In silico ADME/Tox: the state of the art

In Vitro Human Gastrointestinal Tract Simulation Systems: A Panoramic Review

Simulated human gastrointestinal (GI) tract systems are important for their applications in the fields of probiotics, nutrition and health. To date, various in vitro gut systems have been available to study GI tract dynamics and its association with health. In contrast to in vivo investigations, which are constrained by ethical considerations, in vitro models have several benefits despite the challenges involved in mimicking the GI environment. These in vitro models can be used for a range of research, from simple to dynamic, with one compartment to several compartments. In this review, we present a panoramic development of in vitro GI models for the first time through an evolutionary timeline. We tried to provide insight on designing an in vitro gut model, especially for novices. Latest developments and scope for improvement based on the limitations of the existing models were highlighted. In conclusion, designing an in vitro GI model suitable for a particular application is a multifaceted task. The bio-mimicking of the GI tract specific to geometrical, anatomical and mechanical features remains a challenge for the development of effective in vitro GI models. Advances in computer technology, artificial intelligence and nanotechnology are going to be revolutionary for further development. Besides this, in silico high-throughput technologies and miniaturisation are key players in the success of making in vitro modelling cost-effective and reducing the burden of in vivo studies.

In silico approaches for the identification of novel ULK1 inhibitors: pharmacophore model, molecular docking and molecular dynamics simulations

The serine/threonine kinase unc-51-like autophagy activating kinase 1 (ULK1) has been regarded as an attractive target for tumor therapy. In this study, in silico approaches, such as the pharmacophore-based virtual screening strategy, molecular docking and molecular dynamics (MD) simulations, were applied to develop novel potential ULK1 inhibitors. The pharmacophore models based on known aminopyrimidine ULK1 inhibitors were constructed to screen the dataset of 1.68 million compounds, which were obtained via screening the 2.30 million compounds in ChEMBL database by Lipinski's rule of five. Seven novel compounds and 1 known ULK1 inhibitor stand out for the strong virtual biological activity by molecular docking, cluster analysis, Molecular Mechanics/Generalized Born Surface Area (MM/GBSA) calculation and Absorption Distribution Metabolism Excretion Toxicity (ADMET) prediction. Their results of MD included principal component analysis (PCA) and Free Energy Landscapes surface (FELs) indicated that the protein-ligand complex was stable in simulated trajectories of 100 ns. The binding free energy (BFE) calculations showed that a total of 6 novel compounds (CL130, CL834, CL961, CL966, CL163 and CL329) with the stable binding state and stronger BFE (-61.17 to -37.01 kcal/mol) than that of original ligand 3RF (-36.66 kcal/mol). With reference to the ULK1 inhibition of 3RF (IC50 = 160 nM), it can be inferred that these compounds could be used as a new type of potential ULK1 inhibitors and be worthy of further investigation for tumor treatments.Communicated by Ramaswamy H. Sarma.

Broad-Spectrum Inhibitors against Class A, B, and C Type β-Lactamases to Block the Hydrolysis against Antibiotics: Kinetics and Structural Characterization

The emergence of antibiotic resistance has led to a global crisis for the physician to handle infection control issues. All antibiotics, including colistin, have lost efficiency against emerging drug-resistant bacterial strains due to the production of metallo-β-lactamases (MBLs) and serine-β-lactamases (SBLs). Therefore, it is of the utmost importance to design inhibitors against these enzymes to block the hydrolytic action against antibiotics being used. Although various novel β-lactamase inhibitors are being authorized or are under clinical studies, the coverage of their activity spectrum does not include MDR organisms expressing multiple classes of β-lactamases at a single time. This study reports three novel broad-spectrum inhibitors effective against both SBLs and MBLs. Virtual screening, molecular docking, molecular dynamics simulations, and an in silico pharmacokinetic study were performed to identify the lead molecules with broad-spectrum ability to inhibit the hydrolysis of β-lactam. The selected compounds were further assessed by in vitro cell assays (MIC, 50% inhibitory concentration [IC₅₀], kinetics, and fluorescence against class A, B, and C type β-lactamases) to confirm their efficacies. A 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) assay was performed to check the toxicity of screened lead molecules. All three selected inhibitors were found to reduce MIC and showed good affinity against all the SBLs and MBLs produced by class A, B, and C type β-lactamases. These nontoxic novel non-β-lactam broad-spectrum inhibitors bind to the active site residues of selected β-lactamases, which are crucial for β-lactam antibiotic hydrolysis. These inhibitors may be proposed as a future drug candidate in combination with antibiotics as a single formulation to control infection caused by resistant strains. Hence, this study plays a significant role in the cure of infections caused by antibiotic-resistant bacteria.

IMPORTANCE Several inhibitors for usage in conjunction with antibiotics have been developed. However, to date, there is no commercially available broad-spectrum β-lactamase inhibitor that targets both MBLs and SBLs. Here, we showed three novel broad-spectrum inhibitors with promising results through computational techniques and in vitro studies. These inhibitors are effective against both SBLs and MBLs and hence could be used as future drug candidates to treat infections caused by multidrug-resistant bacteria producing both types of enzymes (SBLs and MBLs).

