MOViDA: multiomics visible drug activity prediction with a biologically informed neural network model.

MOTIVATION: The process of drug development is inherently complex, marked by extended intervals from the inception of a pharmaceutical agent to its eventual launch in the market. Additionally, each phase in this process is associated with a significant failure rate, amplifying the inherent challenges of this task. Computational virtual screening powered by machine learning algorithms has emerged as a promising approach for predicting therapeutic efficacy. However, the complex relationships between the features learned by these algorithms can be challenging to decipher. RESULTS: We have engineered an artificial neural network model designed specifically for predicting drug sensitivity. This model utilizes a biologically informed visible neural network, thereby enhancing its interpretability. The trained model allows for an in-depth exploration of the biological pathways integral to prediction and the chemical attributes of drugs that impact sensitivity. Our model harnesses multiomics data derived from a different tumor tissue sources, as well as molecular descriptors that encapsulate the properties of drugs. We extended the model to predict drug synergy, resulting in favorable outcomes while retaining interpretability. Given the imbalanced nature of publicly available drug screening datasets, our model demonstrated superior performance to state-of-the-art visible machine learning algorithms. AVAILABILITY AND IMPLEMENTATION: MOViDA is implemented in Python using PyTorch library and freely available for download at https://github.com/Luigi-Ferraro/MOViDA. Training data, RIS score and drug features are archived on Zenodo https://doi.org/10.5281/zenodo.8180380.

CACHE Challenge #1: Targeting the WDR Domain of LRRK2, A Parkinson's Disease Associated Protein.

The CACHE challenges are a series of prospective benchmarking exercises to evaluate progress in the field of computational hit-finding. Here we report the results of the inaugural CACHE challenge in which 23 computational teams each selected up to 100 commercially available compounds that they predicted would bind to the WDR domain of the Parkinson's disease target LRRK2, a domain with no known ligand and only an apo structure in the PDB. The lack of known binding data and presumably low druggability of the target is a challenge to computational hit finding methods. Of the 1955 molecules predicted by participants in Round 1 of the challenge, 73 were found to bind to LRRK2 in an SPR assay with a KD lower than 150 µM. These 73 molecules were advanced to the Round 2 hit expansion phase, where computational teams each selected up to 50 analogs. Binding was observed in two orthogonal assays for seven chemically diverse series, with affinities ranging from 18 to 140 µM. The seven successful computational workflows varied in their screening strategies and techniques. Three used molecular dynamics to produce a conformational ensemble of the targeted site, three included a fragment docking step, three implemented a generative design strategy and five used one or more deep learning steps. CACHE #1 reflects a highly exploratory phase in computational drug design where participants adopted strikingly diverging screening strategies. Machine learning-accelerated methods achieved similar results to brute force (e.g., exhaustive) docking. First-in-class, experimentally confirmed compounds were rare and weakly potent, indicating that recent advances are not sufficient to effectively address challenging targets.

Enzymatic Synthesis and Structural Modeling of Bio-Based Oligoesters as an Approach for the Fast Screening of Marine Biodegradation and Ecotoxicity.

Given the widespread use of esters and polyesters in products like cosmetics, fishing nets, lubricants and adhesives, whose specific application(s) may cause their dispersion in open environments, there is a critical need for stringent eco-design criteria based on biodegradability and ecotoxicity evidence. Our approach integrates experimental and computational methods based on short oligomers, offering a screening tool for the rapid identification of sustainable monomers and oligomers, with a special focus on bio-based alternates. We provide insights into the relationships between the chemical structure and properties of bio-based oligomers in terms of biodegradability in marine environments and toxicity in benchmark organisms. The experimental results reveal that the considered aromatic monomers (terephthalic acid and 2,5-furandicarboxylic acid) accumulate under the tested conditions (OECD 306), although some slight biodegradation is observable when the inoculum derives from sites affected by industrial and urban pollution, which suggests that ecosystems adapt to non-natural chemical pollutants. While clean seas are more susceptible to toxic chemical buildup, biotic catalytic activities offer promise for plastic pollution mitigation. Without prejudice to the fact that biodegradability inherently signifies a desirable trait in plastic products, nor that it automatically grants them a sustainable "license", this study is intended to facilitate the rational design of new polymers and materials on the basis of specific uses and applications.

Combining machine learning and quantum mechanics yields more chemically aware molecular descriptors for medicinal chemistry applications.

