Using Machine Learning and Molecular Docking to Leverage Urease Inhibition Data for Virtual Screening.

Urease is a metalloenzyme that catalyzes the hydrolysis of urea, and its modulation has an important role in both the agricultural and medical industry. Even though numerous molecules have been tested against ureases of different species, their clinical translation has been limited due to chemical and metabolic stability as well as side effects. Therefore, screening new compounds against urease would be of interest in part due to rising concerns regarding antibiotic resistance. In this work, we collected and curated a diverse set of 2640 publicly available small-molecule inhibitors of jack bean urease and developed a classifier using a random forest machine learning method with high predictive performance. In addition, the physicochemical features of compounds were paired with molecular docking and protein-ligand fingerprint analysis to gather insight into the current activity landscape. We observed that the docking score could not differentiate active from inactive compounds within each chemical family, but scores were correlated with compound activity when all compounds were considered. Additionally, a decision tree model was built based on 2D and 3D Morgan fingerprints to mine patterns of the known active-class compounds. The final machine learning model showed good prediction performance against the test set (81% and 77% precision for active and inactive compounds, respectively). Finally, this model was employed, as a proof-of-concept, on an in-house library to predict new hits that were then tested against urease and found to be active. This is, to date, the largest, most diverse dataset of compounds used to develop predictive in silico models. Overall, the results highlight the usefulness of using machine learning classifiers and molecular docking to predict novel urease inhibitors.

Harnessing Protein-Ligand Interaction Fingerprints to Predict New Scaffolds of RIPK1 Inhibitors.

Necroptosis has emerged as an exciting target in oncological, inflammatory, neurodegenerative, and autoimmune diseases, in addition to acute ischemic injuries. It is known to play a role in innate immune response, as well as in antiviral cellular response. Here we devised a concerted in silico and experimental framework to identify novel RIPK1 inhibitors, a key necroptosis factor. We propose the first in silico model for the prediction of new RIPK1 inhibitor scaffolds by combining docking and machine learning methodologies. Through the data analysis of patterns in docking results, we derived two rules, where rule #1 consisted of a four-residue signature filter, and rule #2 consisted of a six-residue similarity filter based on docking calculations. These were used in consensus with a machine learning QSAR model from data collated from ChEMBL, the literature, in patents, and from PubChem data. The models allowed for good prediction of actives of >90, 92, and 96.4% precision, respectively. As a proof-of-concept, we selected 50 compounds from the ChemBridge database, using a consensus of both molecular docking and machine learning methods, and tested them in a phenotypic necroptosis assay and a biochemical RIPK1 inhibition assay. A total of 7 of the 47 tested compounds demonstrated around 20−25% inhibition of RIPK1's kinase activity but, more importantly, these compounds were discovered to occupy new areas of chemical space. Although no strong actives were found, they could be candidates for further optimization, particularly because they have new scaffolds. In conclusion, this screening method may prove valuable for future screening efforts as it allows for the exploration of new areas of the chemical space in a very fast and inexpensive manner, therefore providing efficient starting points amenable to further hit-optimization campaigns.

Exploring the Chemical Space of Urease Inhibitors to Extract Meaningful Trends and Drivers of Activity.

