Prediction of Human Clearance Using In Silico Models with Reduced Bias.

Predicting human clearance with high accuracy from in silico-derived parameters alone is highly desirable, as it is fast, saves in vitro resources, and is animal-sparing. We derived random forest (RF) models from 1340 compounds with human intravenous pharmacokinetic (PK) data, the largest data set publicly available today. To assess the general applicability of the RF models, we systematically removed structural-therapeutic class analogues and other compounds with structural similarity from the training sets. For a quasi-prospective test set of 343 compounds, we show that RF models devoid of structurally similar compounds in the training set predict human clearance with a geometric mean fold error (GMFE) of 3.3. While the observed GMFE illustrates how difficult it is to generate a useful model that is broadly applicable, we posit that our RF models yield a more realistic assessment of how well human clearance can be predicted prospectively. We deployed the conformal prediction formalism to assess the model applicability and to determine the prediction confidence intervals for each prediction. We observed that clearance can be predicted better for renally cleared compounds than for other clearance mechanisms. We show that applying a classification model for predicting renal clearance identifies a subset of compounds for which clearance can be predicted with higher accuracy, yielding a GMFE of 2.3. In addition, our in silico RF human clearance models compared well to models derived from scaling human hepatocytes or preclinical in vivo data.

Kinase Crystal Miner: A Powerful Approach to Repurposing 3D Hinge Binding Fragments and Its Application to Finding Novel Bruton Tyrosine Kinase Inhibitors.

Protein kinases represent an important target class for drug discovery because of their role in signaling pathways involved in disease areas such as oncology and immunology. A key element of many ATP-competitive kinase inhibitors is their hinge-binding motif. Here, we describe Kinase Crystal Miner (KCM)-a new approach developed at Boehringer Ingelheim (BI) that harvests the existing crystallographic information on kinase-inhibitor co-crystal structures from internal and external databases. About 1000 unique three-dimensional kinase inhibitor hinge binding motifs have been extracted from structures covering more than 180 different protein kinases. These hinge binding motifs along with their attachment vectors have been combined in the KCM for the purpose of scaffold hopping, kinase screening deck design, and interactive structure-based design. Prospective scaffold hopping using the KCM identified two potent and selective Bruton tyrosine kinase (BTK) inhibitors with hinge binding fragments novel to BTK.

Discovery of a potent and dissociated non-steroidal glucocorticoid receptor agonist containing an alkyl carbinol pharmacophore.

Synthesis and structure-activity relationship (SAR) of a series of alkyl and cycloalkyl containing non-steroidal dissociated glucocorticoid receptor (GR) agonists is reported. This series of compounds was identified as part of an effort to replace the CF3 group in a scaffold represented by 1a. The study culminated in the identification of compound 14, a t-butyl containing derivative, which has shown potent activity for GR, selectivity against the progesterone receptor (PR) and the mineralocorticoid receptor (MR), in vitro anti-inflammatory activity in an IL-6 transrepression assay, and dissociation in a MMTV transactivation counter-screen. In a collagen-induced arthritis mouse model, 14 displayed prednisolone-like efficacy, and lower impact on body fat and free fatty acids than prednisolone at an equivalent anti-inflammatory dose.

Perspectives on current approaches to virtual screening in drug discovery.

INTRODUCTION: For the past two decades, virtual screening (VS) has been an efficient hit finding approach for drug discovery. Today, billions of commercially accessible compounds are routinely screened, and many successful examples of VS have been reported. VS methods continue to evolve, including machine learning and physics-based methods. AREAS COVERED: The authors examine recent examples of VS in drug discovery and discuss prospective hit finding results from the critical assessment of computational hit-finding experiments (CACHE) challenge. The authors also highlight the cost considerations and open-source options for conducting VS and examine chemical space coverage and library selections for VS. EXPERT OPINION: The advancement of sophisticated VS approaches, including the use of machine learning techniques and increased computer resources as well as the ease of access to synthetically available chemical spaces, and commercial and open-source VS platforms allow for interrogating ultra-large libraries (ULL) of billions of molecules. An impressive number of prospective ULL VS campaigns have generated potent and structurally novel hits across many target classes. Nonetheless, many successful contemporary VS approaches still use considerably smaller focused libraries. This apparent dichotomy illustrates that VS is best conducted in a fit-for-purpose way choosing an appropriate chemical space. Better methods need to be developed to tackle more challenging targets.

