Yusuf Tanrikulu

The holistic integration of virtual screening in drug discovery

During the past decade, virtual screening (VS) has come of age. In this review, we document the evolution and maturation of VS from a rather exotic, stand-alone method toward a versatile hit and lead identification technology. VS campaigns have become fully integrated into drug discovery campaigns, evenly matched and complementary to high-throughput screening (HTS) methods. Here, we propose a novel classification of VS applications to help to monitor the advances in VS and to support future improvement of computational hit and lead identification methods. Several relevant VS studies from recent publications, in both academic and industrial settings, were selected to demonstrate the progress in this area. Furthermore, we identify challenges that lie ahead for the development of integrated VS campaigns.

Voyages to the (un)known: adaptive design of bioactive compounds

De novo drug design has emerged as a valuable concept for the rapid identification of lead structure candidates. In particular, fragment-based molecular assembly methods have been successfully employed for the automated design of screening compounds. Here, we review the current status of these approaches, with an emphasis on adaptive techniques that can be used to artificially evolve novel bioactive molecules. Evolutionary algorithms (EAs) and particle swarm optimization (PSO) are presented as preferred techniques for iterative virtual synthesis and testing. By the inclusion of straightforward synthesis rules, druglike compounds can be obtained. Evolving compound libraries are particularly suited for hit and lead finding in situations where resources are limited and the complete testing of a large screening compound collection is prohibitive.

Pseudoreceptor models in drug design: bridging ligand- and receptor-based virtual screening.

Rational drug design is based on explicit or implicit structure–activity relationship models. Typically, receptor-based or ligand-based strategies are pursued, depending on the information available about known ligands and the receptor structure. Pseudoreceptor models combine the advantages of these two strategies and represent a unifying concept for both receptor mapping and ligand matching. They can provide an entry point for structure-based modelling in drug discovery projects that lack a high-resolution structure of the target. Here, we review the field of pseudoreceptor modelling techniques along with recent hit and lead finding applications, and critically discuss prerequisites, advantages and limitations of the various approaches.

The molecular mechanism of the inhibition by licofelone of the biosynthesis of 5-lipoxygenase products.

Licofelone is a dual inhibitor of the cyclooxygenase and 5‐lipoxygenase (5‐LO) pathway, and has been developed for the treatment of inflammatory diseases. Here, we investigated the molecular mechanisms underlying the inhibition by licofelone of the formation of 5‐LO products.

2,4-Diaminopyrimidines as histamine H4 receptor ligands--Scaffold optimization and pharmacological characterization.

The human histamine H(4) receptor (hH(4)R) is a promising new target in the therapy of inflammatory diseases and disorders of the immune system. For the development of new H(4)R antagonists a broad ligand-based virtual screening was performed resulting in two hits. The dissection of their common annelated aromatic core into its heteromonocyclic components showed that 2,4-diaminopyrimidine is a potent hH(4)R affinity scaffold, which was comprehensively investigated. Structure-activity relationship studies revealed that slight structural changes evoke extensive differences in functional activities and potencies: while o- and p-substituted benzyl amines mainly showed partial agonism, m-substituted and rigidified ones exhibited inverse agonist efficacy.

Scaffold Hopping by “Fuzzy” Pharmacophores and its Application to RNA Targets

Ideally, a novel lead structure represents a different chemotype from known ligands of a given target macromolecule. Virtual screening techniques and molecular de novo design are able to perform “scaffold-hopping”, and can thus be used as idea generators for medicinal chemistry. The task is to find isofunctional bioactive molecules with different backbone architectures. Here, we present the successful application of our “fuzzy” pharmacophore method LIQUID (“ligand-based quantification of interaction distributions”) to finding novel RNA ligand scaffolds and new inhibitors of in vitro protein expression. LIQUID is akin to our previously introduced SQUID approach. Both techniques model pharmacophoric features as Gaussian densities. While SQUID operates with univariate distributions, LIQUID employs trivariate models. In a retrospective virtual-screening study, we first compared LIQUID and SQUID, which both work on 3D pharmacophores, with the corresponding 2D CATS approach. All three methods encode pharmacophoric information as a correlation vector; this allows for rapid alignment-free similarity searching (by pair-wise Euclidian distance) in large compound libraries. We assessed their ability to retrieve known ligands of six drug targets from a collection of drugs and lead compounds compiled from recent literature. Overall, the CATS 2D method outperformed the 3D methods in this study, and SQUID retrieved slightly more hits than LIQUID (Table 1). This outcome is not entirely surprising, as we employed only a single calculated conformation for each of the molecules, and owing to the fact that the trivariate pharmacophore points in the LIQUID models are more restrictive than their univariate counterparts in SQUID. Notably, within the error margins, all methods performed comparably and clearly demonstrated their ability to retrieve active species from a collection of drug-like compounds. As all targets employed for this preliminary retrospective assessment were proteins, we checked the scaffold-hopping abilities of SQUID and LIQUID by using an RNA target, namely the trans-activation response element (TAR) of human immunodeficiency virus (HIV) 1. AIDS is caused by infection with HIV, which belongs to the class of retroviruses. This viral class is able to transpose its genome and use the host’s replication machinery for proliferation. Regulatory viral units have been selected as targets for blocking viral replication. Specific interaction of TAR RNA with the Tat protein is essential for virus Table 1. Average enrichment of active species expressed as average enrichment factor <ef>=hitsfound/hitsexpected ( SD), yielded by retrospective virtual screening.

