A cheminformatic study on chemical space characterization and diversity analysis of 5-LOX inhibitors

https://doi.org/10.1016/j.jmgm.2020.107699Get rights and content

Highlights

  • Diversity and complexity of the 5-LOX and FLAP inhibitors sets has been explored.

  • These databases show a comparable polarity and flexibility to the approved drugs.

  • 5-LOX inhibitors cover most medicinal property space, but some are broadened.

  • New chemical structures must be created as FLAP inhibitors due to limited diversity.

  • Enamine LOX virtual library could provide novel potent 5-LOX inhibitor scaffolds.

Abstract

The process of blocking 5-lipoxygenase (5-LOX) catalyzed leukotriene biosynthesis has been recognized for the past few decades as a promising therapeutic strategy for acute inflammatory, allergic, and respiratory diseases. Due to the toxicity effect of FDA approved 5-LOX inhibitor zileuton, novel 5-LOX inhibitors have been sought by the scientific community. As a result, a significant and relevant amount of information on the structure-activity of 5-LOX inhibitors has been released and stored in public databases. In this study, we aimed at the comprehensive cheminformatic characterization of the diversity and complexity of the chemical space of 5-LOX inhibitors and its activating protein FLAP inhibitors by comparing it with the Approved drug space and virtual LOX library. The visual representation of the property space indicates some compounds in the 5-LOX inhibitors space broaden the traditional medicinal space. The structural diversity of the databases is computed using complementary approaches, including Physicochemical Property (PCP) descriptors, molecular fingerprints, and molecular scaffold. With the apparent exception of approved drugs, the 5-LOX dataset shows more diversity compared to FLAP and LOX virtual library set. This study was able to identify the underlying patterns in the chemical and pharmacological properties space that were decisive for the drug discovery and development of 5-LOX inhibitors.

Introduction

Leukotrienes (LTs) are a class of potent inflammatory mediators produced by 5-LOX from arachidonic acid. The overproduction of LT causes severe allergic and inflammatory conditions. Inhibition of LT formation, therefore, has valuable therapeutic advantages in excessive and chronic inflammatory and allergic responses like asthma, rhinitis, rheumatoid arthritis, gastroesophageal reflux disease, and atherosclerosis [1]. Moreover, the role of LTs in carcinogenesis has also been documented recently [[2], [3], [4], [5]]. Because traditional anti-inflammatory treatments, like treatment with NSAID, are still far from effective in many of these diseases, new and improved approaches are being actively sought to counter these conditions. Methods that inhibit the biosynthesis of LTs are, therefore, of interest as potential therapies for such diseases. The non-heme iron-containing enzyme 5-LOX catalyzes the first step in the biosynthesis of leukotriene. 5-LOX selective inhibition offers a definite means of reducing the effects of all leukotrienes, and such an inhibitor could form a new class of therapeutic agents [6].

Scientists have been discovered several types of 5-LOX inhibitors depending upon on the mechanism of action such as (i) iron-chelating inhibitors, (ii) competitive, reversible (non-redox) inhibitors (iii) inhibitors of the 5-lipoxygenase activating protein (FLAP) and (IV) redox inhibitors [7]. These large amounts of structure-activity data of 5-LOX inhibitors obtained as a result of general screening, high-throughput screening (HTS) and combinatorial chemistry is collected and deposited not only in research papers but also in public domain databases such as ChEMBL [8], PubChem [9], BindingDB [10], etc. These compound collections contribute to the development of ’Biologically Relevant Chemical Space (BRCS) of 5-LOX. Cheminformatic characterization of these chemical spaces is a significant step towards virtual or experimental testing to recognize novel biologically active molecules [[11], [12], [13]]. Besides, data mining of these chemical spaces could help the development of models that can be beneficial to predict the activity of a new compound [14]. Computational approaches have been playing an essential role in identifying 5-LOX inhibitors without any limitations of known 5-LOX inhibitors found. With the increase in the quantity and complexity of 5-LOX inhibitor’s structural data, chemical space analysis, and cheminformatic modeling are becoming increasingly important to understand and predict the interactions between inhibitors and 5-LOX protein.

So, visualization and characterization of biologically relevant chemical space of 5-LOX and their compound distributions are critical to medicinal chemists as it can assist in understanding molecular features that are pharmaceutically important [11,15]. Different computational data mining and visualization methods and machine learning algorithms that originally developed for computer science is now used mainly for the understanding the chemical space of biological interest [13,[16], [17], [18], [19]]. Standard statistical and classification techniques can also be used to organize datasets and evaluate the chemical neighborhood of potent hit [[20], [21], [22], [23]]. With the help of these methods, cheminformatic characterization of chemical space of inhibitors of various targets or disease has been reported for the last few decades by many eminent scientists, especially the team lead by Jose L. Medina-Franco [12,[24], [25], [26], [27], [28], [29]].

In recent years, several Structure-Activity Relationship (SAR), Quantitative Structure-Activity Relationship (QSAR), and other predictive models of 5-LOX and its activating protein ’FLAP’ inhibitors have been developed [[30], [31], [32], [33], [34]]. However, there are no systematic cheminformatic studies on the structural diversity and chemical space distribution analysis of 5-LOX and FLAP inhibitor’s chemical space. Therefore, exploring and navigating the biologically relevant 5-LOX and FLAP inhibitor’s chemical space is of the utmost importance. It can provide an opportunity for the analysis and enumeration of the compound in the chemical space of 5-LOX and FLAP.

