Elucidating the binding specificity of interactive compounds targeting ATP binding cassette subfamily G member 2 (ABCG2)

doi:10.21203/rs.3.rs-4954223/v1

Download PDF

Research Article

Elucidating the binding specificity of interactive compounds targeting ATP binding cassette subfamily G member 2 (ABCG2)

https://doi.org/10.21203/rs.3.rs-4954223/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

The ATP-binding cassette (ABC) transporter superfamily plays a pivotal role in cellular detoxification and drug efflux. ATP binding cassette subfamily G member 2 (ABCG2) referred to as the Breast cancer resistance protein (BCRP) has emerged as a key member involved in multidrug resistance displayed by cancer cells. Understanding the molecular basis of substrate and inhibitor recognition, and binding within the transmembrane domain of ABCG2 is crucial for the development of effective therapeutic strategies. Herein, utilizing state-of-the-art molecular docking algorithms and molecular dynamic (MD) simulations, substrates and inhibitors molecular binding with ABCG2 are defined, distinctly. We performed extensive virtual screening of Drugbank to identify the potential candidates, and MD simulations of docked complexes were carried out in POPC lipid bound membrane. Further, the binding affinities of compounds were estimated by free binding energy employing MM-GBSA. Docking results show three compounds, Docetaxel > Tariquidar > Tezacaftor having highest binding affinities (> 12.00 Kcal/mol) for ABCG2. Remarkably, MM-GBSA results suggest the most stable binding of Tariquidar with ABCG2 as compared to the other inhibitors. These results provide valuable insights into the key residues that may govern substrate/inhibitor recognition, shedding light on the molecular determinants influencing substrate specificity, transport kinetics and ABCG2-mediated drug efflux. Hence, Tariquidar could serve as a potential candidate for the design of highly efficient inhibitors targeting ABCG2 in breast cancer therapy.

ABC transporter

Multidrug resistance

Molecular docking

MD simulations

Tariquidar

ATP-binding cassette (ABC) transporters belong to the omnipresent ABC superfamily of proteins that facilitate transport of a large and diverse spectrum of molecules [1]. ABC transporters play a fundamental role in the selective translocation of an array of molecules across biological membranes. While substrates of ABC transporters include amino acids, peptides, lipids and many other crucial molecules, their role in exporting therapeutic molecules has major implications in drug resistance in cancer and microbial cells [2, 3]. As a result, understanding the nuances of ABC transporters has paramount implications, as they not only influence essential cellular processes but also contribute significantly to human health and disease.

The core architecture of an ABC transporter consists of two transmembrane domains (TMDs) that create a channel for substrate transport, and two cytoplasmic nucleotide-binding domains (NBDs) that bind and hydrolyze ATP to power substrate translocation [1]. Despite this common blueprint, the transporter proteins are embellished with structural variations that confer the unique substrate specificity and transport mechanisms observed within the ABC transporter family[4]. Human ABC transporters have been recently grouped into seven categories (I-VII) depending on their TMD fold [4]. While the prokaryotic ABC systems encompass all the VII types, eukaryotic ABCA-ABCG subfamilies belong to the type IV and V only [3].

Among the human ABC transporters, ABCG2, also known as the breast cancer resistance protein (BCRP) has emerged as an extremely critical clinical target due to its multifaceted roles in human health and disease[5]. ABCG2 is known for its remarkable capacity to efflux a diverse array of substrates that include xenobiotics and toxins, endogenous metabolites, and anticancer drugs [6]. Due to the involvement of ABCG2 in drug resistance, it significantly compromises the efficacy of cancer treatment [6].

