computer-aided drug discovery (CADD) techniques, especially those that fall in the structure-based drug design (SBDD) category, in facing present and future pandemics, by showcasing several successful examples of drug discovery campaigns where commonly used methods such as docking and molecular dynamics have been employed in the rational design of effective therapeutic entities against COVID-19.

Thermal Titration Molecular Dynamics (TTMD): Not Your Usual Post-Docking Refinement

Molecular docking is one of the most widely used computational approaches in the field of rational drug design, thanks to its favorable balance between the rapidity of execution and the accuracy of provided results. Although very efficient in exploring the conformational degrees of freedom available to the ligand, docking programs can sometimes suffer from inaccurate scoring and ranking of generated poses. To address this issue, several post-docking filters and refinement protocols have been proposed throughout the years, including pharmacophore models and molecular dynamics simulations. In this work, we present the first application of Thermal Titration Molecular Dynamics (TTMD), a recently developed method for the qualitative estimation of protein-ligand unbinding kinetics, to the refinement of docking results. TTMD evaluates the conservation of the native binding mode throughout a series of molecular dynamics simulations performed at progressively increasing temperatures with a scoring function based on protein-ligand interaction fingerprints. The protocol was successfully applied to retrieve the native-like binding pose among a set of decoy poses of drug-like ligands generated on four different pharmaceutically relevant biological targets, including casein kinase 1δ, casein kinase 2, pyruvate dehydrogenase kinase 2, and SARS-CoV-2 main protease.

"Dual Anta-Inhibitors" of the A2A Adenosine Receptor and Casein Kinase CK1delta: Synthesis, Biological Evaluation, and Molecular Modeling Studies

Based on a screening of a chemical library of A2A adenosine receptor (AR) antagonists, a series of di- and tri-substituted adenine derivatives were synthesized and tested for their ability to inhibit the activity of the enzyme casein kinase 1 delta (CK1δ) and to bind adenosine receptors (ARs). Some derivatives, here called "dual anta-inhibitors", demonstrated good CK1δ inhibitory activity combined with a high binding affinity, especially for the A2AAR. The N6-methyl-(2-benzimidazolyl)-2-dimethyamino-9-cyclopentyladenine (17, IC50 = 0.59 μM and KiA2A = 0.076 μM) showed the best balance of A2AAR affinity and CK1δ inhibitory activity. Computational studies were performed to simulate, at the molecular level, the protein-ligand interactions involving the compounds of our series. Hence, the dual anta-inhibitor 17 could be considered the lead compound of new therapeutic agents endowed with synergistic effects for the treatment of chronic neurodegenerative and cancer diseases.

Qualitative Estimation of Protein-Ligand Complex Stability through Thermal Titration Molecular Dynamics Simulations

The prediction of ligand efficacy has long been linked to thermodynamic properties such as the equilibrium dissociation constant, which considers both the association and the dissociation rates of a defined protein-ligand complex. In the last 15 years, there has been a paradigm shift, with an increased interest in the determination of kinetic properties such as the drug-target residence time since they better correlate with ligand efficacy compared to other parameters. In this article, we present thermal titration molecular dynamics (TTMD), an alternative computational method that combines a series of molecular dynamics simulations performed at progressively increasing temperatures with a scoring function based on protein-ligand interaction fingerprints for the qualitative estimation of protein-ligand-binding stability. The protocol has been applied to four different pharmaceutically relevant test cases, including protein kinase CK1δ, protein kinase CK2, pyruvate dehydrogenase kinase 2, and SARS-CoV-2 main protease, on a variety of ligands with different sizes, structures, and experimentally determined affinity values. In all four cases, TTMD was successfully able to distinguish between high-affinity compounds (low nanomolar range) and low-affinity ones (micromolar), proving to be a useful screening tool for the prioritization of compounds in a drug discovery campaign.