Blocking the catalytic activity of urease has been shown to have a key role in different diseases as well as in different agricultural applications. A vast array of molecules have been tested against ureases of different species, but the clinical translation of these compounds has been limited due to challenges of potency, chemical and metabolic stability as well as promiscuity against other proteins. The design and development of new compounds greatly benefit from insights from previously tested compounds; however, no large-scale studies surveying the urease inhibitors' chemical space exist that can provide an overview of developed compounds to data. Therefore, given the increasing interest in developing new compounds for this target, we carried out a comprehensive analysis of the activity landscape published so far. To do so, we assembled and curated a data set of compounds tested against urease. To the best of our knowledge, this is the largest data set of urease inhibitors to date, composed of 3200 compounds of diverse structures. We characterized the data set in terms of chemical space coverage, molecular scaffolds, distribution with respect to physicochemical properties, as well as temporal trends of drug development. Through these analyses, we highlighted different substructures and functional groups responsible for distinct activity and inactivity against ureases. Furthermore, activity cliffs were assessed, and the chemical space of urease inhibitors was compared to DrugBank. Finally, we extracted meaningful patterns associated with activity using a decision tree algorithm. Overall, this study provides a critical overview of urease inhibitor research carried out in the last few decades and enabled finding underlying SAR patterns such as under-reported chemical functional groups that contribute to the overall activity. With this work, we propose different rules and practical implications that can guide the design or selection of novel compounds to be screened as well as lead optimization.

Revisiting Proteasome Inhibitors: Molecular Underpinnings of Their Development, Mechanisms of Resistance and Strategies to Overcome Anti-Cancer Drug Resistance.

Proteasome inhibitors have shown relevant clinical activity in several hematological malignancies, namely in multiple myeloma and mantle cell lymphoma, improving patient outcomes such as survival and quality of life, when compared with other therapies. However, initial response to the therapy is a challenge as most patients show an innate resistance to proteasome inhibitors, and those that respond to the therapy usually develop late relapses suggesting the development of acquired resistance. The mechanisms of resistance to proteasome inhibition are still controversial and scarce in the literature. In this review, we discuss the development of proteasome inhibitors and the mechanisms of innate and acquired resistance to their activity-a major challenge in preclinical and clinical therapeutics. An improved understanding of these mechanisms is crucial to guiding the design of new and more effective drugs to tackle these devastating diseases. In addition, we provide a comprehensive overview of proteasome inhibitors used in combination with other chemotherapeutic agents, as this is a key strategy to combat resistance.

An In Silico and an In Vitro Inhibition Analysis of Glycogen Phosphorylase by Flavonoids, Styrylchromones, and Pyrazoles.

Glycogen phosphorylase (GP) is a key enzyme in the glycogenolysis pathway. GP inhibitors are currently under investigation as a new liver-targeted approach to managing type 2 diabetes mellitus (DM). The aim of the present study was to evaluate the inhibitory activity of a panel of 52 structurally related chromone derivatives; namely, flavonoids, 2-styrylchromones, 2-styrylchromone-related derivatives [2-(4-arylbuta-1,3-dien-1-yl)chromones], and 4- and 5-styrylpyrazoles against GP, using in silico and in vitro microanalysis screening systems. Several of the tested compounds showed a potent inhibitory effect. The structure-activity relationship study indicated that for 2-styrylchromones and 2-styrylchromone-related derivatives, the hydroxylations at the A and B rings, and in the flavonoid family, as well as the hydroxylation of the A ring, were determinants for the inhibitory activity. To support the in vitro experimental findings, molecular docking studies were performed, revealing clear hydrogen bonding patterns that favored the inhibitory effects of flavonoids, 2-styrylchromones, and 2-styrylchromone-related derivatives. Interestingly, the potency of the most active compounds increased almost four-fold when the concentration of glucose increased, presenting an IC50 < 10 µM. This effect may reduce the risk of hypoglycemia, a commonly reported side effect of antidiabetic agents. This work contributes with important considerations and provides a better understanding of potential scaffolds for the study of novel GP inhibitors.

Exploring EZH2-Proteasome Dual-Targeting Drug Discovery through a Computational Strategy to Fight Multiple Myeloma.