Chemically synthesised flavone and coumarin based isoxazole derivatives as broad spectrum inhibitors of serine β-lactamases and metallo-β-lactamases: a computational, biophysical and biochemical study

The β-lactam antibiotics are the most effective medicines for treating bacterial infections. Resistance to them, particularly through the production of β-lactamases, which can hydrolyse all kinds of β-lactams, poses a threat to their continued use. The synthesised flavone and coumarin based isoxazole derivatives have the potential to be used as broad-spectrum inhibitors of the mechanistically different serine-(SBL) and metallo-β-lactamases (MBL). The synthesised compounds were discovered as potent β-lactamase inhibitors using molecular docking and in silico pharmacokinetic analysis. We studied the binding of chemically synthesised inhibitors to clinically significant β-lactamases of class A, B, and C using biophysical and biochemical approaches, and computational analyses. These molecules follow Lipinski's rule of five and have acceptable solubility, permeability, and oral bioavailability. These molecules were found to be non-toxic and non-carcinogenic. MIC results suggest that these molecules restore the antibiotic efficacy against class A, B, and C β-lactamases. Kinetics data showed that these molecules reduce the catalytic efficiency of clinically relevant class A, B, and C β-lactamases. Fluorescence study showed significant interaction between these flavone-/coumarin-based isoxazole derivatives and class A/B/ C β-lactamases. This study showed promising effect of these new generation compounds as broad spectrum β-lactamase inhibitors of both SBLs and MBLs.Communicated by Ramaswamy H. Sarma.

How fullerene derivatives (FDs) act on therapeutically important targets associated with diabetic diseases

•

Five proteins related to diabetic disease were selected from Protein Data Bank.

•

Binding scores were calculated for five proteins with 169 fullerene derivatives.

•

Correlation between drug-like descriptors and binding scores activity was examined.

•

The contribution of descriptors to protein-ligand binding was demonstrated.

•

The QSARs models for prediction of binding scores activity were built.

Fullerene derivatives (FDs) belong to a relatively new family of nano-sized organic compounds. They are widely applied in materials science, pharmaceutical industry, and (bio) medicine. This research focused on the study of FDs in terms of their potential inhibitory effect on therapeutic targets associated with diabetic disease, as well as analysis of protein–ligand binding in order to identify the key binding characteristics of FDs.

Therapeutic drug compounds when entering the biological system usually inevitably encounter and interact with a vast variety of biomolecules that are responsible for many different functions in organisms. Protein biomolecules are the most important functional components and used in this study as target structures. The structures of proteins [(PDB ID: 1BMQ, 1FM6, 1GPB, 1H5U, 1US0)] belonging to the class of anti-diabetes targets were obtained from the Protein Data Bank (PDB). Protein binding activity data (binding scores) were calculated for the dataset of 169 FDs related to these five proteins. Subsequently, the resulting data were analyzed using various machine learning and cheminformatics methods, including artificial neural network algorithms for variable selection and property prediction.

The Quantitative Structure-Activity Relationship (QSAR) models for prediction of binding scores activity were built up according to five Organization for Economic Co-operation and Development (OECD) principles.

All the data obtained can provide important information for further potential use of FDs with different functional groups as promising medical antidiabetic agents. Binding scores activity can be used for ranking of FDs in terms of their inhibitory activity (pharmacological properties) and potential toxicity.

In silico virtual screening of potent inhibitor to hamper the interaction between HIV-1 integrase and LEDGF/p75 interaction using E-pharmacophore modeling, molecular docking, and dynamics simulations

The rapid increase of HIV-1 infection throughout the globe has a high demand for a superior drug with lesser side effects. LEDGF/p75, the human Lens Epithelium-Derived Growth Factor is identified as a promising cellular cofactor with integrase in facilitating the viral replication in an early stage by acting as a tethering factor in the pre-integration to the chromatin. Therefore, the present study was designed to identify a potent inhibitor by applying an E-pharmacophore based virtual screening, molecular docking, and dynamics simulation approaches. Finally, ZINC22077550 and ZINC32124441 were best identified potent molecules with the efficient binding affinity, strong hydrogen bonding, and acceptable pharmacological properties to hamper the interaction between integrase and LEDGF/p75. Further, the DFT and MDS studies were also analyzed, and shown a favorable energetic state and dynamic stability then reference compound. In conclusion, we suggest that these findings could be novel therapeutics in the future and may increase the lifespan of individuals suffering from viral infection.

SolTranNet-A Machine Learning Tool for Fast Aqueous Solubility Prediction

While accurate prediction of aqueous solubility remains a challenge in drug discovery, machine learning (ML) approaches have become increasingly popular for this task. For instance, in the Second Challenge to Predict Aqueous Solubility (SC2), all groups utilized machine learning methods in their submissions. We present SolTranNet, a molecule attention transformer to predict aqueous solubility from a molecule's SMILES representation. Atypically, we demonstrate that larger models perform worse at this task, with SolTranNet's final architecture having 3,393 parameters while outperforming linear ML approaches. SolTranNet has a 3-fold scaffold split cross-validation root-mean-square error (RMSE) of 1.459 on AqSolDB and an RMSE of 1.711 on a withheld test set. We also demonstrate that, when used as a classifier to filter out insoluble compounds, SolTranNet achieves a sensitivity of 94.8% on the SC2 data set and is competitive with the other methods submitted to the competition. SolTranNet is distributed via pip, and its source code is available at https://github.com/gnina/SolTranNet.

Natural Chain-Breaking Antioxidants and Their Synthetic Analogs as Modulators of Oxidative Stress

Oxidative stress is associated with the increased production of reactive oxygen species or with a significant decrease in the effectiveness of antioxidant enzymes and nonenzymatic defense. The penetration of oxygen and free radicals in the hydrophobic interior of biological membranes initiates radical disintegration of the hydrocarbon “tails” of the lipids. This process is known as “lipid peroxidation”, and the accumulation of the oxidation products as peroxides and the aldehydes and acids derived from them are often used as a measure of oxidative stress levels. In total, 40 phenolic antioxidants were selected for a comparative study and analysis of their chain-breaking antioxidant activity, and thus as modulators of oxidative stress. This included natural and natural-like ortho-methoxy and ortho-hydroxy phenols, nine of them newly synthesized. Applied experimental and theoretical methods (bulk lipid autoxidation, chemiluminescence, in silico methods such as density functional theory (DFT) and quantitative structure–activity relationship ((Q)SAR) modeling) were used to clarify their structure–activity relationship. Kinetics of non-inhibited and inhibited lipid oxidation in close connection with inhibitor transformation under oxidative stress is considered. Special attention has been paid to chemical reactions resulting in the initiation of free radicals, a key stage of oxidative stress. Effects of substituents in the side chains and in the phenolic ring of hydroxylated phenols and biphenols, and the concentration were discussed.

