Deciphering ligand dependent degree of binding site closure and its implication in inhibitor design: A modeling study on human adenosine kinase

https://doi.org/10.1016/j.jmgm.2009.12.001Get rights and content

Abstract

Protein flexibility plays a significant role in drug research due to its effect on accurate prediction of ligand binding mode and activity. Adenosine kinase (AK) represents a highly flexible binding site and is known to exhibit large conformational changes as a result of substrate or inhibitor binding. Here we propose a semi-open conformation for ligand binding in human AK, in addition to the known closed and open forms. The modeling study illustrates the necessity of thorough understanding of the conformational states of protein for docking and binding mode prediction. It has been shown that predicting activity in the context of correct binding mode can improve the insight into conserved interactions and mechanism of action for inhibition of AK. Integrating the knowledge about the binding modes of ligands in different conformational states of the protein, separate pharmacophore models were generated and used for virtual screening to explore potential novel hits. In addition, 2D descriptor based clustering was done to differentiate the ligands, binding to closed, semi-open and open conformations of human AK. The results indicated that binding of all AK inhibitors cannot be described by same rules, instead, they represent a rule based preference for inhibition. This inference about tubercidins binding to semi-open conformation of human AK may facilitate in finding much extensive space for AK inhibitors.

Introduction

Human Adenosine Kinase (AK), a promising therapeutic drug target [1], [2] against a variety of diseases including hypertension [3], epilepsy [4], pain [5], [6], diabetes [7], and inflammation [8], [9], has recently attracted great interest in the search of potent and selective inhibitors. Inhibition of AK results in increased intracellular adenosine, which passes out of the cell via passive diffusion or via nucleoside transporter(s) to activate nearby cell-surface adenosine receptors and produce a wide variety of therapeutically beneficial activities against the said diseases [10], [11]. There are a number of reports on synthesis, SAR [12], [13] and QSAR studies of nucleosidic analogues as AK inhibitors [14], [15]. Nucleoside kinases, such as AK, cytidine kinase, thymidine kinase, catalyze phosphorylation of a variety of nucleosides, including nucleoside-based drugs, giving rise to nucleoside-5′-O-phosphates which are incorporated into RNA or DNA via the corresponding triphosphates [16]. Thus, 5′-O-phosphates of these nucleosides are responsible not only for the pharmacological activity but also for their cytotoxic side effects [12]. Therefore, designing molecules having different scaffold than the nucleosides is demanding. With the objective of finding new scaffolds, docking and pharmacophore analysis were carried out. In the present study, we have discussed and illustrated the crucial interactions for inhibition and proposed a rationale for new inhibitor design.

Two categories of compounds are known as inhibitors of AK, nucleosides [17] and non-nucleosides [18]. In an earlier study, a pharmacophore model for nucleoside and non-nucleoside AK inhibitors was developed, considering their similar binding modes [19]. Like other ribokinase family members, adenosine kinase undergoes large conformational changes as a result of ligand binding [20]. Recently, it has been found that nucleoside and non-nucleoside inhibitors bind in two different, closed and open conformations of enzyme, respectively. [20]. Nucleoside binding triggers a hinge bending by Glycine-Glycine (GG) switch on account of sugar anchoring with the apo form of enzyme.

The present study involves a series of aryl and non-aryl nucleosidic inhibitors (tubercidins) with varying degrees of substitutions at C4 and C5 positions (Table 1) [12], [13]. Based on an exhaustive docking analysis, using various parameterizations, we have found that due to larger size than the required size in the closed binding site, aryl nucleoside analogues cannot accommodate in the closed binding site as adenosine. This study has illustrated that even all nucleosidic inhibitors cannot bind in similar fashion in the adenosine binding site. Hence, a need for another relaxed conformation of protein was realized. Earlier report on the existence of semi-open conformation of T. gondii AK [21], a homologue of human AK, tempted us to further explore the binding of larger tubercidins in semi-open form of human AK. Ligand induced semi-open form of binding site is also reported recently in case of HIV–RT [22], [23].

In this report we have proposed for the first time that, like T. gondii AK, human adenosine kinase also possesses a semi-open/semi-close conformation to accommodate the larger analogues. The docking results indicate that the key ligand–protein interactions are same for all tubercidin inhibitors, irrespective of their binding to close or semi-open protein conformations, thereby preserving all crucial interactions for activity. We started this study by presuming that only aryl molecules can not bind in the adenosine binding site due to large aryl groups at C4 and C5 position but at the end we could find that it is not only the presence of aryl groups, instead, few spatial requirements force the molecules to bind in different ways in the active site. Presence of large hydrophobic aromatic group at C4 position of the ligands has been predicted to be the main discriminating factor for separating the aryl binders from the non-aryl binders.