Molecular interaction fields (MIFs), describing molecules in terms of their ability to interact with any chemical entity, are one of the most established and versatile concepts in drug discovery. Improvement of this molecular description is highly desirable for in silico drug discovery and medicinal chemistry applications. In this work, we revised a well-established molecular mechanics' force field and applied a hybrid quantum mechanics and machine learning approach to parametrize the hydrogen-bonding (HB) potentials of small molecules, improving this aspect of the molecular description. Approximately 66,000 molecules were chosen from available drug databases and subjected to density functional theory calculations (DFT). For each atom, the molecular electrostatic potential (EP) was extracted and used to derive new HB energy contributions; this was subsequently combined with a fingerprint-based description of the structural environment via partial least squares modeling, enabling the new potentials to be used for molecules outside of the training set. We demonstrate that parameter prediction for molecules outside of the training set correlates with their DFT-derived EP, and that there is correlation of the new potentials with hydrogen-bond acidity and basicity scales. We show the newly derived MIFs vary in strength for various ring substitution in accordance with chemical intuition. Finally, we report that this derived parameter, when extended to non-HB atoms, can also be used to estimate sites of reaction.

Playing with opening and closing of heterocycles: using the cusmano-ruccia reaction to develop a novel class of oxadiazolothiazinones, active as calcium channel modulators and P-glycoprotein inhibitors.

As a result of the ring-into-ring conversion of nitrosoimidazole derivatives, we obtained a molecular scaffold that, when properly decorated, is able to decrease inotropy by blocking L-type calcium channels. Previously, we used this scaffold to develop a quantitative structure-activity relationship (QSAR) model, and we used the most potent oxadiazolothiazinone as a template for ligand-based virtual screening. Here, we enlarge the diversity of chemical decorations, present the synthesis and in vitro data for 11 new derivatives, and develop a new 3D-QSAR model with recent in silico techniques. We observed a key role played by the oxadiazolone moiety: given the presence of positively charged calcium ions in the transmembrane channel protein, we hypothesize the formation of a ternary complex between the oxadiazolothiazinone, the Ca2+ ion and the protein. We have supported this hypothesis by means of pharmacophore generation and through the docking of the pharmacophore into a homology model of the protein. We also studied with docking experiments the interaction with a homology model of P-glycoprotein, which is inhibited by this series of molecules, and provided further evidence toward the relevance of this scaffold in biological interactions.

Gaucher Disease: A Glance from a Medicinal Chemistry Perspective.

Rare diseases are particular pathological conditions affecting a limited number of people and few drugs are known to be effective as therapeutic treatment. Gaucher disease, caused by a deficiency of the lysosomal enzyme glucocerebrosidase, belongs to this class of disorders, and it is considered the most common among the Lysosomal Storage Diseases. The two main therapeutic approaches are the Enzyme Replacement Therapy (ERT) and the Substrate Reduction Therapy (SRT). ERT, consisting in replacing the defective enzyme by administering a recombinant enzyme, is effective in alleviating the visceral symptoms, hallmarks of the most common subtype of the disease whereas it has no effects when symptoms involve CNS, since the recombinant protein is unable to significantly cross the Blood Brain Barrier. The SRT strategy involves inhibiting glucosylceramide synthase (GCS), the enzyme responsible for the production of the associated storage molecule. The rational design of new inhibitors of GCS has been hampered by the lack of either the crystal structure of the enzyme or an in-silico model of the active site which could provide important information regarding the interactions of potential inhibitors with the target, but, despite this, interesting results have been obtained and are herein reviewed.

Modelling cytochromes P450 binding modes to predict P450 inhibition, metabolic stability and isoform selectivity.

The cytochromes P450 (P450) superfamily is a diverse group of enzymes involved in the metabolism of xenobiotics, whose orientations within the catalytic site can lead to different binding modes, namely productive, nonproductive, and inhibitory. This article collects the most recent approaches that individually study P450- ligand interactions, including a novel in silico technology, developed in the framework of the Human Cytochrome P450 Consortium initiative, that provides reliable in silico predictions of P450 inhibition, metabolic stability and isoform selectivity.

QSAR modeling and data mining link Torsades de Pointes risk to the interplay of extent of metabolism, active transport, and HERG liability.

