Current Pharmaceutical Design - Volume 12, Issue 17, 2006

The very foundations of drug discovery research are being rapidly transformed by high throughput systems, automated assays, robotics and advanced computational applications in medicinal chemistry. Costs are dropping, the time to complete a cycle of discovery and compound characterization is lessening, and all the while the ability to assess a compound's possible therapeutic role is improving. Indeed, the drive to increase the speed and efficiency of drug discovery, from hit identification all the way through to the creation of development candidates has seen huge investments by major pharmaceutical companies in many new technologies, with the primary aim of synthesizing more compounds and screening them faster; all at reduced cost per compound or assay. Nowadays, computational medicinal chemistry brings together the most powerful concepts in modern chemistry, biology and pharmacology, linking medicinal chemistry with genomics and proteomics. However, following our experience, both ligand-based and structure-based approaches to drug discovery in the absence, but probably also in the presence, of the real 3D-structures require a multidisciplinary approach, where molecular models represent a structural context to efficiently integrate experimental data and inferences derived from molecular biological, biophysical, bioinformatic, pharmacological and organic chemical methods. Although not always achievable, the success of a synergistic effect among these disciplines is highly dependent on the experimental design. Synergy is best achieved when mutations are structurally interpretable, structural hypotheses are experimentally testable, ligands are well characterized pharmacologically, and the necessary chemical modifications of the ligands are feasible. The present edition opens with the interesting report by Silvio Tosatto and Stefano Toppo [1] on the large scale prediction of protein structure and function from primary sequence obtained by the state of art chemioinformatics methodologies. This is an instance of a likely future contribution of bioinformatics research towards an improvement in the druggable target structure and function prediction. Giovanna Scapin [2] recalls the crucial role of macromolecular crystallography in structural determination and structural-based drug design (SBDD) approach. This process has been aided by recent technological innovations such as high-throughput crystallization, high performance synchrotron beamlines, and new methods in structural bioinformatics and computational chemistry prompted by the structural genomics effort. Alfonso Pozzan [3] updates the role of molecular descriptors and methods for ligand based virtual high throughput screening in drug discovery. Christoph Steinbeck, Christian Hoppe, Stefan Kuhn, Matteo Floris, Rajarshi Guha and Egon L. Willighagen [4] deal with the potentiality of Java applications in molecular informatics. The Authors outline the perspective to develop chemoinformatics tools, such as Chemistry Development Kit (CDK), with the aim to speed up the drug discovery process. Nohad Gresh [5] reports about the development, validation, and applications of anisotropic polarizable molecular mechanics to study ligand and drug-receptor interactions. In fact, considering molecular docking one of the most useful application in structural-based drug design (SBDD) approach, a correct representation of intermolecular interaction energies is necessary for reliable drug-receptor docking studies. Gabriele Costantino [6] introduces the role of in silico approaches towards the understanding of the structurefunction relationships in metabotropic glutamate receptors (mGluRs) and other family C GPRCs. The present review will discuss the evolution of our perception in family C GPCRs structure and function as emerged from the critical comparison of in silico methods with molecular biology and crystallographic experiments Francesca Deflorian, Magdalena Bacilieri, Giampiero Spalluto and myself [7], address the concept of receptor multi-conformational state applied to G-protein coupled receptors (GPCRs). Particularly, we define ligand-based homology modeling as new approach to simulate the reorganization of the receptor induced by the ligand binding. Finally, we discuss its possible application in new GPCR ligand discovery. 2066 Current Pharmaceutical Design, 2006, Vol. 12, No. 17 Computational Applications in Medicinal Chemistry References [1] Tosatto SCE, Toppo S. Large-Scale Prediction of Protein Structure and Function from Sequence. Curr Pharm Design 2006; 12(17): 2067-2086. [2] Scapin G. Structural Biology and Drug Discovery. Curr Pharm Design 2006; 12(17): 2087-2097. [3] Pozzan A. Molecular Descriptors and Methods for Ligand Based Virtual High Throughput Screening in Drug Discovery. Curr Pharm Design 2006; 12(17): 2099-2110......

The identification of novel drug targets from genomic data involves the large-scale analysis of many protein sequences. Methods for automated structure and function prediction are an essential tool for this purpose. In this review we concentrate on the recent developments in the field of protein structure prediction and how these can be used to gain hints about the function of proteins. The current state-of-the-art is highlighted through recent community-wide experiments aimed at comparing different approaches. For structure prediction this allows the identification of key improvements to increase the crucial sequence to structure alignment needed for accurate models. Function prediction is a rapidly maturing field that is still being benchmarked. Definitions for protein function are presented and available methods, mostly concentrating on functional site descriptors and structural motifs, presented.