Multiple myeloma is an incurable plasma cell neoplastic disease representing about 10-15% of all haematological malignancies diagnosed in developed countries. Proteasome is a key player in multiple myeloma and proteasome inhibitors are the current first-line of treatment. However, these are associated with limited clinical efficacy due to acquired resistance. One of the solutions to overcome this problem is a polypharmacology approach, namely combination therapy and multitargeting drugs. Several polypharmacology avenues are currently being explored. The simultaneous inhibition of EZH2 and Proteasome 20S remains to be investigated, despite the encouraging evidence of therapeutic synergy between the two. Therefore, we sought to bridge this gap by proposing a holistic in silico strategy to find new dual-target inhibitors. First, we assessed the characteristics of both pockets and compared the chemical space of EZH2 and Proteasome 20S inhibitors, to establish the feasibility of dual targeting. This was followed by molecular docking calculations performed on EZH2 and Proteasome 20S inhibitors from ChEMBL 25, from which we derived a predictive model to propose new EZH2 inhibitors among Proteasome 20S compounds, and vice versa, which yielded two dual-inhibitor hits. Complementarily, we built a machine learning QSAR model for each target but realised their application to our data is very limited as each dataset occupies a different region of chemical space. We finally proceeded with molecular dynamics simulations of the two docking hits against the two targets. Overall, we concluded that one of the hit compounds is particularly promising as a dual-inhibitor candidate exhibiting extensive hydrogen bonding with both targets. Furthermore, this work serves as a framework for how to rationally approach a dual-targeting drug discovery project, from the selection of the targets to the prediction of new hit compounds.

Pyrazoles as novel protein tyrosine phosphatase 1B (PTP1B) inhibitors: An in vitro and in silico study.

Type 2 diabetes mellitus (DM) is a complex chronic disorder and a major global health problem. Insulin resistance is the primary detectable abnormality and the main characteristic feature in individuals with type 2 DM. Protein tyrosine phosphatase 1B (PTP1B) is a key negative regulator of the insulin signaling pathway, which dephosphorylates insulin receptor and insulin receptor substrates, suppressing the insulin signaling cascade. Therefore, the inhibition of PTP1B has become a potential strategy in the management of type 2 DM. In this study, a library of 22 pyrazoles was evaluated here for the first time against human PTP1B activity, using a microanalysis screening system. The results showed that 5-(2-hydroxyphenyl)-3-{2-[3-(4-nitrophenyl)-1,2,3,4-tetrahydronaphthyl]}-1-phenylpyrazole 20 and 3-(2-hydroxyphenyl)-5-{2-[3-(4-methoxyphenyl)]naphthyl}pyrazole 22 excelled as the most potent inhibitors of PTP1B, through noncompetitive inhibition mechanism. These findings suggest that the presence of additional benzene rings as functional groups in the pyrazole moiety increases the ability of pyrazoles to inhibit PTP1B. The most active compounds showed selectivity over the homologous T-cell protein tyrosine phosphatase (TCPTP). Molecular docking analyses were performed and revealed a particular contact signature involving residues like TYR46, ASP48, PHE182, TYR46, ALA217 and ILE219. This study represents a significant beginning for the design of novel PTP1B inhibitors.

Phenotypic high-throughput screening platform identifies novel chemotypes for necroptosis inhibition.

Regulated necrosis or necroptosis, mediated by receptor-interacting kinase 1 (RIPK1), RIPK3 and pseudokinase mixed lineage kinase domain-like protein (MLKL), contributes to the pathogenesis of inflammatory, infectious and degenerative diseases. Recently identified necroptosis inhibitors display moderate specificity, suboptimal pharmacokinetics, off-target effects and toxicity, preventing these molecules from reaching the clinic. Here, we developed a cell-based high-throughput screening (HTS) cascade for the identification of small-molecule inhibitors of necroptosis. From the initial library of over 250,000 compounds, the primary screening phase identified 356 compounds that strongly inhibited TNF-α-induced necroptosis, but not apoptosis, in human and murine cell systems, with EC50 < 6.7 µM. From these, 251 compounds were tested for RIPK1 and/or RIPK3 kinase inhibitory activity; some were active and several have novel mechanisms of action. Based on specific chemical descriptors, 110 compounds proceeded into the secondary screening cascade, which then identified seven compounds with maximum ability to reduce MLKL activation, IC50 >100 µM, EC50 2.5-11.5 µM under long-term necroptosis execution in murine fibroblast L929 cells, and full protection from ATP depletion and membrane leakage in human and murine cells. As a proof of concept, compound SN-6109, with binding mode to RIPK1 similar to that of necrostatin-1, confirmed RIPK1 inhibitory activity and appropriate pharmacokinetic properties. SN-6109 was further tested in mice, showing efficacy against TNF-α-induced systemic inflammatory response syndrome. In conclusion, a phenotypic-driven HTS cascade promptly identified robust necroptosis inhibitors with in vivo activity, currently undergoing further medicinal chemistry optimization. Notably, the novel hits highlight the opportunity to identify new molecular mechanisms of action in necroptosis.

