The Promise of AI for DILI Prediction.

Drug-induced liver injury (DILI) is a common reason for the withdrawal of a drug from the market. Early assessment of DILI risk is an essential part of drug development, but it is rendered challenging prior to clinical trials by the complex factors that give rise to liver damage. Artificial intelligence (AI) approaches, particularly those building on machine learning, range from random forests to more recent techniques such as deep learning, and provide tools that can analyze chemical compounds and accurately predict some of their properties based purely on their structure. This article reviews existing AI approaches to predicting DILI and elaborates on the challenges that arise from the as yet limited availability of data. Future directions are discussed focusing on rich data modalities, such as 3D spheroids, and the slow but steady increase in drugs annotated with DILI risk labels.

In silico approaches in organ toxicity hazard assessment: current status and future needs in predicting liver toxicity.

Hepatotoxicity is one of the most frequently observed adverse effects resulting from exposure to a xenobiotic. For example, in pharmaceutical research and development it is one of the major reasons for drug withdrawals, clinical failures, and discontinuation of drug candidates. The development of faster and cheaper methods to assess hepatotoxicity that are both more sustainable and more informative is critically needed. The biological mechanisms and processes underpinning hepatotoxicity are summarized and experimental approaches to support the prediction of hepatotoxicity are described, including toxicokinetic considerations. The paper describes the increasingly important role of in silico approaches and highlights challenges to the adoption of these methods including the lack of a commonly agreed upon protocol for performing such an assessment and the need for in silico solutions that take dose into consideration. A proposed framework for the integration of in silico and experimental information is provided along with a case study describing how computational methods have been used to successfully respond to a regulatory question concerning non-genotoxic impurities in chemically synthesized pharmaceuticals.

In silico approaches in organ toxicity hazard assessment: Current status and future needs for predicting heart, kidney and lung toxicities.

The kidneys, heart and lungs are vital organ systems evaluated as part of acute or chronic toxicity assessments. New methodologies are being developed to predict these adverse effects based on in vitro and in silico approaches. This paper reviews the current state of the art in predicting these organ toxicities. It outlines the biological basis, processes and endpoints for kidney toxicity, pulmonary toxicity, respiratory irritation and sensitization as well as functional and structural cardiac toxicities. The review also covers current experimental approaches, including off-target panels from secondary pharmacology batteries. Current in silico approaches for prediction of these effects and mechanisms are described as well as obstacles to the use of in silico methods. Ultimately, a commonly accepted protocol for performing such assessment would be a valuable resource to expand the use of such approaches across different regulatory and industrial applications. However, a number of factors impede their widespread deployment including a lack of a comprehensive mechanistic understanding, limited in vitro testing approaches and limited in vivo databases suitable for modeling, a limited understanding of how to incorporate absorption, distribution, metabolism, and excretion (ADME) considerations into the overall process, a lack of in silico models designed to predict a safe dose and an accepted framework for organizing the key characteristics of these organ toxicants.

Design and Synthesis of a Pan-Janus Kinase Inhibitor Clinical Candidate (PF-06263276) Suitable for Inhaled and Topical Delivery for the Treatment of Inflammatory Diseases of the Lungs and Skin.

By use of a structure-based computational method for identification of structurally novel Janus kinase (JAK) inhibitors predicted to bind beyond the ATP binding site, a potent series of indazoles was identified as selective pan-JAK inhibitors with a type 1.5 binding mode. Optimization of the series for potency and increased duration of action commensurate with inhaled or topical delivery resulted in potent pan-JAK inhibitor 2 (PF-06263276), which was advanced into clinical studies.

Developing irreversible inhibitors of the protein kinase cysteinome.

Protein kinases are a large family of approximately 530 highly conserved enzymes that transfer a γ-phosphate group from ATP to a variety of amino acid residues, such as tyrosine, serine, and threonine, that serves as a ubiquitous mechanism for cellular signal transduction. The clinical success of a number of kinase-directed drugs and the frequent observation of disease causing mutations in protein kinases suggest that a large number of kinases may represent therapeutically relevant targets. To date, the majority of clinical and preclinical kinase inhibitors are ATP competitive, noncovalent inhibitors that achieve selectivity through recognition of unique features of particular protein kinases. Recently, there has been renewed interest in the development of irreversible inhibitors that form covalent bonds with cysteine or other nucleophilic residues in the ATP-binding pocket. Irreversible kinase inhibitors have a number of potential advantages including prolonged pharmacodynamics, suitability for rational design, high potency, and ability to validate pharmacological specificity through mutation of the reactive cysteine residue. Here, we review recent efforts to develop cysteine-targeted irreversible protein kinase inhibitors and discuss their modes of recognizing the ATP-binding pocket and their biological activity profiles. In addition, we provided an informatics assessment of the potential "kinase cysteinome" and discuss strategies for the efficient development of new covalent inhibitors.