Substituted phenyl as a steroid A-ring mimetic: providing agonist activity to a class of arylsulfonamide nonsteroidal glucocorticoid ligands.

A class of arylsulfonamide glucocorticoid receptor agonists that contains a substituted phenyl group as a steroid A-ring mimetic is reported. The structural design and SAR that provide the functional switching of a GR antagonist to an agonist is described. A combination of specific hydrogen bonding and lipophilic elements on the A-ring moiety is required to achieve potent GR agonist activity. This study culminated in the identification of compound 23 as a potent GR agonist with selectivity over the PR and MR nuclear hormone receptors.

Function-regulating pharmacophores in a sulfonamide class of glucocorticoid receptor agonists.

A class of α-methyltryptamine sulfonamide glucocorticoid receptor (GR) modulators was optimized for agonist activity. The design of ligands was aided by molecular modeling, and key function-regulating pharmacophoric points were identified that are critical in achieving the desired agonist effect in cell based assays. Compound 27 was profiled in vitro and in vivo in models of inflammation. Analogs could be rapidly prepared in a parallel approach from aziridine building blocks.

Non-steroidal dissociated glucocorticoid agonists: indoles as A-ring mimetics and function-regulating pharmacophores.

We report a SAR of non-steroidal glucocorticoid mimetics that utilize indoles as A-ring mimetics. Detailed SAR is discussed with a focus on improving PR and MR selectivity, GR agonism, and in vitro dissociation profile. SAR analysis led to compound (R)-33 which showed high PR and MR selectivity, potent agonist activity, and reduced transactivation activity in the MMTV and aromatase assays. The compound is equipotent to prednisolone in the LPS-TNF model of inflammation. In mouse CIA, at 30 mg/kg compound (R)-33 inhibited disease progression with an efficacy similar to the 3 mg/kg dose of prednisolone.

DemQSAR: predicting human volume of distribution and clearance of drugs.

In silico methods characterizing molecular compounds with respect to pharmacologically relevant properties can accelerate the identification of new drugs and reduce their development costs. Quantitative structure-activity/-property relationship (QSAR/QSPR) correlate structure and physico-chemical properties of molecular compounds with a specific functional activity/property under study. Typically a large number of molecular features are generated for the compounds. In many cases the number of generated features exceeds the number of molecular compounds with known property values that are available for learning. Machine learning methods tend to overfit the training data in such situations, i.e. the method adjusts to very specific features of the training data, which are not characteristic for the considered property. This problem can be alleviated by diminishing the influence of unimportant, redundant or even misleading features. A better strategy is to eliminate such features completely. Ideally, a molecular property can be described by a small number of features that are chemically interpretable. The purpose of the present contribution is to provide a predictive modeling approach, which combines feature generation, feature selection, model building and control of overtraining into a single application called DemQSAR. DemQSAR is used to predict human volume of distribution (VD(ss)) and human clearance (CL). To control overtraining, quadratic and linear regularization terms were employed. A recursive feature selection approach is used to reduce the number of descriptors. The prediction performance is as good as the best predictions reported in the recent literature. The example presented here demonstrates that DemQSAR can generate a model that uses very few features while maintaining high predictive power. A standalone DemQSAR Java application for model building of any user defined property as well as a web interface for the prediction of human VD(ss) and CL is available on the webpage of DemPRED: http://agknapp.chemie.fu-berlin.de/dempred/ .

In Silico Models of Human PK Parameters. Prediction of Volume of Distribution Using an Extensive Data Set and a Reduced Number of Parameters.

