GPCR-Targeted Library
Medicinal and Computational Chemistry Dept., ChemDiv, Inc., 6605 Nancy Ridge Drive, San Diego, CA 92121 USA, Service: +1 877 ChemDiv, Tel: +1 858-794-4860, Fax: +1 858-794-4931, Email:
玫瑰花刺
Introduction
Owing to historic inefficiency of mass random bioscreening, the current paradigm suggests that target-specific and pharmacokinetic properties of small molecule libraries should be addresd as early as possible in the discovery process. Computational medicinal chemistry can address this problem at the level of pre-synthetic library design. A number of advanced in silico methods have recently been developed and applied to combinatorial templates to enhance their target-specific informational content. Appropriate strategies for the design of combinatorial libraries are developed in accordance with the target, dia area, resources on hand and the specific project goals.
In this description, we prent a rational, practical approach to the design of GPCR-targeted combinatorial library. The goal of the combinatorial synthesis planning strategy prented here is to construct an algorithm utilizing simple, automated procedures for designing combinatorial libraries that are expected to show GPCR-activity.
1. GPCRs as promising drug target
什么是彩礼
The superfamily of G-protein-coupled receptors (GPCRs) is a diver group of transmembrane proteins crucial to eukaryotic cellular signaling i. GPCRs initiate cascades of cellular respons to diver extracellular mediators and are involved in all common human dias. Nearly 40% of marketed drugs act through modulation of GPCR functions ii and up to 70% of novel therapeutics in development target known GPCRs iii. In addition, veral hundred “orphan” GPCRs (which have no natural ligands identified as yet), are the focus of an inten drug discovery effort in many programs. Characterization of orphan GPCRs will substantially facilitate rearch in human physiology and pharmacology. GPCR family consists of ven basic class: Rhodopsin- and Secretin-like receptors, Metabotropic glutamate and Fungal pheromone receptors, cAMP receptors, Ocular albinism proteins and Frizzled/Smoothened subfamily. All the class listed are additionally subdivided into veral categories, for example, Rhodopsin-like GPCRs include amine subclass (Mu
scarinic acetylcholine, Dopamine, Histamine, Serotonin, Bradykinin, Adrenoceptors, etc.), peptide subclass (Angiotensin, Chemokine, Endothelin, Neurotensin, Opioid, Somatostatin, Tachykinin, Vasopressin-like, etc.), Nucleotide-like receptors (Adenosine and Purinoceptors) and other subclass iv.
The key to harnessing the clinical potential of particular GPCRs lies in the ability to elucidate their tissue- and dia-specific functions and identify the lective ligands for the receptors. The optimal ligands would be the potent small molecules with ADME/Tox properties required for the orally available drugs. Specifically, the optimal ligands need to posss high affinity and specificity for the target protein, and reasonable membrane permeability for biological activity in whole cell assays and in vivo models. The prime source of drug candidates is the focud small molecule libraries developed against particular receptors which are often compiled into protein class “GPCR-targeted libraries”. Such libraries are being built by drug discovery companies in hou and are available commercially from medicinal chemistry companies. Due to significant diversity of natural GPCRs ligands and the complexity of downstream events of GPCR signaling, the optimal choice of GPCR lib
rary construction strategy reprents a non-trivial and highly important problem.
2. Neural networks in the design of GPCR-targeted library
There are veral approaches to the design of GPCR-focud compound libraries, ranging from 2D simulation algorithms to the analysis of ligand receptor spatial arrangement and neural network (NN) learning QSAR systems. Over the last years, the methods bad on neural networks became popular due to their efficiency in solving the problem. Several recent studies described successful employment of neural network methods for gregation of pharmaceutical compounds in categories bad on different properties v. Recently, we have applied NN classification methodology for property-bad design of GPCR-targeted library vi. In particular, we have found that a proper combination of specific physicochemical allows to successfully differentiating GPCR ligands from compounds active against other target-specific class. Using the findings, the NN classification models were created with excellent discriminatory power. We have also attempted to solve the next level, more difficult problem: differentiation between specific class of GPCR ligands. The key goal of vii was to develop in silico procedure for the design of small-molecule libraries that would show a receptor-specific GPCR activity. In the fundamental work viii we have comprehensively investigated and reviewed peptidergic G-protein coupled receptors (pGPCRs), their small-molecule modulators a
s well as the related structure-bad design of such agents.