Chemical Patterns of Proteasome Inhibitors: Lessons Learned from Two Decades of Drug Design.

Drug discovery now faces a new challenge, where the availability of experimental data is no longer the limiting step, and instead, making sense of the data has gained a new level of importance, propelled by the extensive incorporation of cheminformatics and bioinformatics methodologies into the drug discovery and development pipeline. These enable, for example, the inference of structure-activity relationships that can be useful in the discovery of new drug candidates. One of the therapeutic applications that could benefit from this type of data mining is proteasome inhibition, given that multiple compounds have been designed and tested for the last 20 years, and this collection of data is yet to be subjected to such type of assessment. This study presents a retrospective overview of two decades of proteasome inhibitors development (680 compounds), in order to gather what could be learned from them and apply this knowledge to any future drug discovery on this subject. Our analysis focused on how different chemical descriptors coupled with statistical tools can be used to extract interesting patterns of activity. Multiple instances of the structure-activity relationship were observed in this dataset, either for isolated molecular descriptors (e.g., molecular refractivity and topological polar surface area) as well as scaffold similarity or chemical space overlap. Building a decision tree allowed the identification of two meaningful decision rules that describe the chemical parameters associated with high activity. Additionally, a characterization of the prevalence of key functional groups gives insight into global patterns followed in drug discovery projects, and highlights some systematically underexplored parts of the chemical space. The various chemical patterns identified provided useful insight that can be applied in future drug discovery projects, and give an overview of what has been done so far.

Applying synergy metrics to combination screening data: agreements, disagreements and pitfalls.

Synergistic drug combinations are commonly sought to overcome monotherapy resistance in cancer treatment. To identify such combinations, high-throughput cancer cell line combination screens are performed; and synergy is quantified using competing models based on fundamentally different assumptions. Here, we compare the behaviour of four synergy models, namely Loewe additivity, Bliss independence, highest single agent and zero interaction potency, using the Merck oncology combination screen. We evaluate agreements and disagreements between the models and investigate putative artefacts of each model's assumptions. Despite at least moderate concordance between scores (Pearson's r >0.32, Spearman's ρ>0.34), multiple instances of strong disagreement were observed. Those disagreements are driven by, among others, large differences in tested concentrations, maximum response values and median effective concentrations.

Conformal Regression for Quantitative Structure-Activity Relationship Modeling-Quantifying Prediction Uncertainty.

Making predictions with an associated confidence is highly desirable as it facilitates decision making and resource prioritization. Conformal regression is a machine learning framework that allows the user to define the required confidence and delivers predictions that are guaranteed to be correct to the selected extent. In this study, we apply conformal regression to model molecular properties and bioactivity values and investigate different ways to scale the resultant prediction intervals to create as efficient (i.e., narrow) regressors as possible. Different algorithms to estimate the prediction uncertainty were used to normalize the prediction ranges, and the different approaches were evaluated on 29 publicly available data sets. Our results show that the most efficient conformal regressors are obtained when using the natural exponential of the ensemble standard deviation from the underlying random forest to scale the prediction intervals, but other approaches were almost as efficient. This approach afforded an average prediction range of 1.65 pIC50 units at the 80% confidence level when applied to bioactivity modeling. The choice of nonconformity function has a pronounced impact on the average prediction range with a difference of close to one log unit in bioactivity between the tightest and widest prediction range. Overall, conformal regression is a robust approach to generate bioactivity predictions with associated confidence.