Homology Model Adjustment and Ligand Screening with a Pseudoreceptor of the Human Histamine H4 Receptor

A new pseudoreceptor modeling method (PRPS) was applied to the refinement of a homology model of the human histamine H4 receptor (H4R), the prediction of a ligand binding site, and virtual screening. Retrieval of two new H4R ligands demonstrates the biological relevance of the pseudoreceptor model and provides a means for finding new hits and leads in the early phases of drug discovery.

Scaffold-Hopping Cascade Yields Potent Inhibitors of 5-Lipoxygenase

5-lipoxygenase (5-LO) is a validated drug target for the treatment of inflammation and allergic reactions as this enzyme is involved in catalyzing the conversion of arachidonic acid into leukotrienes. Inhibitors with dual activity towards both 5-LO and cyclooxygenase (COX), such as Licofelone, have been described as potent analgesic, anti-inflammatory and antiasthmatic agents lacking the gastrointestinal side effects seen with selective COX inhibitors. More recently, dual 5-LO/COX-2 inhibitors have been suggested as potential anticancer drugs. In this study, ligand-based virtual screening methods were used in an iterative fashion to identify new inhibitors of 5-LO product formation. The study consisted of four subsequent cycles of virtual screening, including 3Dand 2D-based methods and substructure searching, as well as biochemical testing. The iterative steps led to the discovery of a pyridine–imidazole-based lead structure series with nanomolar inhibitory activity in a cellular assay, demonstrating the applicability of advanced virtual screening techniques for designing small, focused, screening libraries that yield high hit rates in cell-based assays. We selected 11 dual 5-LO/COX reference inhibitors from the literature (Supporting Information, figure S1), with the aim to explore the ability of ligand-based virtual screening methods to retrieve isofunctional chemotypes with different backbone architecture (“scaffold-hopping”) from a large compound collection. The first step involved a broad, virtual screening process following the protocol outlined in Figure 1. For each of the 11 queries, two alignment-free similarity searches were performed in the Asinex Gold (November 2005: 231812 compounds) and Platinum (132250 compounds) collections (Asinex Ltd. , Moscow, Russia) using the “Charge3D” and “TripleCharge3D” methods. Briefly, “Charge3D” is an implementation of the correlation vector approach developed by Gasteiger and co-workers. The method compares two molecules based on their three-dimensional distribution of partial atom charges: Euclidian distances of all atom-pair combinations in one molecule are calculated (distances within a certain range are allocated to the same bin), and the charge values of the two atoms that form a pair are multiplied to yield a single value per atom-pair (charge values that were assigned to the same bin were added). Equation (1) describes the autocorrelation vector (CV) calculation used by “Charge3D”, where d is the distance in F, qi and qj are partial atomic charges, A is the number of atoms in a molecule and d defines the Kronecker delta (1 if a given atom pair exist, 0 otherwise). “TripleCharge3D” is an extension of this technique but makes a distinction between atom-pair types according to charge signs (++ , + , ).

Molecular characterization of EP6—A novel imidazo[1,2-a]pyridine based direct 5-lipoxygenase inhibitor

5-Lipoxygenase (5-LO) is a crucial enzyme of the arachidonic acid (AA) cascade and catalyzes the formation of bioactive leukotrienes (LTs) which are involved in inflammatory diseases and allergic reactions. The pathophysiological effects of LTs are considered to be prevented by 5-LO inhibitors. In this study we present cyclohexyl-[6-methyl-2-(4-morpholin-4-yl-phenyl)-imidazo[1,2-a]pyridin-3-yl]-amine (EP6), a novel imidazo[1,2-a]pyridine based compound and its characterization in several in vitro assays. EP6 suppresses 5-LO activity in intact polymorphonuclear leukocytes with an IC(50) value of 0.16μM and exhibits full inhibitory potency in cell free assays (IC(50) value of 0.05μM for purified 5-LO). The efficacy of EP6 was not affected by the redox tone or the concentration of exogenous AA, characteristic drawbacks known for the class of nonredox-type 5-LO inhibitors. Furthermore, EP6 suppressed 5-LO activity independently of the cell stimulus or the activation pathway of 5-LO contrary to what is known for some nonredox-type inhibitors. Using molecular modeling and site-directed mutagenesis studies, we were able to derive a feasible binding region within the C2-like domain of 5-LO that can serve as a new starting point for optimization and development of new 5-LO inhibitors targeting this site. EP6 has promising effects on cell viability of tumor cells without mutagenic activity. Hence the drug may possess potential for intervention with inflammatory and allergic diseases and certain types of cancer including leukemia.

Self-organizing molecular fingerprints: a ligand-based view on drug-like chemical space and off-target prediction

BACKGROUND Reliable prediction of multiple ligand-receptor interactions for a given bioactive compound helps recognize and understand off-target effects, and enables drug re-purposing and scaffold-hopping in lead discovery. We developed a ligand-based computational method for drug-target prediction that is independent from protein structural analysis. METHOD The idea is to infer drug targets from the pharmacophoric feature similarity of known ligands, and define functional target similarity from a ligand perspective, which also provides access to targets with unknown structures. First, known ligands were represented by topological pharmacophoric features. Then, the self-organizing map technique was used to generate fingerprint patterns for similarity analysis, where each resulting fingerprint represents a drug target. Target fingerprints were clustered and analyzed for correlations. Well-structured dendrograms were obtained presenting interpretable and meaningful relationships between drug targets. CONCLUSION Self-organization of fingerprints reduces noise from molecular pharmacophore descriptors, captures their essential features, and reveals potential cross-activities of ligand classes and off-target effects of bioactive compounds.