Investigating into the molecular interactions of flavonoids targeting NS2B-NS3 protease from ZIKA virus through in-silico approaches

Zika virus (ZIKV), belongs to the flavivirus genus and Flaviviridae family that associated with serious diseased conditions like microcephaly and other neurological disorders (Guillan-Barré syndrome). As there is no vaccine or therapies available against ZIKV to date. Hence, it is an unmet need to find potential drug candidates and target sites against Zika virus infection. NS2B-NS3 protease making an attractive target for therapeutic intervention in ZIKV infections because of its critical role in hydrolysis of a single polyprotein encoded by Zika virus. Recently, there are some experimental evidence about the flavonoids as Zika virus NS2B-NS3 protease inhibitors. However, molecular interaction between protease complex and inhibitors at atomic levels has not been explored. Here, we have taken the experimentally validated thirty-eight flavonoids inhibitors against NS2B-NS3 protease to examine the molecular interaction using molecular docking and molecular dynamics simulations. We found out few flavonoids such as EGCG and its two derivatives, isoquercetin, rutin and sanggenon O showing interaction with catalytic triad (His51, Asp75, and Ser135) of the active site of NS2B-NS3 protease and found to be stable throughout the simulation. Therefore it is evident that interaction with the catalytic triad playing a vital role in the inhibition of the enzyme activity as a result inhibition of the virus propagation. However these compounds can be explored further for understanding the mechanism of action of these compounds targeting NS2B-NS3 protease for inhibition of Zika virus.

Adversarial Threshold Neural Computer for Molecular de Novo Design

In this article, we propose the deep neural network Adversarial Threshold Neural Computer (ATNC). The ATNC model is intended for the de novo design of novel small-molecule organic structures. The model is based on generative adversarial network architecture and reinforcement learning. ATNC uses a Differentiable Neural Computer as a generator and has a new specific block, called adversarial threshold (AT). AT acts as a filter between the agent (generator) and the environment (discriminator + objective reward functions). Furthermore, to generate more diverse molecules we introduce a new objective reward function named Internal Diversity Clustering (IDC). In this work, ATNC is tested and compared with the ORGANIC model. Both models were trained on the SMILES string representation of the molecules, using four objective functions (internal similarity, Muegge druglikeness filter, presence or absence of sp³-rich fragments, and IDC). The SMILES representations of 15K druglike molecules from the ChemDiv collection were used as a training data set. For the different functions, ATNC outperforms ORGANIC. Combined with the IDC, ATNC generates 72% of valid and 77% of unique SMILES strings, while ORGANIC generates only 7% of valid and 86% of unique SMILES strings. For each set of molecules generated by ATNC and ORGANIC, we analyzed distributions of four molecular descriptors (number of atoms, molecular weight, logP, and tpsa) and calculated five chemical statistical features (internal diversity, number of unique heterocycles, number of clusters, number of singletons, and number of compounds that have not been passed through medicinal chemistry filters). Analysis of key molecular descriptors and chemical statistical features demonstrated that the molecules generated by ATNC elicited better druglikeness properties. We also performed in vitro validation of the molecules generated by ATNC; results indicated that ATNC is an effective method for producing hit compounds.

Designing of inhibitors against CTX-M-15 type β-lactamase: potential drug candidate against β-lactamases-producing multi-drug-resistant bacteria

CTX-M-15 are the most prevalent types of β-lactamases that hydrolyze almost all antibiotics of β-lactam group lead to multiple-antibiotic resistance in bacteria. Three β-lactam inhibitors are available for use in combination with different antibiotics of cephalosporine group against the CTX-M-15-producing strains. Therefore, strategies to identify novel anti β-lactamase agents with specific mechanisms of action are the need of an hour. In this study, we screened three novel non-β-lactam inhibitors against CTX-M-15 by multi-step virtual screening approach. The potential for virtually screened drugs was estimated through in vitro cell assays. Hence, we proposed a study to understand the binding mode of CTX-M-15 with inhibitors using bioinformatics and experimental approach. We calculated the dissociation constants (K_d), association constant (K_a), stoichiometry (n) and binding energies (ΔG) of compounds with the respective targets. Molecular dynamic simulation carried out for 25 ns, revealed that these complexes were found stable throughout the simulation with relative RMSD in acceptable range. Moreover, microbiological and kinetic studies further confirmed high efficacies of these inhibitors by reducing the minimum inhibitory concentration (MIC) and catalysis of antibiotics by β-lactamases in the presence of inhibitors. Therefore, we conclude that these potential inhibitors may be used as a lead molecule for future drug candidates against β-lactamases-producing bacteria.

ADME/Tox Properties and Biochemical Interactions of Silybin Congeners: In silico Study

Silymarin, the active constituent of Silybum marianum (milk thistle), and its main component, silybin, are products with well-known hepatoprotective, cytoprotective, antioxidant, and chemopreventative properties. Despite substantial in vitro and in vivo investigations of these flavonolignans, their mechanisms of action and potential toxic effects are not fully defined. In this study we explored important ADME/Tox properties and biochemical interactions of selected flavonolignans using in silico methods. A quantitative structure–activity relationship (QSAR) model based on data from a parallel artificial membrane permeability assay (PAMPA) was used to estimate bioavailability after oral administration. Toxic effects and metabolic transformations were predicted using the knowledge-based expert systems Derek Nexus and Meteor Nexus (Lhasa Ltd). Potential estrogenic activity of the studied silybin congeners was outlined. To address further the stereospecificity of this effect the stereoisomeric forms of silybin were docked into the ligand-binding domain of the human estrogen receptor alpha (ERα) (MOE software, CCG). According to our results both stereoisomers can be accommodated into the ERα active site, but different poses and interactions were observed for silybin A and silybin B.