Population Approach to Efavirenz Therapy.

Efavirenz (EFV) is a nonnucleoside reverse transcriptase inhibitor commonly used as first-line therapy in the treatment of human immunodeficiency virus (HIV), with a narrow therapeutic range and a high between-subject variability which can lead to central nervous system toxicity or therapeutic failure. To characterize the sources of variability and better predict EFV steady-state plasma concentrations, a population pharmacokinetic model was developed from 96 HIV-positive individuals, using a nonlinear mixed-effect method with Monolix® software. A one-compartment with first-order absorption and elimination model adequately described the data. To explain between-subject variability, demographic characteristics, biochemical parameters, hepatitis C virus-HIV coinfection, and genetic polymorphisms were tested. A combination of the single-nucleotide polymorphisms rs2279343 and rs3745274, both in the CYP2B6 gene, were the only covariates influencing clearance, included in the final model. Oral clearance was estimated to be 19.6 L/h, 14.15 L/h, and 6.08 L/h for wild-type, heterozygous mutated and homozygous mutated individuals, respectively. These results are in accordance with the current knowledge of EFV metabolism and also suggest that in homozygous mutated individuals, a dose adjustment is necessary. Hepatitis C virus-HIV coinfection does not seem to be a predictive indicator of EFV pharmacokinetic disposition.

Simultaneous Prediction of four ATP-binding Cassette Transporters' Substrates Using Multi-label QSAR.

Efflux by the ATP-binding cassette (ABC) transporters affects the pharmacokinetic profile of drugs and it has been implicated in drug-drug interactions as well as its major role in multi-drug resistance in cancer. It is therefore important for the pharmaceutical industry to be able to understand what phenomena rule ABC substrate recognition. Considering a high degree of substrate overlap between various members of ABC transporter family, it is advantageous to employ a multi-label classification approach where predictions made for one transporter can be used for modeling of the other ABC transporters. Here, we present decision tree-based QSAR classification models able to simultaneously predict substrates and non-substrates for BCRP1, P-gp/MDR1 and MRP1 and MRP2, using a dataset of 1493 compounds. To this end, two multi-label classification QSAR modelling approaches were adopted: Binary Relevance (BR) and Classifier Chain (CC). Even though both multi-label models yielded similar predictive performances in terms of overall accuracies (close to 70 %), the CC model overcame the problem of skewed performance towards identifying substrates compared with non-substrates, which is a common problem in the literature. The models were thoroughly validated by using external testing, applicability domain and activity cliffs characterization. In conclusion, a multi-label classification approach is an appropriate alternative for the prediction of ABC efflux.

Substrate-specific effects of pirinixic acid derivatives on ABCB1-mediated drug transport.

Pirinixic acid derivatives, a new class of drug candidates for a range of diseases, interfere with targets including PPARα, PPARγ, 5-lipoxygenase (5-LO), and microsomal prostaglandin and E2 synthase-1 (mPGES1). Since 5-LO, mPGES1, PPARα, and PPARγ represent potential anti-cancer drug targets, we here investigated the effects of 39 pirinixic acid derivatives on prostate cancer (PC-3) and neuroblastoma (UKF-NB-3) cell viability and, subsequently, the effects of selected compounds on drug-resistant neuroblastoma cells. Few compounds affected cancer cell viability in low micromolar concentrations but there was no correlation between the anti-cancer effects and the effects on 5-LO, mPGES1, PPARα, or PPARγ. Most strikingly, pirinixic acid derivatives interfered with drug transport by the ATP-binding cassette (ABC) transporter ABCB1 in a drug-specific fashion. LP117, the compound that exerted the strongest effect on ABCB1, interfered in the investigated concentrations of up to 2µM with the ABCB1-mediated transport of vincristine, vinorelbine, actinomycin D, paclitaxel, and calcein-AM but not of doxorubicin, rhodamine 123, or JC-1. In silico docking studies identified differences in the interaction profiles of the investigated ABCB1 substrates with the known ABCB1 binding sites that may explain the substrate-specific effects of LP117. Thus, pirinixic acid derivatives may offer potential as drug-specific modulators of ABCB1-mediated drug transport.

