Discriminating Agonist from Antagonist Ligands of the Nuclear Receptors Using Different Chemoinformatics Approaches

Abstract

[a] 1Introduction Nuclear receptors (NRs) are transcription factors naturally switched on and off by small-molecule hormones that monitor aw ide range of physiological functions. NRs can be targeted in numerous diseases [1] and synthetic ligands can artificially modulate the action of NRs,m ainly by activating (agonist ligands) or inhibiting (antagonist ligands) its activity.I nt his context, identifying the best ligand of ag iven targeti sn ot sufficient,i ti sn ecessary to find al igand with the suitablep harmacological profilea nd thus to be able to predict the agonist or antagonist behaviour of aN Rl igand. Virtual screening methods are widely used to predict the activity of small compounds [2] and can also be applied to the prediction of the pharmacologicalp rofile of NRs ligands using the knowledgeo ft he molecular bases of NRs agonism and antagonism.[3] However,t he different virtual screening tools evaluated in this purpose lead to mixed results [4] and being able to predict the pharmacological profile of NRs ligands remains ac hallenge.In this study, we review the ability of currently available virtual screeningm ethodst oc haracterize specifically the pharmacologicalp rofile of NRs ligands using the 27 NRLiSt BDB datasets.[5] After at horough description of the results we obtained using molecular descriptors and ar ecall of the results we obtainedu sing ad ocking method [4i] and a3 D ligand-based (LB) and structure-based (SB) pharmacophore modeling method, [6] we will emphasize on the advantages and drawbacks of each approach. 2Material and MethodsNuclearR eceptors Ligands and Structures Benchmarking DataBase (NRLiStB DB). The NRLiSt BDB [5] is ap ublic benchmarkingd atabase dedicated to the NRs and constructed to be used for the evaluation of both SB and LB methods. The NRLiSt BDB includest he 27 NRs (out of the 48 known human NRs) for whichm ore thano ne agonist ligand, one antagonistl igand, and at least one experimental structure were described. For each NR, all of the ligands foundt ob e agonisto ra ntagonist in the scientific literature are provided in two separated datasets and all available human holo PDB structures are provided (except for RXRg,f or which Abstract:N uclear receptors (NRs) constitute an important class of therapeutic targets. During the last 4years, we tackled the pharmacological profilea ssessment of NR ligands for which we constructedt he NRLiSt BDB. We evaluated and compared the performance of different virtual screeninga pproaches:m ean of molecular descriptor distribution values, molecular dockinga nd 3D pharmacophore models. The simple comparison of the distributionp rofiles of 4885 molecular descriptors between the agonista nd antagonist datasets didn'tp rovide satisfying results. We obtained an overall good performance with the docking method we used, Surflex-Dock which was able to discriminate agonist from antagonistl igands. But the availability of PDB structures in the "pharmacological-profile-to-predictbound-state" (agonist-bound or antagonist-bound) and the a...

Meglumine acridone acetate, the ionic salt of CMA and N-methylglucamine, induces apoptosis in human PBMCs via the mitochondrial pathway

“…Along with high-affinity receptors, low-affinity ligands which must be present in high concentrations for the realization of a physiological response are characteristic of nuclear receptors 47,48 . Nuclear receptors having more than one agonist and one antagonist ligand 49 exist and are involved in a wide range of key physiological functions 50–52 . They are potential drug targets for numerous diseases and constitute an important class of therapeutic targets.…”

Decoys Selection in Benchmarking Datasets: Overview and Perspectives

“…Unwanted compounds, i.e., compounds that display unwanted activity or binding, can also be used as negatives. For instance, a recent study used ligands of the NRLiSt BDB (Lagarde et al, 2014a ) either as active compounds or decoy compounds, depending on their activity for each nuclear receptor; antagonist (or agonist) ligands of a given nuclear receptor were used as decoys to evaluate agonistic (or antagonistic) pharmacophores (Lagarde et al, 2016 , 2017 ). This strategy has shown successful results in the past: Guasch et al ( 2012 ) focused on PPAR γ partial agonists to avoid side effects accompanying full receptor activation and built an anti-pharmacophore model with known full agonist compounds to remove all potential full agonist compounds from their initial set of 89,165 natural products and natural product derivatives.…”

“…Additionally, only one single structure was considered for each protein while many docking studies pointed out that the structure selection is crucial for screening and docking, particularly for proteins that accommodate ligands with different binding modes (May and Zacharias, 2005 ; Ben Nasr et al, 2013 ; Lionta et al, 2014 ). A recent study has shown that the ligand pharmacological profile should be considered for both the active set design and the structure selection (Lagarde et al, 2017 ). For instance, nuclear receptors (NR) can be inhibited by antagonists or activated by agonists that differ in their structure and properties: agonists should be considered in the active set if the screening is performed on an agonist-bound structure while antagonists should be used in the active set if the screening is performed on an antagonist-bound structure.…”

Extending in Silico Protein Target Prediction Models to Include Functional Effects

“…Additional problems with MDDR are inconsistent annotation, since many activity classes are not on the target level (for example the activity class ‘anti-helminthic activity’) and relatively small numbers of compound-target pairs are available for modeling, compared to other current databases ( Lagunin et al, 2000 ). Other cheminformatics approaches discriminate between agonist from antagonist classifications of ligands at nuclear receptors across targets simultaneously (within a single-model architecture) ( Lagarde et al, 2017 ). This architecture could negatively affect performance due to the imbalance between the functional data and the requirement to assign probability scores across all target proteins.…”