Open Source Bayesian Models. 1. Application to ADME/Tox and Drug Discovery Datasets

On the order of hundreds of absorption, distribution, metabolism, excretion, and toxicity (ADME/Tox) models have been described in the literature in the past decade which are more often than not inaccessible to anyone but their authors. Public accessibility is also an issue with computational models for bioactivity, and the ability to share such models still remains a major challenge limiting drug discovery. We describe the creation of a reference implementation of a Bayesian model-building software module, which we have released as an open source component that is now included in the Chemistry Development Kit (CDK) project, as well as implemented in the CDD Vault and in several mobile apps. We use this implementation to build an array of Bayesian models for ADME/Tox, in vitro and in vivo bioactivity, and other physicochemical properties. We show that these models possess cross-validation receiver operator curve values comparable to those generated previously in prior publications using alternative tools. We have now described how the implementation of Bayesian models with FCFP6 descriptors generated in the CDD Vault enables the rapid production of robust machine learning models from public data or the user’s own datasets. The current study sets the stage for generating models in proprietary software (such as CDD) and exporting these models in a format that could be run in open source software using CDK components. This work also demonstrates that we can enable biocomputation across distributed private or public datasets to enhance drug discovery.

Computational approaches to analyse and predict small molecule transport and distribution at cellular and subcellular levels

Quantitative structure-activity relationship (QSAR) studies and mechanistic mathematical modeling approaches have been independently employed for analysing and predicting the transport and distribution of small molecule chemical agents in living organisms. Both of these computational approaches have been useful for interpreting experiments measuring the transport properties of small molecule chemical agents, in vitro and in vivo. Nevertheless, mechanistic cell-based pharmacokinetic models have been especially useful to guide the design of experiments probing the molecular pathways underlying small molecule transport phenomena. Unlike QSAR models, mechanistic models can be integrated from microscopic to macroscopic levels, to analyse the spatiotemporal dynamics of small molecule chemical agents from intracellular organelles to whole organs, well beyond the experiments and training data sets upon which the models are based. Based on differential equations, mechanistic models can also be integrated with other differential equations-based systems biology models of biochemical networks or signaling pathways. Although the origin and evolution of mathematical modeling approaches aimed at predicting drug transport and distribution has occurred independently from systems biology, we propose that the incorporation of mechanistic cell-based computational models of drug transport and distribution into a systems biology modeling framework is a logical next step for the advancement of systems pharmacology research.

The role of tumor tissue architecture in treatment penetration and efficacy: an integrative study

Despite the great progress that has been made in understanding cancer biology and the potential molecular targets for its treatment, the majority of drugs fail in the clinical trials. This may be attributed (at least in part) to the complexity of interstitial drug transport in the patient’s body, which is hard to test experimentally. Similarly, recent advances in molecular imaging have led to the development of targeted biomarkers that can predict pharmacological responses to therapeutic interventions. However, both the drug and biomarker molecules need to access the tumor tissue and be taken up into individual cells in concentrations sufficient to exert the desired effect. To investigate the process of drug penetration at the mesoscopic level we developed a computational model of interstitial transport that incorporates the biophysical properties of the tumor tissue, including its architecture and interstitial fluid flow, as well as the properties of the agents. This model is based on the method of regularized Stokeslets to describe the fluid flow coupled with discrete diffusion-advection-reaction equations to model the dynamics of the drugs. Our results show that the tissue cellular porosity and density influence the depth of penetration in a non-linear way, with sparsely packed tissues being traveled through more slowly than the denser tissues. We demonstrate that irregularities in the cell spatial configurations result in the formation of interstitial corridors that are followed by agents leading to the emergence of tissue zones with less exposure to the drugs. We describe how the model can be integrated with in vivo experiments to test the extravasation and penetration of the targeted biomarkers through the tumor tissue. A better understanding of tissue- or compound-specific factors that limit the penetration through the tumors is important for non-invasive diagnoses, chemotherapy, the monitoring of treatment responses, and the detection of tumor recurrence.

Testing physical models of passive membrane permeation

The biophysical basis of passive membrane permeability is well-understood, but most methods for predicting membrane permeability in the context of drug design are based on statistical relationships that indirectly capture the key physical aspects. Here, we investigate molecular mechanics-based models of passive membrane permeability and evaluate their performance against different types of experimental data, including parallel artificial membrane permeability assays (PAMPA), cell-based assays, in vivo measurements, and other in silico predictions. The experimental data sets we use in these tests are diverse, including peptidomimetics, congeneric series, and diverse FDA approved drugs. The physical models are not specifically trained for any of these data sets; rather, input parameters are based on standard molecular mechanics force fields, such as partial charges, and an implicit solvent model. A systematic approach is taken to analyze the contribution from each component in the physics-based permeability model. A primary factor in determining rates of passive membrane permeation is the conformation-dependent free energy of desolvating the molecule, and this measure alone provides good agreement with experimental permeability measurements in many cases. Other factors that improve agreement with experimental data include deionization and estimates of entropy losses of the ligand and the membrane, which lead to size-dependence of the permeation rate.