So, in this work, we have tried to locate the chemical space of 5-LOX and FLAP currently stored in a major public databases ’ChEMBL’ and characterized these spaces using multiple criteria including physicochemical properties (PCP), structural fingerprints, and molecular scaffolds. In all these studies, we have compared the chemical space of 5-LOX, and FLAP inhibitors to the dataset of FDA approved drugs from ’DrugBank’ because this FDA approved drug database is a commonly used reference compound database in drug discovery campaigns. We have also tried to understand and validate the Target focused virtual LOX library created and offered by the Enamine database by comparing it to the 5-LOX dataset.

Section snippets

5-LOX and FLAP inhibitor dataset

The study of the complexity and diversity of large molecular databases allows for the creation of high-quality leads. It thus increases the performance of real and virtual drug design compound libraries. Here we too tried to understand the complexity and diversity of 5-LOX and FLAP chemical space. For this purpose, molecules in the ChEMBL database, which are experimentally tested for 5-LOX and FLAP protein inhibitory potency, were chosen. The chemical structures and activity data of 5-LOX and

Physicochemical properties of 5-LOX chemical space

The property distribution was evaluated with SPSS box plots. Fig. 2 shows the box and whisker plots spread out of the basic PCP of all four datasets. The primary portion of the chart is a box, which indicates the interquartile range, i.e., it shows where the central portion of the data is. The first quartile, Q1 (25% mark), and the third quartile, Q3 (75% mark), are situated at lower and upper ends of the box, respectively. The ’median’ represents a horizontal bar in the middle of the box. The

Conclusion

This study discusses a comprehensive cheminformatic characterization of the chemical space of 5-LOX and FLAP inhibitors obtained from the CHEMBL database by comparing it with the Approved Drug space. Also, we have validated the LOX virtual library created by the Enamine database by comparing it to the 5-LOX dataset. Analysis of the distributions of PCPs like HBA, HDB, and TPSA indicated that the compounds screened as inhibitors of 5-LOX and FLAP are, in general, less or comparable polar than an

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

The author, T.K. Shameera Ahamed, express her sincere gratitude to Human Resource Development Group’s Council of Scientific & Industrial Research (CSIR), India, for the financial support.

References (53)

  • L.G. Melstrom et al.

    Overexpression of 5-lipoxygenase in colon polyps and cancer and the effect of 5-LOX Inhibitors in vitro and in a murine model

    Clin. Canc. Res.

    (2008)
  • R. Hennig et al.

    5-Lipoxygenase, a marker for early pancreatic intraepithelial neoplastic lesions

    Canc. Res.

    (2005)
  • Y. Chen et al.

    Loss of the Alox5 gene impairs leukemia stem cells and prevents chronic myeloid leukemia

    Nat. Genet.

    (2009)
  • R. Alan

    Brash, lipoxygenases: occurrence, functions , catalysis , and acquisition of substrate

    J. Biol. Chem.

    (1999)
  • A. Gaulton et al.

    ChEMBL: a large-scale bioactivity database for drug discovery

    Nucleic Acids Res.

    (2012)
  • S. Kim et al.

    PubChem substance and compound databases

    Nucleic Acids Res.

    (2016)
  • T. Liu et al.

    BindingDB: a web-accessible database of experimentally determined protein-ligand binding affinities

    Nucleic Acids Res.

    (2007)
  • J.-L. Reymond

    The chemical space project

    Acc. Chem. Res.

    (2015)
  • E.F. Gortari et al.

    Epigenetic relevant chemical space: a chemoinformatic characterization of inhibitors of DNA methyltransferases

    RSC Adv.

    (2015)
  • N. Singh et al.

    Chemoinformatic analysis of combinatorial libraries, drugs, natural products, and molecular libraries small molecule repository

    J. Chem. Inf. Model.

    (2009)
  • A. Yosipof et al.

    Data mining and machine learning models for predicting drug likeness and their disease or organ category

    Front. Chem.

    (2018)
  • J.L. Medina-Franco

    Interrogating novel areas of chemical space for drug discovery using chemoinformatics

    Drug Dev. Res.

    (2012)
  • G. Sliwoski et al.

    Computational methods in drug discovery

    Pharmacol. Rev.

    (2013)
  • C. Lipinski et al.

    Navigating chemical space for biology and medicine

    Nature

    (2004)
  • A.S. Rifaioglu et al.

    Recent applications of deep learning and machine intelligence on in silico drug discovery: methods, tools and databases

    Briefings Bioinf.

    (2018)
  • A. Golbraikh et al.
  • Cited by (1)

    • Concepts and applications of chemical fingerprint for hit and lead screening

      2022, Drug Discovery Today
      Citation Excerpt :

      For example, CSNAP3D, proposed by Lo et al., combines shape and pharmacophore characteristics to measure the 3D similarities among ligands and showed a significant improvement in target prediction.71 Recently, Ahamed et al. proposed a novel methodology to analyze the cheminformatic characteristics and diversity of chemical space of 5-LOX through combining fingerprints of ECFP, MACC, PC, molecular scaffolds, and basic pharmaceutical relevance property descriptors.72 Capecchi et al. designed a universal fingerprint (suitable for drugs, biomolecules, and the metabolome) to describe and search chemical space called MAP4 (MinHashed atom-pair fingerprint up to a diameter of four bonds) by combining advantages of AP (i.e., ability to describe large molecules and with excellent perception of molecular shape) and circular fingerprints (i.e., ability to encode molecular substructures in detail, performing well in small-molecule benchmarking studies).73

    View full text