Again by comparing the interactions of nucleosides with those of non-nucleosidic inhibitors, we have analyzed the importance of some interactions, which are conserved for all the molecules interacting with the enzyme and play important role in molecular recognition. For all kinds of inhibitors and protein conformations, three separate pharmacophore models for lead identification have been generated. The interpretation of binding interactions was carried out in the light of pharmacophore modeling data. Coherent result for diverse binding patterns of inhibitors was obtained by 2D descriptor based cluster analysis. We have also discussed why and how molecules choose the mode of binding and in what way the information could be utilized for designing new inhibitors. The pharmacophore models were validated and finally used as queries for 3D database mining.

Thus to gain more insight into structure–activity relationships we carried out this docking and pharmacophore analysis. More importantly, we have shown that information encoded in this approach can be effective to create virtual libraries for selectively targeting particular conformation of protein. Incorporation of this knowledge into virtual screening queries represents a much wider chemical space of molecules, which could be identified as potent AK inhibitors.

Section snippets

Preparation of proteins and ligands

Adenosine and tubercidin analogues are known to bind similarly in the same binding site [24]. Comparison of AK bound to 5-iodotubercidin (PDB code 2I6A) and to adenosine (PDB code 1BX4) shows that the iodine atom of 5-iodotubercidin in 2I6A replaces a conserved and highly stabilized water molecule, which is present in 1BX4, deeply buried into the pocket. Adenosine binding induces protein closing, and subsequent retention of two water molecules W1 and W2 (near C4 and C5 substituents,

Overall docking results, binding mode prediction and analysis of scores

To assess the ability of docking strategy, ADO was initially docked as a control and the result was compared with the crystal structure of the enzyme inhibitor complex (1BX4). A docked structure of ADO, the natural substrate, in the protein (Fig. 1a) reveals all interactions of bound mode in 1BX4, with RMSD less than 0.5 Å, in all the four docking runs. The validity of GOLD protocol for the water system was assessed by docking 5-iodotubercidin (compound 2 in Table 1) into protein (1BX4 after

Conclusions

The purpose of this study was (1) to know the binding modes within the tubercidin analogues, (2) generate pharmacophores for validation of different binding modes for tubercidins and utilization of those pharmacophores as search tools to identify potent chemical entities. This work illustrates that the molecules of same chemical series, having different degrees of substitutions, bind to different protein conformations, showing similar interactions. As far our knowledge, this is the first report

Acknowledgements

The authors thank Council of Scientific and Industrial Research (CSIR), New Delhi, for providing financial grant under Mission Mode Program CMM 0017. SB thanks UGC for Senior Research Fellowship. We wish to thank Dr Peter Carlqvist and Gary Battle from CCDC for the useful suggestions, Katalin Nadassy from Accelrys for helping on some matters while dealing with Catalyst.

References (47)

  • G. Jones et al.

    Development and validation of a genetic algorithm for flexible docking

    J. Mol. Biol.

    (1997)
  • G. Klebe

    Virtual ligand screening: strategies, perspectives and limitations

    Drug Discov. Today

    (2006)
  • D. Boison

    Adenosine as a modular of brain activity

    Drug News Perspect.

    (2007)
  • E.A. Kowaluk et al.

    Therapeutic potential of adenosine kinase inhibitors

    Expert Opin. Investig. Drugs

    (2000)
  • K.J. Broadley

    Drugs modulating adenosine receptors as potential therapeutic agents for cardiovascular diseases

    Expert Opin. Ther. Pat.

    (2005)
  • M.F. Jarvis

    Purinergic modulation of pain

    Rev. Analgesia

    (2003)
  • J. Sawynok et al.

    Adenosine and pain: recent findings with directly and indirectly acting agent

    Drug Dev. Res.

    (1998)
  • M. Sakowicz-Burkiewicz et al.

    Diabetes-induced decrease of adenosine kinase expression impairs the proliferation potential of diabetic rat T lymphocytes

    Immunology

    (2006)
  • S. McGaraughty et al.

    Recent developments in the discovery of novel adenosine kinase inhibitors: mechanism of action and therapeutic potential

    CNS Drug Rev.

    (2001)
  • H.B. Cottam

    New adenosine kinase inhibitors with oral antiinflammatory activity

    Drugs Future

    (1994)
  • K.A. Jacobson et al.

    Adenosine receptors: pharmacology, structure-activity relationships, and therapeutic potential

    J. Med. Chem.

    (1992)
  • B.G. Ugarkar et al.

    Adenosine kinase inhibitors. 1. Synthesis, enzyme inhibition and antiseizure activity of 5-iodotubercidin analogues

    J. Med. Chem.

    (2000)
  • B.G. Ugarkar et al.

    Adenosine kinase inhibitors. 2. Synthesis, enzyme inhibition and antiseizure activity of diaryltubercidin analogues

    J. Med. Chem.

    (2000)
  • View full text