Use of mathematical models to understand anticancer drug delivery and its effect on solid tumors

The transport of anticancer drugs and their effect on tumor cells involve a number of physical and biochemical processes. Mathematical modeling provides a tool to help us understand the interaction of these complex processes, thereby contributing to the improvement and optimization of drug delivery. This article starts with a discussion of the biological and physiological properties of tumors, which are often found as barriers to anticancer drug transport and effect. A broad spectrum of mathematical models is reviewed to give an overview of the current state of modeling approaches and different categories of models are outlined. These include pharmacokinetic and transport-based models for the prediction of temporal and temporal-spatial profiles of antidrug concentrations, as well as empirical or deterministic models to describe the effect of drug. We conclude that the systematic elucidation and integration of cellular signal transduction with the biophysical aspects of drug transport will lead to a better understanding of the entire drug-delivery process.

Synthesis, in silico metabolic and toxicity prediction of some novel imidazolinones derivatives as potent anticonvulsant agents

A series of 1,2,4-trisubstituted 5-imidazolinone derivatives were synthesized by Erlenmeyer condensation of benzoylglycine (hippuric acid) with different aldehydes in the presence of sodium acetate and acetic anhydride. The derivatives of the compounds were prepared by condensation of some known sulpha drugs with 5-oxazolone derivatives. The anticonvulsant activity of the compounds was determined by the protection of pentylenetetrazole-induced convulsions that was ranged from 10 to 60%. The compounds with p-OCH₃, p-OH and o-Cl substitutions in the phenyl ring on 4(th) position of the imidazolinone ring exhibited good anticonvulsant activity. In silico metabolic and toxicity studies showed that all the compounds in the series are not likely to exhibit toxicity except the compounds IIIa, IIIb, VIa and VIb, that is predicted to show 29% mutagenicity and 53% irritation in comparison to the other compounds. The predicted lethal effect and hERG toxicity of the compounds showed that IIa, IVa, Va and Vb might be toxic at higher concentrations. The results successfully establish the synthesized imidazolinone derivatives as novel compounds with anticonvulsant properties, low predicted cardiotoxicity and lethal effects thus can be promising leads for further development as novel anticonvulsants.

QSARs for PBPK modelling of environmental contaminants

Physiologically-based pharmacokinetic (PBPK) models are increasingly finding use in risk assessment applications of data-rich compounds. However, it is a challenge to determine the chemical-specific parameters for these models, particularly in time- and resource-limiting situations. In this regard, SARs, QSARs and QPPRs are potentially useful for computing the chemical-specific input parameters of PBPK models. Based on the frequency of occurrence of molecular fragments (CH(3), CH(2), CH, C, C=C, H, benzene ring and H in benzene ring structure) and exposure conditions, the available QSAR-PBPK models facilitate the simulation of tissue and blood concentrations for some inhaled volatile organic chemicals. The application domain of existing QSARs for developing PBPK models is limited, due to lack of relevant data for diverse chemicals and mechanisms. Even though this approach is conceptually applicable to non-volatile and high molecular weight organics as well, it is more challenging to predict the other PBPK model parameters required for modelling the kinetics of these chemicals (particularly tissue diffusion coefficients, association constants for binding and oral absorption rates). As the level of our understanding of the mechanistic basis of toxicokinetic processes improves, QSARs to provide a priori predictions of key chemical-specific PBPK parameters can be developed to expedite the internal dose-based health risk assessments in data-poor situations.

Automated selection of compounds with physicochemical properties to maximize bioavailability and druglikeness

Adequate bioavailability is one of the essential properties for an orally administered drug. Lipinski and others have formulated simplified rules in which compounds that satisfy selected physiochemical properties, for example, molecular weight (MW) ≤ 500 or the logarithm of the octanol-water partition coefficient, log P(o/w) < 5, are anticipated to likely have pharmacokinetic properties appropriate for oral administration. However, these schemes do not simultaneously consider the combination of the physiochemical properties, complicating their application in a more automated fashion. To overcome this, we present a novel method to select compounds with a combination of physicochemical properties that maximize bioavailability and druglikeness based on compounds in the World Drug Index database. In the study four properties, MW, log P(o/w), number of hydrogen bond donors, and number of hydrogen acceptors, were combined into a 4-dimensional (4D) histogram, from which a scoring function was defined on the basis of a 4D dependent multivariate Gaussian model. The resulting equation allows for assigning compounds a bioavailability score, termed 4D-BA, such that chemicals with higher 4D-BA scores are more likely to have oral druglike characteristics. The descriptor is validated by applying the function to drugs previously categorized in the Biopharmaceutics Classification System, and examples of application of the descriptor are given in the context of previously published studies targeting heme oxygenase and SHP2 phosphatase. The approach is anticipated to be useful in early lead identification studies in combination with clustering methods to maximize chemical and structural diversity when selecting compounds for biological assays from large database screens. It may also be applied to prioritize synthetically feasible chemical modifications during lead compound optimization.

Prediction of human pharmacokinetics—evaluation of methods for prediction of hepatic metabolic clearance

Methods for prediction of hepatic clearance (CLH) in man have been evaluated. A physiologically-based in-vitro to in-vivo (PB-IVIV) method with human unbound fraction in blood (fu,bl) and hepatocyte intrinsic clearance (CLint)-data has a good rationale and appears to give the best predictions (maximum ∼2-fold errors; &lt; 25% errors for half of CL-predictions; appropriate ranking). Inclusion of an empirical scaling factor is, however, needed, and reasons include the use of cryopreserved hepatocytes with low activity, and inappropriate CLint- and fu,bl-estimation methods. Thus, an improvement of this methodology is possible and required. Neglect of fu,bl or incorporation of incubation binding does not seem appropriate. When microsome CLint-data are used with this approach, the CLH is underpredicted by 5- to 9-fold on average, and a 106-fold underprediction (attrition potential) has been observed. The poor performance could probably be related to permeation, binding and low metabolic activity. Inclusion of scaling factors and neglect of fu,bl for basic and neutral compounds improve microsome predictions. The performance is, however, still not satisfactory. Allometry incorrectly assumes that the determinants for CLH relate to body weight and overpredicts human liver blood flow rate. Consequently, allometric methods have poor predictability. Simple allometry has an average overprediction potential, &gt; 2-fold errors for ∼1/3 of predictions, and 140-fold underprediction to 5800-fold overprediction (potential safety risk) range. In-silico methodologies are available, but these need further development. Acceptable prediction errors for compounds with low and high CLH should be ∼50 and ∼10%, respectively. In conclusion, it is recommended that PB-IVIV with human hepatocyte CLint and fu,bl is applied and improved, limits for acceptable errors are decreased, and that animal CLH-studies and allometry are avoided.