Identification of type-II inhibitors using kinase structures.

Spleen tyrosine kinase is a non-receptor tyrosine kinase, overactivation of which is thought to contribute to autoimmune diseases as well as allergy and asthma. Protein kinases have a highly conserved ATP binding site, thus making challenging the design of selective small molecule inhibitors. It has been well documented that some protein kinases can be stabilized in their inactive conformations (Type-II inhibitors). Herein, we describe a protein structure/ligand-based approach to successfully identify ligands that bind to novel conformations of spleen tyrosine kinase. By utilizing kinase protein crystal structures both in the public domain (RCSB) and within Pfizer's protein crystal database, we report the discovery of the first spleen tyrosine kinase Type-II ligands. Compounds 1 and 3 were found to bind to the DFG-out conformation of spleen tyrosine kinase, while compound 2 binds to a DFG-in, C-Helix-out conformation. In this instance, the C-helix moved significantly to create a large hydrophobic pocket rarely seen in kinase protein crystal structures.

Discovery and synthesis of a new class of opioid ligand having a 3-azabicyclo[3.1.0]hexane core. An example of a 'magic methyl' giving a 35-fold improvement in binding.

In looking for a novel achiral µ opioid receptor antagonist for the treatment of pruritus, we designed and synthesised azabicyclo[3.1.0]hexane compounds as a new class of opioid ligand. During optimisation, an addition of a single methyl resulted in a 35-fold improvement in binding. An early example from the series had excellent µ opioid receptor antagonist antagonist activity and was very effective in an in vivo pruritus study.

Is it possible to increase hit rates in structure-based virtual screening by pharmacophore filtering? An investigation of the advantages and pitfalls of post-filtering.

We have investigated the influence of post-filtering virtual screening results, with pharmacophoric features generated from an X-ray structure, on enrichment rates. This was performed using three docking softwares, zdock+, Surflex and FRED, as virtual screening tools and pharmacophores generated in UNITY from co-crystallized complexes. Sets of known actives along with 9997 pharmaceutically relevant decoy compounds were docked against six chemically diverse protein targets namely CDK2, COX2, ERalpha, fXa, MMP3, and NA. To try to overcome the inherent limitations of the well-known docking problem, we generated multiple poses for each compound. The compounds were first ranked according to their scores alone and enrichment rates were calculated using only the top scoring pose of each compound. Subsequently, all poses for each compound were passed through the different pharmacophores generated from co-crystallized complexes and the enrichment factors were re-calculated based on the top-scoring passing pose of each compound. Post-filtering with a pharmacophore generated from only one X-ray complex was shown to increase enrichment rates in all investigated targets compared to docking alone. This indicates that this is a general method, which works for diverse targets and different docking softwares.

Evaluation of a diverse set of potential P1 carboxylic acid bioisosteres in hepatitis C virus NS3 protease inhibitors.

There is an urgent need for more efficient therapies for people infected with hepatitis C virus (HCV). HCV NS3 protease inhibitors have shown proof-of-concept in clinical trials, which make the virally encoded NS3 protease an attractive drug target. Product-based NS3 protease inhibitors comprising a P1 C-terminal carboxylic acid have shown to be effective and we were interested in finding alternatives to this crucial carboxylic acid group. Thus, a series of diverse P1 functional groups with different acidity and with possibilities to form a similar, or an even more powerful, hydrogen bond network as compared to the carboxylic acid were synthesized and incorporated into potential inhibitors of the NS3 protease. Biochemical evaluation of the inhibitors was performed in both enzyme and cell-based assays. Several non-acidic C-terminal groups, such as amides and hydrazides, were evaluated but failed to produce inhibitors more potent than the corresponding carboxylic acid inhibitor. The tetrazole moiety, although of similar acidity to a carboxylic acid, provided an inhibitor with mediocre potencies in both assays. However, the acyl cyanamide and the acyl sulfinamide groups rendered compounds with low nanomolar inhibitory potencies and were more potent than the corresponding carboxylic acid inhibitor in the enzymatic assay. Additionally, results from a pH-study suggest that the P(1) C-terminal of the inhibitors comprising a carboxylic acid, an acyl sulfonamide or an acyl cyanamide group binds in a similar mode in the active site of the NS3 protease.

Phenylglycine as a novel P2 scaffold in hepatitis C virus NS3 protease inhibitors.