We collected 1173 hERG patch clamp (PC) data (IC50) from the literature to derive twelve classification models for hERG inhibition, covering a large variety of chemical descriptors and classification algorithms. Models were generated using 545 molecules and validated through 258 external molecules tested in PC experiments. We also evaluated the suitability of the best models to predict the activity of 26 proprietary compounds tested in radioligand binding displacement (RBD). Results proved the necessity to use multiple validation sets for a true estimation of model accuracy and demonstrated that using various descriptors and algorithms improves the performance of ligand-based models. Intriguingly, one of the most accurate models uncovered an unexpected link between extent of metabolism and hERG liability. This hypothesis was fairly reinforced by using the Biopharmaceutics Drug Disposition Classification System (BDDCS) that recognized 94% of the hERG inhibitors as extensively metabolized in vivo. Data mining suggested that high Torsades de Pointes (TdP) risk results from an interplay of hERG inhibition, extent of metabolism, active transport, and possibly solubility. Overall, these new findings might improve both the decision making skills of pharmaceutical scientists to mitigate hERG liability during the drug discovery process and the TdP risk assessment during drug development.

Absolute configuration and biological profile of two thiazinooxadiazol-3-ones with L-type calcium channel activity: a study of the structural effects.

In the framework of our interest in racemic thiazinooxadiazol-3-ones we determined the absolute configuration and the biological activity as L-type calcium channel blockers of two compounds that differ in the length of the acetal chain, which could affect the pharmacological profile. We observed an interesting inversion of the stereoselectivity, with the activity residing on the R-form for a short chain compound (n = 1) and on the S-form for a long chain one (n = 12). The length of the linear acetal chain appears to be able to invert the stereoselectivity of such a class of compounds, and in silico simulations suggested that this different behaviour might be explained by different hydrophilic and hydrophobic interactions with the binding site.

Destabilizers of the thymidylate synthase homodimer accelerate its proteasomal degradation and inhibit cancer growth.

Drugs that target human thymidylate synthase (hTS), a dimeric enzyme, are widely used in anticancer therapy. However, treatment with classical substrate-site-directed TS inhibitors induces over-expression of this protein and development of drug resistance. We thus pursued an alternative strategy that led us to the discovery of TS-dimer destabilizers. These compounds bind at the monomer-monomer interface and shift the dimerization equilibrium of both the recombinant and the intracellular protein toward the inactive monomers. A structural, spectroscopic, and kinetic investigation has provided evidence and quantitative information on the effects of the interaction of these small molecules with hTS. Focusing on the best among them, E7, we have shown that it inhibits hTS in cancer cells and accelerates its proteasomal degradation, thus causing a decrease in the enzyme intracellular level. E7 also showed a superior anticancer profile to fluorouracil in a mouse model of human pancreatic and ovarian cancer. Thus, over sixty years after the discovery of the first TS prodrug inhibitor, fluorouracil, E7 breaks the link between TS inhibition and enhanced expression in response, providing a strategy to fight drug-resistant cancers.

FLAP: GRID molecular interaction fields in virtual screening. validation using the DUD data set.

The performance of FLAP (Fingerprints for Ligands and Proteins) in virtual screening is assessed using a subset of the DUD (Directory of Useful Decoys) benchmarking data set containing 13 targets each with more than 15 different chemotype classes. A variety of ligand and receptor-based virtual screening approaches are examined, using combinations of individual templates 2D structures of known actives, a cocrystallized ligand, a receptor structure, or a cocrystallized ligand-biased receptor structure. We examine several data fusion approaches to combine the results of the individual virtual screens. In doing so, we show that excellent chemotype enrichment is achieved in both single target ligand-based and receptor-based approaches, of approximately 17-fold over random on average at a false positive rate of 1%. We also show that using as much starting knowledge as possible improves chemotype enrichment, and that data fusion using Pareto ranking is an effective method to do this giving up to 50% improvement in enrichment over the single methods. Finally we show that if inactivity or decoy data is incorporated, automatically training the scoring function in FLAP improves recovery still further, with almost 2-fold improvement over the enrichments shown by the single methods. The results clearly demonstrate the utility of FLAP for virtual screening when either a limited or wide range of prior knowledge is available.

Ligand-based virtual screening and ADME-tox guided approach to identify triazolo-quinoxalines as folate cycle inhibitors.