Modeling free energies of solvation in olive oil

Olive oil partition coefficients are useful for modeling the bioavailability of drug-like compounds. We have recently developed an accurate solvation model called SM8 for aqueous and organic solvents (Marenich, A. V.; Olson, R. M.; Kelly, C. P.; Cramer, C. J.; Truhlar, D. G. J. Chem. Theory Comput. 2007, 3, 2011) and a temperature-dependent solvation model called SM8T for aqueous solution (Chamberlin, A. C.; Cramer, C. J.; Truhlar, D. G. J. Phys. Chem. B 2008, 112, 3024). Here we describe an extension of SM8T to predict air-olive oil and water-olive oil partitioning for drug-like solutes as functions of temperature. We also describe the database of experimental partition coefficients used to parametrize the model; this database includes 371 entries for 304 compounds spanning the 291-310 K temperature range.

A modified uncorrelated linear discriminant analysis model coupled with recursive feature elimination for the prediction of bioactivity

To meet the requirements of providing accurate, robust, and interpretable prediction of bioactivity, a modified uncorrelated linear discriminant analysis (M-ULDA) model was developed. In addition, a feature selection method called recursive feature elimination (RFE), originally used for support vector machine (SVM), was introduced and modified to fit the scheme of ULDA. From the evaluation of six pharmaceutical datasets, the M-UDLA coupled with RFE showed better or comparable classification accuracy with respect to other well-studied methods such as SVM and decision trees. The RFE used for ULDA has the advantage of increasing the computational speed and provides useful insights into biochemical mechanisms related to pharmaceutical activity by significantly reducing the number of variables used for the final model.

Parameters for Carbamate Pesticide QSAR and PBPK/PD Models for Human Risk Assessment

Our interest in providing parameters for the development of quantitative structure physiologically based pharmacokinetic/pharmacodynamic (QSPBPK/PD) models for assessing health risks to carbamates (USEPA 2005) comes from earlier work with organophosphorus (OP) insecticides (Knaak et al. 2004). Parameters specific to each carbamate are needed in the construction of PBPK/PD models along with their metabolic pathways. Parameters may be obtained by (1) development of QSAR models, (2) collecting pharmacokinetic data, and (3) determining pharmacokinetic parameters by fitting to experimental data. The biological parameters are given in Table 1 (Blancato et al. 2000). Table 1 Biological Parameters Required for Carbamate Pesticide Physiologically Based Pharmacokinetic/Pharmacodynamic (PBPK/PD) Models.(a).

An atomistic model of passive membrane permeability: application to a series of FDA approved drugs

We apply an atomistic model of passive membrane permeability to a series of weakly basic drugs. The computational model uses conformational sampling in combination with an all-atom force field and implicit solvent model to estimate relative passive membrane permeabilities. The model does not require the use of training data for rank-ordering compounds, and as such represents a different approach from the more commonly employed QSPR models. We compare the computational results to previously published experimental PAMPA and Caco-2 permeabilities.

Application of data mining approaches to drug delivery

Computational approaches play a key role in all areas of the pharmaceutical industry from data mining, experimental and clinical data capture to pharmacoeconomics and adverse events monitoring. They will likely continue to be indispensable assets along with a growing library of software applications. This is primarily due to the increasingly massive amount of biology, chemistry and clinical data, which is now entering the public domain mainly as a result of NIH and commercially funded projects. We are therefore in need of new methods for mining this mountain of data in order to enable new hypothesis generation. The computational approaches include, but are not limited to, database compilation, quantitative structure activity relationships (QSAR), pharmacophores, network visualization models, decision trees, machine learning algorithms and multidimensional data visualization software that could be used to improve drug delivery after mining public and/or proprietary data. We will discuss some areas of unmet needs in the area of data mining for drug delivery that can be addressed with new software tools or databases of relevance to future pharmaceutical projects.

Conformational Flexibility, Internal Hydrogen Bonding, and Passive Membrane Permeability: Successful in Silico Prediction of the Relative Permeabilities of Cyclic Peptides

We report an atomistic physical model for the passive membrane permeability of cyclic peptides. The computational modeling was performed in advance of the experiments and did not involve the use of "training data". The model explicitly treats the conformational flexibility of the peptides by extensive conformational sampling in low (membrane) and high (water) dielectric environments. The passive membrane permeabilities of 11 cyclic peptides were obtained experimentally using a parallel artificial membrane permeability assay (PAMPA) and showed a linear correlation with the computational results with R(2) = 0.96. In general, the results support the hypothesis, already well established in the literature, that the ability to form internal hydrogen bonds is critical for passive membrane permeability and can be the distinguishing factor among closely related compounds, such as those studied here. However, we have found that the number of internal hydrogen bonds that can form in the membrane and the solvent-exposed polar surface area correlate more poorly with PAMPA permeability than our model, which quantitatively estimates the solvation free energy losses upon moving from high-dielectric water to the low-dielectric interior of a membrane.