Recent years have witnessed extensive research being carried out on unraveling the intricacies and at the same time shedding light on its structural architecture, molecular mechanism involved in substrate transport and its inhibition and/or modulation[7, 8]. Due to the advancement in cryogenic electron microscopy (cryo-EM), a significant number of structural data has emerged, not only in the apo-state but also in complex with some common substrates and inhibitors of ABCG2 [7–10]. Despite these breakthroughs, there is still absence of a robust understanding of specific interactions that distinguish a substrate from inhibitor and features that are seen in molecules which can act both as substrate and an inhibitor. More specifically, how the binding affinities of various substrates and inhibitors fair against each other, is something we seek an answer for. This is extremely relevant in the context of pharmacophore designing because it is desirable to have anticancer drugs with least propensity to act as substrates for multidrug transporters like ABCG2 and it is always going to be beneficial to have drugs that can play a dual role by acting as inhibitor for efflux pump too. This paper aims to provide a comprehensive exploration of ABCG2, by collating insights obtained from molecular docking and molecular dynamic (MD) simulation approaches.

We utilized the drug bank data (https://go.drugbank.com/categories/DBCAT002663) which lists all the ABCG2 substrates identified so far and compared their affinities and specific interactions in reference to imatinib bound ABCG2 structure [8]. To contrast the transporter-substrate and transporter-inhibitor interactions, we also studied tariquidar and ko143 interactions with ABCG2. Notably, tariquidar also acts as a substrate/inhibitor of ABCG2, whereas ko143 is a non-substrate inhibitor[11, 12]. Altogether, we endeavor to contribute to mechanistic underpinnings governing substrate and inhibitor interaction with ABCG2. By amalgamating the findings, we sought to develop a holistic understanding of the features which may dictate the molecule to act as substrate and/or inhibitor. The present study thus aspires to pave the way for innovative therapeutic strategies exploiting the critical transporter with implications in multidrug resistance (MDR) phenomenon.

Chemical library preparation and virtual screening

On December 2022, Drugbank category BCRP/ABCG2 substrates (https://go.drugbank.com/categories/DBCAT002663) had 120 drugs name which was downloaded in sdf format and prepared for 3D structural geometry using ligprep module of Schrodinger suites. Each drug molecule was prepared for hydrogen atoms addition, ionization, and chemical stereo state calculation at biological pH. The 3D macromolecular structure of ABCG2 protein (PDB-ID : 6VXH) was downloaded from the PDB database [13] and scrutinized for the missing regions/residues. From N-terminal to C-terminal, five regions (1–34, 46–60, 302–327, 355–369, 655) were identified and corresponding missing were filled using the AlphaFold model entry (AF-Q9UNQ0-F1)[14]. As N-terminal region 1–35 region was a very coiled flexible region, this region was subtracted from the modeled 6VXH structure. Further, Modeled Protein was prepared using the protein preparation wizard of Schrodinger by adding the missing hydrogen atoms, removing water and any other metal ions, and protonation at pH 7[15]. Grid box was created in the trans-membrane region on ABCG2 protein using the bound STI ligand as reference binding site which was further used by the glide program for docking the chemical library in Extra precision (XP) mode[16]. After docking, the resultant receptor-ligand complexes were visualized and studied by UCSF chimera 1.9 [17], and Ligplot+[18].

Molecular Dynamics Simulation

To assess the dynamical stability and estimation of ligand binding affinity against the ABCG2 receptor protein, both ligands bound, and unbound forms were simulated using the Amber23 package[19]. Six screening hit compounds selected at the virtual screening stage were used as ligand bound complex while native receptor was used as unbound form. 20% cholesterol and 80% POPC lipid membrane was constructed around the transmembrane region of ABCG2 receptor protein using the packmol module of the Ambertools23 with box dimensions choose to ensure sufficient membrane coverage (10 Å minimum distance from protein to box edge) and 15 Å water layer thickness was added to surround the membrane receptor protein. During the packmol run, 0.15 M of NaCl was added in the water layer to neutralize the simulation box. tLeap module is used for final simulation box checking and ff14SB, lipid21 and GAFF force fields were used to parameterize the protein, lipid and small molecules respectively [20].

MPI version of PMEMD is used for 10ps minimization using steepest descent (SD) and then conjugate gradient (CG) (5ps for each) followed by six stages of gradual heating from 50 K to 300 K at the step size of 50K. After the completion of the heating process, the simulation box was further minimized for 10ps followed by 500 ps equilibration run under NVT ensemble condition. Final production run of 200 ns was done using the GPU version of PMEMD for each system with a time step of 2 fs and simulation frames were written after every 10 ps time interval. Total 20000 simulation frames were collected, and the last 50 ns (5000) frames were used for MM-(GBSA) binding energy calculation for each receptor complex. Cpptraj was utilized to evaluate the global structural order parameters, such as RMSD (root mean square deviation), Rg (radius of gyration) SASA (solvent accessible surface area), RMSF (root mean square fluctuation), PCA (Principal component analysis)[21–23].