Molecular modeling and inhibitory potencies of tetrapeptide protease inhibitors of HCV NS3 proposed phenylglycine as a new promising P2 residue. The results suggest that phenylglycine might be capable of interacting with the NS3 (protease-helicase/NTPase) in ways not possible for the common P2 proline-based inhibitors. Thus, a series of tripeptides, both linear and macrocyclic, based on p-hydroxy-phenylglycine in the P2 position were prepared and their inhibitory effect determined. When the p-hydroxy group was replaced by methoxy, isoquinolin-, or quinolinyloxy functions, inhibitors with improved potencies were obtained. The P2 phenylglycine-based inhibitors were further optimized by C-terminal extension to acyl sulfonamides and by P1-P3 cyclization, which gave products with inhibition constants in the nanomolar range ( approximately 75nM).

A new structural theme in C2-symmetric HIV-1 protease inhibitors: ortho-substituted P1/P1' side chains.

In this report, the rapid syntheses of 24 novel C2-symmetric HIV-1 protease inhibitors are described. Two ortho-iodobenzyloxy containing C-terminal duplicated inhibitors served as starting materials for microwave-enhanced palladium(0)-catalyzed carbon-carbon bond forming reactions (Suzuki, Sonogashira, Heck, and Negishi). Highly potent inhibitors equipped with ortho-functionalized P1/P1' side chains as the structural theme were identified. Computational efforts were applied to study the binding mode of this class of inhibitors and to establish structure-activity relationships. The overall orientation of the inhibitors in the active site was reproduced by docking which suggested three possible conformations of the P1/P1' groups of which two seem more plausible.

Exploration of acyl sulfonamides as carboxylic acid replacements in protease inhibitors of the hepatitis C virus full-length NS3.

The hepatitis C virus (HCV) NS3 protease has emerged as a promising anti-HCV drug target. Herein, we present an investigation of NS3 inhibitors comprising the acyl sulfonamide functionality. A series of tetra- and tripeptide based acyl sulfonamide inhibitors and their structure-activity relationships from both enzymatic and cell-based in vitro assays are presented. In summary, the acidity of the acyl sulfonamide functionality, the character of the P1 side chain, and the acyl sulfonamide substituent were found to be important for the inhibitory potencies.

Synthesis, biological evaluation, and modeling studies of inhibitors aimed at the malarial proteases plasmepsins I and II.

The increasing resistance of the malarial parasite to antimalarial drugs is a major contributor to the reemergence of the disease and increases the need for new drug targets. The two aspartic proteases, plasmepsins I and II, from Plasmodium falciparum have recently emerged as potential targets. In an effort to inhibit these hemoglobinases, a series of inhibitors encompassing a basic hydroxyethylamine transition state isostere as a central fragment were prepared. The synthesized compounds were varied in the P1' position and exhibited biological activities in the range of 31 to >2000 nM. To try to rationalize the results, molecular docking and 3D-QSAR analysis were used.

Identification of novel parasitic cysteine protease inhibitors using virtual screening. 1. The ChemBridge database.

Trypanosomiasis, leishmaniasis, and malaria are major parasitic diseases in developing countries. The existing chemotherapy of these diseases suffers from lack of safe and effective drugs and/or the presence of widespread drug resistance. Cysteine proteases are exciting novel targets for antiparasitic drug design. Virtual screening was performed in an attempt to identify novel druglike nonpeptide inhibitors of parasitic cysteine proteases. The ChemBridge database consisting of approximately 241 000 compounds was screened against homology models of falcipain-2 and falcipain-3 in three consecutive stages of docking. A total of 24 diverse inhibitors were identified from an initial group of 84, of which 12 compounds appeared to be dual inhibitors of falcipain-2 and falcipain-3. Four compounds showed inhibition of both the malarial cysteine proteases as well as Leishmania donovani cysteine protease.

Structure-based approach to falcipain-2 inhibitors: synthesis and biological evaluation of 1,6,7-trisubstituted dihydroisoquinolines and isoquinolines.

1,4,7-Trisubstituted isoquinolines were designed, synthesized and evaluated for their inhibition against Plasmodium falciparum cysteine protease falcipain-2. The 1-benzyloxyphenyl-dihydroisoquinoline and -isoquinoline derivatives were found to exhibit better activity against falcipain-2 than their corresponding 1-hydroxyphenyl or 1-methoxyphenyl analogues. The docking scores correlate with the IC(50) values of compounds and give a high coefficient correlation of 0.94.

Probing the structure of falcipain-3, a cysteine protease from Plasmodium falciparum: comparative protein modeling and docking studies.