A novel, descriptor-parsimonious in silico model to predict human VDss (volume of distribution at steady-state) has been derived and thoroughly tested in a quasi-prospective regimen using an independent test set of 213 compounds. The model performs on par with a former benchmark model that relied on far more descriptors. As a result, the new random forest model relying on only six descriptors allows for interpretations that help chemists to design compounds with desired human VDss values. A comparison of in silico predictions of VDss with models using in vitro derived descriptors or in vivo scaling methods supports the strength of the in-silico approach, considering its resource- and animal-sparing nature. The strong performance of the in silico VDss models on structurally novel compounds supports the high degree of confidence that can be placed in using in silico human VDss predictions for compound design and human dose predictions.

An in silico method for predicting Ames activities of primary aromatic amines by calculating the stabilities of nitrenium ions.

In this paper, we describe an in silico first principal approach to predict the mutagenic potential of primary aromatic amines. This approach is based on the so-called "nitrenium hypothesis", which was developed by Ford et al. in the early 1990s. This hypothesis asserts that the mutagenic effect for this class of molecules is mediated through the transient formation of a nitrenium ion and that the stability of this cation is correlated with the mutagenic potential. Here we use quantum mechanical calculations at different levels of theory (semiempirical AM1, ab initio HF/3-21G, HF/6-311G(d,p), and DFT/B3LYP/6-311G(d,p)) to compute the stability of nitrenium ions. When applied to a test set of 257 primary aromatic amines, we show that this method can correctly differentiate between Ames active and inactive compounds, and furthermore that it is able to rationalize and predict SAR trends within structurally related chemical series. For this test set, the AM1 nitrenium stability calculations are found to provide a good balance between speed and accuracy, resulting in an overall accuracy of 85%, and sensitivity and specificity of 91% and 72%, respectively. The nitrenium-based predictions are also compared to the commercial software packages DEREK, MULTICASE, and the MOE-Toxicophore descriptor. One advantage of the approach presented here is that the calculation of relative stabilities results in a continuous spectrum of activities and not a simple yes/no answer. This allows us to observe and rationalize subtle trends due to the different electrostatic properties of the organic molecules. Our results strongly indicate that nitrenium ion stability calculations should be used as a complementary approach to assist the medicinal chemist in prioritizing and selecting nonmutagenic primary aromatic amines during preclinical drug discovery programs.

Crowdsourcing and open innovation in drug discovery: recent contributions and future directions.

The past decade has seen significant growth in the use of 'crowdsourcing' and open innovation approaches to engage 'citizen scientists' to perform novel scientific research. Here, we quantify and summarize the current state of adoption of open innovation by major pharmaceutical companies. We also highlight recent crowdsourcing and open innovation research contributions to the field of drug discovery, and interesting future directions.

Discovery of potent inhibitors of interleukin-2 inducible T-cell kinase (ITK) through structure-based drug design.

Interleukin-2 inducible T-cell kinase (ITK) is a member of the Tec kinase family and is involved with T-cell activation and proliferation. Due to its critical role in acting as a modulator of T-cells, ITK inhibitors could provide a novel route to anti-inflammatory therapy. This work describes the discovery of ITK inhibitors through structure-based design where high-resolution crystal structural information was used to optimize interactions within the kinase specificity pocket of the enzyme to improve both potency and selectivity.

5-Aminomethylbenzimidazoles as potent ITK antagonists.

Benzamide 1 demonstrated good potency as a selective ITK inhibitor, however the amide moiety was found to be hydrolytically labile in vivo, resulting in low oral exposure and the generation of mutagenic aromatic amine metabolites. Replacing the benzamide with a benzylamine linker not only addressed the toxicity issue, but also improved the cellular and functional potency as well as the drug-like properties. SAR studies around the benzylamines and the identification of 10n and 10o as excellent tools for proof-of-concept studies are described.

2-Aminobenzimidazoles as potent ITK antagonists: trans-stilbene-like moieties targeting the kinase specificity pocket.