Ensemble of Linear Models for Predicting Drug Properties

We propose a new classification method for the prediction of drug properties, called random feature subset boosting for linear discriminant analysis (LDA). The main novelty of this method is the ability to overcome the problems with constructing ensembles of linear discriminant models based on generalized eigenvectors of covariance matrices. Such linear models are popular in building classification-based structure-activity relationships. The introduction of ensembles of LDA models allows for an analysis of more complex problems than by using single LDA, for example, those involving multiple mechanisms of action. Using four data sets, we show experimentally that the method is competitive with other recently studied chemoinformatic methods, including support vector machines and models based on decision trees. We present an easy scheme for interpreting the model despite its apparent sophistication. We also outline theoretical evidence as to why, contrary to the conventional AdaBoost ensemble algorithm, this method is able to increase the accuracy of LDA models.

Computational prediction of human drug metabolism

There is an urgent requirement within the pharmaceutical and biotechnology industries, regulatory authorities and academia to improve the success of molecules that are selected for clinical trials. Although absorption, distribution, metabolism, excretion and toxicity (ADME/Tox) properties are some of the many components that contribute to successful drug discovery and development, they represent factors for which we currently have in vitro and in vivo data that can be modelled computationally. Understanding the possible toxicity and the metabolic fate of xenobiotics in the human body is particularly important in early drug discovery. There is, therefore, a need for computational methodologies for uncovering the relationships between the structure and the biological activity of novel molecules. The convergence of numerous technologies, including high-throughput techniques, databases, ADME/Tox modelling and systems biology modelling, is leading to the foundation of systems-ADME/Tox. Results from experiments can be integrated with predictions to globally simulate and understand the likely complete effects of a molecule in humans. The development and early application of major components of MetaDrug (GeneGo, Inc.) software will be described, which includes rule-based metabolite prediction, quantitative structure-activity relationship models for major drug metabolising enzymes, and an extensive database of human protein-xenobiotic interactions. This represents a combined approach to predicting drug metabolism. MetaDrug can be readily used for visualising Phase I and II metabolic pathways, as well as interpreting high-throughput data derived from microarrays as networks of interacting objects. This will ultimately aid in hypothesis generation and the early triaging of molecules likely to have undesirable predicted properties or measured effects on key proteins and cellular functions.

In Silico Prediction of Membrane Permeability from Calculated Molecular Parameters

A data set consisting of 712 compounds was used for classification into two classes with respect to membrane permeation in a cell-based assay: (0) apparent permeability (P(app)) below 4 x 10(-6) cm/s and (1) P(app) on 4 x 10(-6) cm/s or higher. Nine molecular descriptors were calculated for each compound and Nearest-Neighbor classification was applied using five neighbors as optimized by full cross-validation. A model based on five descriptors, number of flex bonds, number of hydrogen bond acceptors and donors, and molecular and polar surface area, was selected by variable selection. In an external test set of 112 compounds, 104 compounds were classified and 8 compounds were judged as "unknown". Among the 104 compounds, 16 were misclassified corresponding to a misclassification rate of 15% and no compounds were falsely predicted in the nonpermeable class.

Rapid Quantum Mechanical Models for the Computational Estimation of C−H Bond Dissociation Energies as a Measure of Metabolic Stability

Several relatively inexpensive levels of theory are surveyed together with alternative algorithmic methods for the estimation of C-H bond dissociation energies (BDEs), such energies being useful for the prediction of metabolic stability in drug-like molecules. In particular, bond stretching potentials of several C-H bonds are computed using the AM1, PM3, HF/MIDI!, and B3LYP/MIDI! levels of electronic structure theory, and selected points are fit to Morse and parabolic potentials. BDEs computed by an AM1 fit to the Morse function show the smallest mean unsigned error in prediction (+/- 3-4 kcal/mol) over 32 diverse C-H bonds. An alternative method for correlating the AM1 parabolic force constant from a two-point unrelaxed potential provides only a slightly decreased accuracy and is computationally particularly inexpensive. Both methods should prove to be useful for the rapid in silico screening of drug-like molecules for metabolic stability to C-H bond oxidizing enzymes.

A Universal Molecular Descriptor System for Prediction of LogP, LogS, LogBB, and Absorption

Predictive models for octanol/water partition coefficient (logP), aqueous solubility (logS), blood-brain barrier (logBB), and human intestinal absorption (HIA) were built from a universal, generic molecular descriptor system, designed on the basis of atom type classification. The atom type classification tree was trained to optimize the logP predictions. With nine components, the final partial least-squares (PLS) model predicted logP of 10850 compounds in Starlist with a regression coefficient (r2) of 0.912, cross-validated r2 (q2) of 0.892, and root-mean-square error of estimation (RMSEE) of 0.50 log units. The PLS models for solubility (logS), blood-brain barrier (logBB), and a PLS-DA (discrimination analysis) model for HIA were established from the same atom type descriptors. The seven-component PLS model derived from a diverse set of 1478 organic compounds predicted a 21-compound test set designed by Yalkowsky with r2 = 0.88 and RMSEP (RMS error of prediction) = 0.64. A predictive r2 = 0.90 and RMSEE = 0.26 were achieved for logBB of a 57-compound "Abraham data set" with a three-component model. The first three components of a five-component PLS-DA model were sufficient to clearly separate the 169 drug molecules, collected by Abraham, into three classes, according to their percentage human intestinal absorption.

ADME-Tox in drug discovery: integration of experimental and computational technologies

Over the past ten years, in vitro experimental tools to characterize ADME-Tox profiles of compounds have been applied in early stages of the drug discovery process to increase the success rate of discovery programmes and to progress better candidates into drug development. Application of in silico ADME-Tox models has further enhanced discovery support, enabling virtual screening of compounds and thus, application of ADME-Tox at every stage of the discovery process. Ultimately, effective and efficient ADME-Tox support of discovery will depend on a complementary and synergistic use of experimental and in silico ADME-Tox.