Prediction of drug distribution in rat and humans using an artificial neural networks ensemble and a PBPK model.

PURPOSE: To develop a QSAR model, based on calculated molecular descriptors and an Artificial Neural Networks Ensemble (ANNE), for the estimation of rat tissue-to-blood partition coefficients (Kt:b), as well as the assessment of the applicability domain of the model and its utility in predicting the drug distribution in humans. METHODS: A total of 1460 individual Kt:b values (75% train and 25% validation), obtained in 13 different rat tissues were collected in the literature. A correlation between simple molecular descriptors for lipophilicity, ionization, size and hydrogen bonding capacity and Kt:b data was attempted by using an ANNE. RESULTS: Similar statistics were observed between the train and validation group of data with correlations, between the observed values and the predicted average ANNE values, of 0.909 and 0.896, respectively. A degradation of the correlations was observed for predicted values with high uncertainty, as judged by the standard deviations of the ANNE outputs. This was further observed when using the ANNE Kt:b values in a Physiologically based pharmacokinetic (PBPK) model for predicting the Human Volume of distribution of another 532 drugs. CONCLUSIONS: This model (available as a MS Excel® workbook in the Supporting material of this article) may be a valuable tool for prediction and simulation in early drug development, allowing the in silico estimation of rat Kt:b values for PBPK purposes and also indicating its applicability domain.

Tissue-to-blood distribution coefficients in the rat: utility for estimation of the volume of distribution in man.

A compilation of rat tissue-to-blood partition coefficient data obtained both in vitro and in vivo in thirteen different tissues for a total of 309 different drugs is presented. An evaluation of the relationship between several fundamental physicochemical molecular descriptors and these distribution parameters was made. In addition, the ability to predict the Human Volume of distribution by regression analysis and by a Physiologically-based approach was also tested. Results have shown different trends between the drug classes and tissues, consistent with earlier described relationships between physicochemical properties and pharmacokinetic behavior. It was also possible to conclude for the acceptable ability to predict the volume of distribution in Humans by both regression and mechanistic approaches, which suggests that this type of data represents a convenient tool to describe the drug distribution on a new drug development context. These observations and analyses, along with the large database of rat tissue distribution data, should enable future efforts aimed toward developing a full in silico quantitative structure-pharmacokinetic relationships and improving our understanding of the correlations between fundamental chemical characteristics and drug distribution.

The antitumor effects of an arsthinol-cyclodextrin complex in a heterotopic mouse model of glioma.

In this paper, we examined arsthinol-cyclodextrin complexes, which display an anticancer activity. The association constants were 17,502±522 M(-1) for hydroxypropyl-ß-cyclodextrin and 12,038±10,168 M(-1) for randomized methylated ß-cyclodextrin. (1)H NMR experiments in solution also confirmed the formation of these complexes and demonstrated an insertion of the arsthinol (STB) with its dithiarsolane extremity into the wide rim of the hydroxypropyl-ß-cyclodextrin cavity. Complexed arsthinol was more effective than arsenic trioxide (As2O3) and melarsoprol on the U87 MG cell line. Importantly, in the in vivo study, we observed significant antitumor activity against heterotopic xenografts after i.p. administration and did not see any signs of toxicity. This remains to be verified using an orthotopic model.