In the process of drug discovery the lead-identification phase may be critical due to the likely poor safety profile of the candidates, causing the delay or even the abandonment of a certain project. Nowadays, combining molecular modeling and in vivo cellular evaluation can help to identify compounds with an enhanced safety profile. Previously, two quinoxalines have been identified as inhibitors of the folate-dependent proteins belonging to the thymidylate synthase cycle. Unfortunately, cytotoxic activity against a panel of cisplatin(cDDP)-sensitive ovarian carcinoma cell lines and their resistant counterparts was coupled with toxicity to non-tumorigenic Vero cells. Here we describe the application of a ligand-based virtual screening, and several [1,2,4]triazolo[4,3-a]quinoxalines were optimized to improve their ADME-tox profile. The resulting 4-(trifluoromethyl)-1-p-tolyl-[1,2,4]triazolo[4,3-a]quinoxaline (24), which interferes intracellularly with DHFR and TS reducing the protein levels like 5-FU, but without inducing TS ternary complex formation, was 2-times less toxic in vitro than cisplatin and 5-FU.

Diltiazem analogues: the last ten years on structure activity relationships.

Cardiovascular diseases as hypertension, angina and/or supraventricular arrhythmias are among the most important death causes in the world. For the treatment of heart pathologies, calcium channel entry blockers are very important drugs, owing to their therapeutic versatility. Although few calcium antagonists described until today are structurally related to diltiazem and to the benzothiazepine class, the still high pharmaceutical interest on diltiazem analogues justifies this review. Diltiazem and its first analogues developed in the early '70s became popular in the '80s, and were pharmacologically characterized for a long time. It is in the '90s that several research groups carried out structural variations identifying novel scaffolds for diltiazem-related compounds, with significant calcium antagonist behaviour. Recently, a series of thiazino-oxadiazolone derivatives were identified as potent and selective antagonists for calcium influx into cardiac cells, and they were subsequently used to search for novel chemotypes by means of virtual screening techniques. The resulting hits could open interesting perspectives for the development of drugs to treat cardiovascular diseases. In the present review, an updated collection of diltiazem analogues is reported over the last ten years. The chemical structure and the structure activity relationships will be given, with additional mention to the potential therapeutic applications.

Virtual screening for novel openers of pancreatic K(ATP) channels.

Ligand-based virtual screening approaches were applied to search for new chemotype KCOs activating Kir6.2/SUR1 KATP channels. A total of 65 208 commercially available compounds, extracted from the ZINC archive, served as database for screening. In a first step, pharmacokinetic filtering via VolSurf reduced the initial database to 1913 compounds. Afterward, six molecules were selected as templates for similarity searches: similarity scores, obtained toward these templates, were calculated with the GRIND, FLAP, and TOPP approaches, which differently encode structural information into potential pharmacophores. In this way, we obtained 32 hit candidates, 16 via GRIND and eight each via FLAP and TOPP. For biological testing of the hit candidates, their effects on membrane potentials in HEK 293 cells expressing Kir6.2/SUR1 were studied. GRIND, FLAP, and TOPP all yielded hits, but no method top-ranked all the actives. Thus, parallel application of different approaches probably improves hit detection.

ADME-Space: a new tool for medicinal chemists to explore ADME properties.

We introduce a new chemical space for drugs and drug-like molecules, exclusively based on their in silico ADME behaviour. This ADME-Space is based on self-organizing map (SOM) applied to 26,000 molecules. Twenty accurate QSPR models, describing important ADME properties, were developed and, successively, used as new molecular descriptors not related to molecular structure. Applications include permeability, active transport, metabolism and bioavailability studies, but the method can be even used to discuss drug-drug interactions (DDIs) or it can be extended to additional ADME properties. Thus, the ADME-Space opens a new framework for the multi-parametric data analysis in drug discovery where all ADME behaviours of molecules are condensed in one map: it allows medicinal chemists to simultaneously monitor several ADME properties, to rapidly select optimal ADME profiles, retrieve warning on potential ADME problems and DDIs or select proper in vitro experiments.

Side-Chain Modified Ergosterol and Stigmasterol Derivatives as Liver X Receptor Agonists.