Based on the information from molecular modeling and X-ray crystal structures, the kinase specificity pocket of ITK could be occupied upon extension of the right-hand-side of the 2-benzimidazole core of the inhibitors. 2-Aminobenzimidazoles with a trans-stilbene-like extension were designed and synthesized as novel ITK antagonists. Significant improvement on binding affinity and cellular activity were obtained through the trans-stilbene-like antagonists. Several compounds showed inhibitory activity in an IL-2 functional assay.

Quinol-4-ones as steroid A-ring mimetics in nonsteroidal dissociated glucocorticoid agonists.

We report on the nuclear receptor binding affinities, cellular activities of transrepression and transactivation, and anti-inflammatory properties of a quinol-4-one and other A-ring mimetic containing nonsteroidal class of glucocorticoid agonists.

Automatically updating predictive modeling workflows support decision-making in drug design.

Using predictive models for early decision-making in drug discovery has become standard practice. We suggest that model building needs to be automated with minimum input and low technical maintenance requirements. Models perform best when tailored to answering specific compound optimization related questions. If qualitative answers are required, 2-bin classification models are preferred. Integrating predictive modeling results with structural information stimulates better decision making. For in silico models supporting rapid structure-activity relationship cycles the performance deteriorates within weeks. Frequent automated updates of predictive models ensure best predictions. Consensus between multiple modeling approaches increases the prediction confidence. Combining qualified and nonqualified data optimally uses all available information. Dose predictions provide a holistic alternative to multiple individual property predictions for reaching complex decisions.

Crowdsourcing in pharma: a strategic framework.

Conceptually, all organizations can be described as coordinated actors working together to deliver a product(s), or provide a service(s). For organizations to remain competitive, it is important to have processes that look outward for external 'innovations' that could improve how work is done, and what is delivered. We present a comprehensive review of a variety of processes that pharmaceutical companies have used to engage external actors ('the crowd') to provide innovation in the service of delivering novel therapeutic agents. This culminates in a framework that provides a consolidated view of crowdsourcing processes, which in turn enables a strategic application of a crowdsourcing methodology based on problem type.

In silico predictions of genotoxicity for aromatic amines.

Insufficient drug safety is one of the major reasons for failure of drug candidates in Phase II and Phase III clinical trials. Determining toxicity early during the drug discovery process can help lower the attrition rate in clinical trials and lead to significant cost savings. In silico approaches can help to prioritize large numbers of compounds quickly and cost effectively in the early phase of drug discovery. One form of toxicity is genotoxicity due to mutagenicity. In this paper different in silico approaches for predicting mutagenicity, in particular in primary aromatic amines, are reviewed.

Identification of highly efficacious glucocorticoid receptor agonists with a potential for reduced clinical bone side effects.

Synthesis and structure-activity relationship (SAR) of a series of nonsteroidal glucocorticoid receptor (GR) agonists are described. These compounds contain "diazaindole" moieties and display different transcriptional regulatory profiles in vitro and are considered "dissociated" between gene transrepression and transactivation. The lead optimization effort described in this article focused in particular on limiting the transactivation of genes which result in bone side effects and these were assessed in vitro in MG-63 osteosarcoma cells, leading to the identification of (R)-18 and (R)-21. These compounds maintained anti-inflammatory activity in vivo in collagen induced arthritis studies in mouse but had reduced effects on bone relevant parameters compared to the widely used synthetic glucocorticoid prednisolone 2 in vivo. To our knowledge, we are the first to report on selective glucocorticoid ligands with reduced bone loss in a preclinical in vivo model.

Crowd computing: using competitive dynamics to develop and refine highly predictive models.

A recent application of a crowd computing platform to develop highly predictive in silico models for use in the drug discovery process is described. The platform, Kaggle™, exploits a competitive dynamic that results in model optimization as the competition unfolds. Here, this dynamic is described in detail and compared with more-conventional modeling strategies. The complete and full structure of the underlying dataset is disclosed and some thoughts as to the broader utility of such 'gamification' approaches to the field of modeling are offered.