Binding Free-energy estimation using MM-GBSA

Molecular Mechanics-Generalized Born Surface Area (MM-GBSA) is a computational approach used to estimate the binding free energies of protein complexes[24]. It uses molecular mechanics (MM) force fields with implicit solvent models, such as the generalized Born (GB) model, to account for solvation effects. MM-GBSA decomposes binding free energy into distinct energy terms which include Gas-phase interaction energy (ΔE_MM), Solvation free energy (ΔG_solv): Electrostatic solvation energy (ΔG_GB), nonpolar solvation energy (ΔG_SA) and Entropic contributions (-TΔS). The calculation of the binding free energy can be performed according to the following equations:

△G_bind = G_complex − (G_receptor + G_ligand) (1)

△G_bind = △H − T△S ~ △E_MM + △G_solv − T△S (2)

△E_MM = △E_inter + △E_{electorstatic} + △E_vdw (3)

△G_solv = △G_GB + △G_SA (4)

Here, the generalized Born (GB) model was used for the polar electrostatic component, whereas SASA calculated the nonpolar electrostatic contribution. The last 50 ns simulation trajectory sampled every 10ps (500 frames) is used for binding free energy calculation.

In the drug discovery process virtual screening plays a vital role in identifying the novel and potential drug candidates and derivatives from the chemical libraries [10, 24, 25]. The identified lead molecules help in designing the target specific chemical scaffolds, facilitating the molecular interactions[22, 24]. The efficacy of the drug compounds relies on the binding stability at the active site of target molecules [26]. Molecular docking has become a routine tool to define the protein-ligand molecular interactions[25]. Moreover, the molecular dynamics (MD) simulations provide comprehensive assessment of protein structural flexibility and atomic interaction[27]. The MM-GBSA analysis from the trajectories obtained from MD simulations allows the accurate estimation of binding free energy and drug-binding consistency[22, 24]. As an extensive continued effort to develop the potential inhibitors targeting ABCG2 in therapy of breast cancer, we explored Drugbank data base to examine the molecular binding of substrate and inhibitors with ABCG2 and compared with Imatinib. MD simulation of docked complexes in the bilayer lipid membranes of POPC and MM-GBSA significantly exhibited the underpinning molecular interactions, governing the substrates and inhibitors binding with ABCG2.

Virtual screening

To evaluate the affinities and interactions of various ABCG2 substrates with different inhibitor and inhibitor-substrates, the virtual screening of BCRP/ABCG2 substrates reported in the drug databank (https://go.drugbank.com/categories/DBCAT002663) was performed, taking the cryo-EM structure of ABCG2 bound with STI (imatinib) as a reference [13]. The missing regions of protein were built by AlphaFold and Schrodinger wizard was used to prepare the protein. The LigPrep module was used to prepare the compounds downloaded from drugbank for virtual screening. Glide based virtual screening protocol involves three steps filtering of compounds [15, 16]. The high-throughput virtual screening (HTVS) of chemical library compounds followed by standard precision (SP) and extra precision (XP) results in the selection of best four compounds: Docetaxel, Ozanimod, Pitavastatin, Tezacaftor, having the Glide scores cutoff > -10.00 kcal/mol.

Molecular docking

The analysis of molecular docking results in the identification of crucial residues involved in interaction with substrate and inhibitors of ABCG2. It is important to state that structural details of the molecular interactions of ABCG2 with Docetaxel, Ozanimod, Pitavastatin and Tezacaftor are lacking. As a result, this study sheds light on the crucial interactions that are responsible for interaction of these substrates with the ABCG2 transport channel. A generalized description of all the identified leads and reference compounds are presented in Table 1. The results of docking and the active site residues involved in molecular interactions with compounds are enumerated in Table 2 and depicted in Fig. 1.