A series of stigmasterol and ergosterol derivatives, characterized by the presence of oxygenated functions at C-22 and/or C-23 positions, were designed as potential liver X receptor (LXR) agonists. The absolute configuration of the newly created chiral centers was definitively assigned for all the corresponding compounds. Among the 16 synthesized compounds, 21, 27, and 28 were found to be selective LXRα agonists, whereas 20, 22, and 25 showed good selectivity for the LXRß isoform. In particular, 25 showed the same degree of potency as 22R-HC (3) at LXRß, while it was virtually inactive at LXRα (EC50 = 14.51 µM). Interestingly, 13, 19, 20, and 25 showed to be LXR target gene-selective modulators, by strongly inducing the expression of ABCA1, while poorly or not activating the lipogenic genes SREBP1 and SCD1 or FASN, respectively.

Calcium channel antagonists discovered by a multidisciplinary approach.

A multidisciplinary project has led to the discovery of novel, structurally diverse, L-type calcium entry blockers (CEBs). The absolute configuration of a recently reported CEB has been determined by vibrational circular dichroism spectroscopy, to assign the stereospecificity of the ligand-channel interaction. Thereafter, a virtual screening procedure was performed with the aim of identifying novel chemotypes for CEBs, starting from a database of purchasable compounds; 340,000 molecules were screened in silico in order to prioritize structures of interest for bioscreening. As a result, 20 compounds were tested in vitro, and functional and binding assays revealed several hits with promising behavior as CEBs.

Understanding Oxadiazolothiazinone Biological Properties: Negative Inotropic Activity versus Cytochrome P450-Mediated Metabolism.

We present a series of oxadiazolothiazinones, selective inotropic agents on isolated cardiac tissues, devoid of chronotropy and vasorelaxant activity. Functional and binding data for the precursor of the series (compound 1) let us hypothesize LTCC blocking activity and the existence of a recognition site specific for this scaffold. We synthesized and tested 22 new derivatives: introducing a para-methoxyphenyl at C-8 led to compound 12 (EC50 = 0.022 µM), twice as potent as its para-bromo analogue (1). For 10 analogues, we extended the characterization of the biological properties by including the assessment of metabolic stability in human liver microsomes and cytochrome P450 inhibition potential. We observed that the methoxy group led to active compounds with low metabolic stability and high CYP inhibition, whereas the protective effect of bromine resulted in enhanced metabolic stability and reduced CYP inhibition. Thus, we identified two para-bromo benzothiazino-analogues as candidates for further studies.

Comparison of ligand-based and structure-based 3D-QSAR approaches: a case study on (aryl-)bridged 2-aminobenzonitriles inhibiting HIV-1 reverse transcriptase.

Ligand- (GRIND) and structure-based (GLUE/GRIND) 3D-QSAR approaches were compared for 55 (aryl-)bridged 2-aminobenzonitriles inhibiting HIV-1 reverse transcriptase (HIV-1 RT). The ligand-based model was built from conformers selected by in vacuo minimization. The available X-ray structure of 3v in complex with HIV-1 RT allowed comparative structure-based calculations using the new docking software GLUE for conformer selection. Both models were validated via statistics and via virtual receptor sites (VRS) considering pharmacophoric regions and mutual distances, which were also compared with experimental evidence. The statistics show slight superiority of the structure-based approach in terms of fitting and prediction. By encoding relevant molecular interaction fields (MIF) into pharmacophoric regions, 10 such regions were derived from both models; they all fit the real receptor except HBD2. Also mutual distances highly agree between the real site and both VRS. Although distances from the structure-based approach are closer to the real receptor, present data prove the validity of the ligand-based GRIND approach.

A new class of selective myocardial calcium channel modulators. 2. Role of the acetal chain in oxadiazol-3-one derivatives.

In the framework of the continuing interest of this research group in the use of 8-aryl-8-hydroxy-8H-[1,4]thiazino[3,4-c][1,2,4]oxadiazol-3-ones (1) as calcium entry blockers, a number of acetals were synthesized and assayed "in vitro". All of them are structurally related to diltiazem and pyrrolobenzothiazines. The effect on the biological profile was measured by functional assays for a wide variety of acetal residues: saturated linear and branched chains, short and long unsaturated E and/or Z chains as well as benzyl and methylcyclohexyl residues. From selective assays on the most active derivative (5b) (EC(50) = 0.04 microM), which is 20 times more active than diltiazem (EC(50) = 0.79 microM), a muscarinic or adenosinic mechanism of action was excluded. A 3D QSAR model was obtained and validated with homologous literature data, and a virtual receptor scheme was derived for the unknown binding site. The following pharmacophoric features favorably affect the potency: one positively charged center, three lipophilic groups, and two hydrogen-bonding acceptor groups.