Generation and validation of rapid computational filters for cyp2d6 and cyp3a4

CYP2D6 and CYP3A4 represent two particularly important members of the cytochrome p450 enzyme family due to their involvement in the metabolism of many commercially available drugs. Avoiding potent inhibitory interactions with both of these enzymes is highly desirable in early drug discovery, long before entering clinical trials. Computational prediction of this liability as early as possible is desired. Using a commercially available data set of over 1750 molecules to train computer models that were generated with commercially available software enabled predictions of inhibition for CYP2D6 and CYP3A4, which were compared with empirical data. The results suggest that using a recursive partitioning (tree) technique with augmented atom descriptors enables a statistically significant rank ordering of test-set molecules (Spearman's rho of 0.61 and 0.48 for CYP2D6 and CYP3A4, respectively), which represents an increased rate of identifying the best compounds when compared with the random rate. This approach represents a valuable computational filter in early drug discovery to identify compounds that may have p450 inhibition liabilities prior to molecule synthesis. Such computational filters offer a new approach in which lead optimization in silico can occur with virtual molecules simultaneously tested against multiple enzymes implicated in drug-drug interactions, with a resultant cost savings from a decreased level of molecule synthesis and in vitro screening.

A multiple in silico program approach for the prediction of mutagenicity from chemical structure

We have conducted an evaluation of three of the most widely used commercial toxicity prediction programs, Toxicity Prediction by Komputer Assisted Technology (TOPKAT), Deductive Estimation of Risk from Existing Knowledge (DEREK) for Windows (DfW) and CASETOX. The three programs were evaluated for their ability to predict Ames test mutagenicity using 520 proprietary drug candidate (Test set 1) and 94 commercial (Test set 2) compounds. The study demonstrates that these three commercially available programs are useful, with limitations in their ability to predict mutagenicity over a wide range of chemical space, i.e. global predictivity. Individually, each of the programs performed at an acceptable level for overall accuracy, i.e. the ability to predict the correct outcome. However, analysis of the predictions indicates that the overall accuracy figure is heavily weighted by the ability of the programs to correctly predict non-mutagens, whereas none of the programs individually performed well in the prediction of novel mutagenic structures, i.e. Ames positive compounds. The performance of these programs' in predicting Ames positive mutagens appeared to be independent of the chemical utility of the compound, i.e. industrial, agricultural or pharmaceutical. The combination of program predictions provided some improvement in overall accuracy, sensitivity and specificity.

Towards a new age of virtual ADME/TOX and multidimensional drug discovery

With the continual pressure to ensure follow-up molecules to billion dollar blockbuster drugs, there is a hurdle in profitability and growth for pharmaceutical companies in the next decades. With each success and failure we increasingly appreciate that a key to the success of synthesized molecules through the research and development process is the possession of drug-like properties. These properties include an adequate bioactivity as well as adequate solubility, an ability to cross critical membranes (intestinal and sometimes blood-brain barrier), reasonable metabolic stability and of course safety in humans. Dependent on the therapeutic area being investigated it might also be desirable to avoid certain enzymes or transporters to circumvent potential drug-drug interactions. It may also be important to limit the induction of these same proteins that can result in further toxicities. We have clearly moved the assessment of in vitro absorption, distribution, metabolism, excretion and toxicity (ADME/TOX) parameters much earlier in the discovery organization than a decade ago with the inclusion of higher throughput systems. We are also now faced with huge amounts of ADME/TOX data for each molecule that need interpretation and also provide a valuable resource for generating predictive computational models for future drug discovery. The present review aims to show what tools exist today for visualizing and modeling ADME/TOX data, what tools need to be developed, and how both the present and future tools are valuable for virtual filtering using ADME/TOX and bioactivity properties in parallel as a viable addition to present practices.

Selection criteria for drug-like compounds

The fast identification of quality lead compounds in the pharmaceutical industry through a combination of high throughput synthesis and screening has become more challenging in recent years. Although the number of available compounds for high throughput screening (HTS) has dramatically increased, large-scale random combinatorial libraries have contributed proportionally less to identify novel leads for drug discovery projects. Therefore, the concept of 'drug-likeness' of compound selections has become a focus in recent years. In parallel, the low success rate of converting lead compounds into drugs often due to unfavorable pharmacokinetic parameters has sparked a renewed interest in understanding more clearly what makes a compound drug-like. Various approaches have been devised to address the drug-likeness of molecules employing retrospective analyses of known drug collections as well as attempting to capture 'chemical wisdom' in algorithms. For example, simple property counting schemes, machine learning methods, regression models, and clustering methods have been employed to distinguish between drugs and non-drugs. Here we review computational techniques to address the drug-likeness of compound selections and offer an outlook for the further development of the field.

A brief history of novel drug discovery technologies

Laypersons, researchers and clinicians alike speak of the biotechnology revolution with excitement. Media coverage of new breakthroughs in medicine often have the public and the investment community on the edge of their seats, eager for the next blockbuster drug to cure everything from high cholesterol levels to cancer. In this perspective, we examine some of the more popularized and influential new technologies in drug discovery and assess their relative impact on the actual attainment of new therapeutics.

In vitro and pharmacophore insights into CYP3A enzymes

The cytochrome P450 3A (CYP3A) enzymes have a major role in the metabolism of drugs in humans. Their wide substrate specificity and induction by a vast array of structurally diverse compounds presents the possibility of metabolic drug-drug interactions. Understanding the enzymes themselves is crucial. Over the past decade, this has occurred mostly with in vitro studies, although more recent approaches incorporate computational models to predict CYP inhibition and substrate potential. The three-dimensional displacement, or pharmacophore, of chemical features in space that are derived from inhibition data have produced pharmacophores for CYP3A4, CYP3A5 and CYP3A7, and provide new insights into ligand binding for each enzyme.