Increasing resistance of malaria parasites to conventional antimalarial drugs is an important factor contributing to the persistence of the disease as a major health threat. The ongoing search for novel targets has resulted in identification and expression of several enzymes including cysteine proteases that are implicated in hemoglobin degradation. Falcipain-2 and falcipain-3 are considered to be the two principal cysteine proteases in this degradation, and hence, are potential drug targets. A homology model of falcipain-3 was built and validated by various structure/geometry verification tools as well as docking studies of known substrates. The correlation coefficient of 0.975 between interaction energies and K(m) values of these substrates provided additional support for the model. On comparison with the previously reported falcipain-2 homology model, the currently constructed falcipain-3 structure showed important differences between the S2 pockets that might explain the variations in the K(m) values of various substrates for these enzymes. Further, docking studies also provided insight into possible binding modes and interactions of ligands with falcipain-3. Results of the current study could be employed in de novo drug design leading to development of new antimalarial agents.

Interaction of artemisinin and its related compounds with hydroxypropyl-beta-cyclodextrin in solution state: experimental and molecular-modeling studies.

Hydroxypropyl-beta-cyclodextrin (HPBCD) was investigated as a possible solubilizer for a series of poorly water-soluble antimalarial drugs. The solubilities of artemisinin, artether, dihydroartemisinin, and 10-deoxoartemisinin in HPBCD solutions were studied. The phase-solubility profile of these drugs in HPBCD solutions, in the concentration range studied, can be classified as type A(L) or soluble 1:1 complexes. The solubilities of artemisinin, artether, dihydroartemisinin, and 10-deoxoartemisinin in 20% w/v solutions of HPBCD are 4.5, 1.3, 6.0, and 5.2 mg/mL, respectively. The stability constants of artemisinin, dihydroartemisinin, artether, and 10-deoxoartemisinin complexes with HPBCD are 475, 405, 327, and 146 M(-1), respectively. Three different docking methods, SYBYL DOCK, FlexiDock, and DOCK 4.0.1 were evaluated to further understand the complexation modes and applicability of the docking programs for the modeling of inclusion complexes. The results showed that DOCK 4.0.1 offers a better correlation in terms of orientation of molecules inside the cyclodextrin cavity and also in terms of docking scores.

Homology modeling of falcipain-2: validation, de novo ligand design and synthesis of novel inhibitors.

Increasing resistance of malaria parasites, in particular Plasmodium falciparum, demands a serious search for novel targets. Cysteine protease in P. falciparum, encoded by a previously unidentified gene falcipain 2, provides one such target to design chemotherapeutic agents for treatment of malaria. In fact, a few cysteine protease inhibitors have been shown to inhibit growth of cultured malarial parasites. In absence of a crystal structure for this enzyme, homology modeling proved to be a reasonable alternative to study binding requirements of the enzyme. A homology model for falcipain 2 was developed and validated by docking of known vinyl sulfone inhibitors. Further, based on the observations of these studies, novel isoquinoline inhibitors were designed and synthesized, which exhibited in vitro enzyme inhibition at micromolar concentrations.

Structure-activity relationships of the antimalarial agent artemisinin. 6. The development of predictive in vitro potency models using CoMFA and HQSAR methodologies.

Artemisinin (1) is a unique sesquiterpene peroxide occurring as a constituent of Artemisia annua L. Because of the effectiveness of Artemisinin in the treatment of drug-resistant Plasmodium falciparum and its rapid clearance of cerebral malaria, development of clinically useful semisynthetic drugs for severe and complicated malaria (artemether, artesunate) was prompt. However, recent reports of fatal neurotoxicity in animals with dihydroartemisinin derivatives such as artemether have spawned a renewed effort to develop nontoxic analogues of artemisinin. In our effort to develop more potent, less neurotoxic agents for the oral treatment of drug-resistant malaria, we utilized comparative molecular field analysis (CoMFA) and hologram QSAR (HQSAR), beginning with a series of 211 artemisinin analogues with known in vitro antimalarial activity. CoMFA models were based on two conformational hypotheses: (a) that the X-ray structure of artemisinin represents the bioactive shape of the molecule or (b) that the hemin-docked conformation is the bioactive form of the drug. In addition, we examined the effect of inclusion or exclusion of racemates in the partial least squares (pls) analysis. Databases derived from the original 211 were split into chiral (n = 157), achiral (n = 34), and mixed databases (n = 191) after leaving out a test set of 20 compounds. HQSAR and CoMFA models were compared in terms of their potential to generate robust QSAR models. The r(2) and q(2) (cross-validated r(2)) were used to assess the statistical quality of our models. Another statistical parameter, the ratio of the standard error to the activity range (s/AR), was also generated. CoMFA and HQSAR models were developed having statistically excellent properties, which also possessed good predictive ability for test set compounds. The best model was obtained when racemates were excluded from QSAR analysis. Thus, CoMFA of the n = 157 database gave excellent predictions with outstanding statistical properties. HQSAR did an outstanding job in statistical analysis and also handled predictions well.