Fundamentally, neural network (NN) modeling allows optimizing a large number of input parameters in different areas of NN applications. In drug development this property of NNs is ud in "property-bad design" approach, by analogy with the terminology propod earlier ix. NN approach is an efficient tool for constraining the size of virtual compound libraries designed for primary bioscreening with target-specific activity. The property-bad approach is an alternative to a variety of more broadly ud target- and ligand-structure focud design methods. Despite of its track record of success for certain targets, target-focud design has rious drawbacks. Namely, the are an inability
to accurately estimate all target-ligand interactions, significant computation time, the ignorance of water microenvironment, the difficulties in correct generation of 3D structures and in the analysis of all possible spatial conformations of it, etc. Ligand structure-bad methods are indispensable in exploring the feasible chemistry space when many ligands for a
target are known and the active chemotypes are defined. However, the method is poorly suitable for the discovery of novel lead chemotypes. It is well documented that most popular ligand structure-bad methodologies (such as bioisosteric approach, or similarity-bad methods) are skewed toward the old scaffolds. In general ca, the target- and ligand structure-bad technologies can not adequately address all the real problems of rational drug design, particularly tho connected with the virtual screening of large compound databas or with the discovery of novel lead chemotypes. A similarity of molecular physico-chemical properties reprents an alternative design basis for target specific libraries. The underlying theory
states that every group of active ligand molecules can be characterized by a unique combination of physico-chemical parameters differentiating it from other target-specific groups of ligands. As a rule, receptors of one type share the structurally conrved ligand binding site. The structure of this site dictates the bundle of properties a receptorlective ligand should posss, such as specific steric, lipophilic, H-binding, and other features influencing the pharmacodynamic requirements. This theory is realized in computation models for quantitative discrimination between the ligand groups. Whenever a large t of active ligands is available for a particular receptor, the mean values of some key molecular properties can be considered as optimal and characteristic of this group of ligands. Ba
d on the values, one can generate a quantitative discrimination function that permits the lection of a ries of compounds to be assayed against the target. Finding such function is a key element for computational virtual screening programs. It is important for this function to be bad on physico-chemical rather than on structural properties to be capable of suggesting novel lead chemotypes.
锐怎么组词2.1. Unsupervid Kohonen-bad learning approach世界五百强公司
In most studies on application of neural networks in drug discovery, a supervid learning strategy was ud. The alternative unsupervid learning method becomes popular for comparative analysis and visualization of large ligands data ts x. For instance, benzodiazepine and dopamine data ts were compared recently with an implementation of a Kohonen network xi. In another study, a datat of 31 steroids binding to the corticosteroid binding globulin (CBG) receptor was modeled xii. Kohonen lf-organizing maps were ud for distinguishing between drugs and non-drugs with a t of descriptors derived from mi-empirical molecular orbital calculations xiii. It was emphasized that Kohonen map-bad classification does not depend on the definition of a non-drug, non-ligand data t, and, therefore, the virtual screening of active compounds can be conducted more objectively. This property of
unsupervid Kohonen learning strategy is particularly important in cas when the negative training t is unavailable or hard to define. In this work, we ud the unsupervid learning methodology for differentiation between various receptor-specific groups of GPCR ligands. With the data available, only positive training lections of molecules can be unambiguously identified, namely, the groups of ligands to particular GPCRs. The definition of a negative training t would be very complicated and, probably, unreliable, as only a few compounds with particular receptor-specific activity have been tested against all groups of GPCRs. This limitation restricts the application of multi-layer neural networks with a supervid learning procedure as an error back-propagation learning algorithm; an unsupervid approach is required. Among the unsupervid methods, we cho Kohonen neural network as the one with the most appropriate learning strategy for GPCR-targeted library design.
3. Concept and Applications4月3日是什么星座
GPCR-targeted library design at CDL involves:
• A combined profiling methodology that provides a connsus score and decision bad on various
advanced computational tools:
1. Unique bioisosteric morphing and funneling procedures in designing novel potential GPCR ligands with high IP value. We apply CDL’s proprietary Chemosoft TM software and commercially available solutions from Accelrys, MOE, Daylight and other platforms.
2. Neural Network tools for target-library profiling, in particular Self-organizing Kohonen maps, performed in SmartMining Software. We have also ud the Sammon mapping and Support vector machine (SVM) methodology as more accurate computational tools to create our GPCR-focud library.
心情郁闷3. In veral cas we have ud 3D-molecular docking approach to the focud library design.
4. Computational-bad `in silico` ADME/Tox asssment for novel compounds includes prediction of human CYP P450-mediated metabolism and toxicity as well as many pharmacokinetic parameters, such as Brain-Blood Barrier (BBB) permeability, Human Intestinal Absorption (HIA), Plasma Protein binding (PPB), Plasma half-life time (T1/2), Volume of distribution in human plasma (V d), etc.
A general approach to limiting the space of virtual libraries of combinatorial reaction products consists of implementation of a ries of special filtering procedures. The typical filtering stages are briefly summarized in Figure 1. A variety of "Rapid Elimination of Swill" (REOS) filters is ud to eliminate compounds that do not meet certain criteria xiv.
Figure 1. General procedures of lection of a rational target-specific subt within an initial virtual居家男人
combinatorial library
The criteria can include: (1) prence of certain non-desirable functional groups, such as reactive moieties and known toxicophores; (2) molecular size, lipophilicity, the number of H-bond donors/acceptors, the number of rotatable bonds. At the next stage the design focus on “lead” and “drug-likeness” of combinatorial molecules xv. The ADME/Tox properties of screening candidates should be taken into consideration as early as possible xvi. Additional filters are therefore ud for in silico prediction of some crucial ADME/Tox parameters, such as solubility in water, logD at different pH values, cytochrome P450-mediated metabolism and toxicity, and fractional absorption. Optimization of structural diversity is another natural and very important way to constrain the size of combinatorial libraries (reviewed in xvii). The fundamentals for the applications are described in a ries of our recent articles on the design of exploratory small molecule chemistry for bioscreening [for related data visit ChemDiv, Inc. online source: ]. Our multi-step in silico approach to GPCR-focud library design is schematically illustrated in Fig. 2.
Figure 2. Multi-step computational approach to GPCR-targeted libraries design
taonbao
This common approach was effectively applied for the developing of our GPCR-focud, in particular for Serotonin, Dopamine, Opioid, Endothelin, Cannabinoid, Bradykinin, Chemokine receptors, Adrenoceptors, etc.