In the past few years macromolecular crystallography has become a standard technique used by many pharmaceutical and biotechnology companies. This methodology offers details of protein-ligand interactions at levels of resolution virtually unmatched by any other technique, and this approach holds the promise of novel, more effective, safer and cheaper drugs. Although crystallography remains a laborious and rather expensive technique, remarkable advances in structure determination and structure based drug design (SBDD) have been made in recent years. This process has been aided by recent technological innovations such as high-throughput crystallization, high performance synchrotron beamlines, and new methods in structural bioinformatics and computational chemistry prompted by the structural genomics effort. As a consequence of the increased availability of structural data, the use of structure-based information has expanded from simple protein-ligand interaction analysis to include other aspects of the drug discovery process like target selection and initial lead discovery that used to be almost the exclusive property of biology and chemistry. This review will cover recent examples to illustrate how macromolecular crystallography has evolved and how structural information is now being used in the different stages of the drug discovery process. Advantages and shortcomings of the methodology will also be discussed.

The aim of virtual high throughput screening is the identification of biologically relevant molecules amongst either tangible or virtual (large) collections of compounds. Amongst the various virtual screening approaches, those that are ligand based are becoming very popular due to the possibility to screen millions of molecules in a timely way. Descriptors and methods are briefly introduced and reviewed with more emphasis for those approaches that are based on fingerprint descriptors and that seems to be more utilized during the drug discovery process.

The Chemistry Development Kit (CDK) provides methods for common tasks in molecular informatics, including 2D and 3D rendering of chemical structures, I/O routines, SMILES parsing and generation, ring searches, isomorphism checking, structure diagram generation, etc. Implemented in Java, it is used both for server-side computational services, possibly equipped with a web interface, as well as for applications and client-side applets. This article introduces the CDK's new QSAR capabilities and the recently introduced interface to statistical software.

A correct representation of intermolecular interaction energies is necessary for reliable drug-receptor docking studies. While ab initio quantum chemistry with extended basis sets is the most accurate tool for that purpose, its use is precluded for very large molecular complexes. This constitutes the incentive for the development of accurate molecular mechanics potentials, in which the first-order electrostatic, and the second-order polarization energy contributions, are of essential importance. In this paper, we review the most important steps in the development of anisotropic, polarizable molecular mechanics (APMM) procedures. Among these, we illustrate validation tests of the ab initio-grounded, polarizable molecular mechanics potential, SIBFA (Sum of Interactions Between Fragments Ab initio computed). These are done by comparisons with parallel quantum-chemical (QC) results on representative multiply hydrogen-bonded complexes and polycoordinated complexes of one, or of two, divalent metal cations. For both kinds of complexes, the need to reproduce the non-additivity of the QC interaction energies is emphasized. One difficulty arises upon handling flexible molecules, due to the need to account simultaneously and consistently for the onset of inter- and intra-molecular polarization and charge-transfer effects. A new approach in the context of SIBFA was recently developed towards this aim, and tested in two cases of conformation-dependent cation-ligand interaction energies. The first relates to the complexes formed between the mecapto-carboxamide anion, an essential building-block of several Zn-metalloenzyme inhibitors, and Zn(II). The second relates to the complexes of the tetra-anionic pyrophosphate anion, a key building-block of ATP and GTP, with one or two divalent Zn(II) cations used as a probe. In the domain of applications, two recent studies are then presented. The first is the docking of the captopril drug to the active site of the binuclear Zn(II)- β-lactamase enzyme. The second is the complex of a non-hydrolyzable analog of ATP with the active site of a binuclear Mg(II)-dependent kinase. An extension to an open-shell cation, Cu(II), is finally presented. The encouraging results presented in this review show that APMM procedures could be used in large-scale studies of ligand and drug-receptor interactions.

Family C of the superfamily of G-protein coupled receptors is a growing family of heptahelical receptors, which includes, among others, metabotropic glutamate receptors (mGluRs) and GABA(B) receptors. A common feature of all the members of family C is a structural architecture much more complex than any other GPCRs. Computational studies, including homology modeling, pharmacophore definitions and molecular dynamics simulations have constantly flanked experimental approaches in the understanding of the receptor functioning. The present review will discuss the evolution of our perception in family C GPCRs structure and function as emerged from the critical comparison of in silico methods with molecular biology and crystallographic experiments.

G protein-coupled receptors (GPCRs) represent the largest family known of signal-transducing molecules. They convey signals for light and many extracellular regulatory molecules. GPCRs have been found to be dysfunctional/ dysregulated in a growing number of human diseases and they have been estimated to be the targets of more than 40% of the drugs used in clinical medicine today. The crystal structure of rhodopsin provides the first three-dimensional GPCR information, which now supports homology modeling studies and structure-based drug design approaches. Here, we review our recent work on adenosine receptors, a family of GPCRs and, in particular, on A3 adenosine receptor subtype antagonists. We will focus on an alternative approach to computationally explore the multi-conformational space of the antagonist-like state of the human A3 receptor. We define ligand-based homology modeling as new approach to simulate the reorganization of the receptor induced by the ligand binding. The success of this approach is due to the synergic interaction between theory and experiment.