Notably, PHE 439 emerges as a crucial residue which is responsible for both substrate as well as inhibitor interactions with ABCG2. Importance of PHE 439 is also highlighted in the available structural data of ABCG2 with various molecules [13]. However, our study also points out that besides PHE 439 there are also some other crucial residues in ABCG2 that are essential for interaction with the transporter, whether it is a substrate or an inhibitor. Such residues include PHE 432, THR 435, ASN 436, SER 440, VAL 546, THR 542, ILE 543, VAL 546 and MET 549. The predominancy of hydrophobic residues in this set means hydrophobic interactions are common in both substrates as well as inhibitors. Interestingly, ARG 482 is one residue that is shown to be exclusively interactive for Tariquidar which acts as inhibitor and is also a good substrate for ABCG2. Furthermore, in the vicinity of ARG 482, there is a cluster of polar residues that surround the active site, pointing at a unique network of residues that sustains in the hydrophobic environment and helps in interaction of tariquidar with ABCG2.

This is essential information because inhibitors that recognize this zone will always tend to be exported by the transporter as well. PHE 431, VAL 401 and LEU 405 are three hydrophobic residues that comprise an important set of interactions for Ko143 and imatinib with ABCG2. Importantly, these residues do not take part in the interaction of Tariquidar with the protein, albeit these are critical for some substrate compounds. LEU 405 is a residue that features in interaction of ozanimod with the protein but is not seen in any other substrate. Further experiments are warranted to evaluate the mechanism of ozanimod translocation by ABCG2.

Table 1

A generalized description of all the identified lead and reference compounds
S. No.	Name	Compound type	Description
1	Docetaxel	Substrate	Docetaxel serves as an antineoplastic agent employed in the treatment of various metastatic and unresectable tumor types. As a more recent addition to the taxane family, it has gained approval, in conjunction with cisplatin, as a first-line treatment for prostate cancer [28]. Furthermore, when combined with oxaliplatin and capecitabine, docetaxel constitutes the third component of the triple-agent TEX regimen, specifically designed for the management of advanced gastric cancer [29]. The primary mode of action of docetaxel involves binding to beta-tubulin, thereby augmenting its proliferation and stabilizing its conformation[28]. This interference disrupts the proper assembly of microtubules into the mitotic spindle, leading to cell cycle arrest during G2/M.
2	Ozanimod	Substrate	Ozanimod is an oral sphingosine 1-phosphate receptor 1 and 5 modulator and frequently used in the treatment of Multiple sclerosis and ulcerative colitis [30]. However, it has been seen to be frequently associated with progression of certain types of cancer [31]. Importantly, it serves as a substrate for both P-glycoprotein as well as ABCG2 [32]
3	Pitavastatin	Substrate	Pitavastatin is used in the management and treatment of hypercholesterolemia and dyslipidemia. Within the statin class, these drugs work to reduce abnormal cholesterol and lipid levels by inhibiting the endogenous production of cholesterol in the liver. Specifically, statins competitively hinder the enzyme hydroxymethylglutaryl-coenzyme A (HMG-CoA) Reductase that catalyzes the conversion of HMG-CoA to mevalonic acid [33]. Involvement of ABCG2 in the biliary excretion of Pitavastatin was demonstrated by Hirano et al. [34].
4	Tezacaftor	Substrate	Alongside Elexacaftor and Ivacaftor, it comprise the First Triple-Combination Cystic Fibrosis Transmembrane Conductance Regulator (CFTR) modulation therapy[35]. It is used to treat both homozygous or heterozygous F508del mutation cystic fibrosis.
5	Tariquidar	Substrate and Inhibitor	Tariquidar is an anthranilamide derivative that was developed as an inhibitor for ABCB1/P-glycoprotein. Later studies identified it as a substrate and inhibitor for ABCG2 as well[11, 36].
6	Imatinib	Inhibitor	Imatinib mesylate, known by the brand names Gleevec and STI571, is a kinase inhibitor that selectively targets Bcr-Abl, activated c-Kit kinases, and platelet-derived growth factor receptor tyrosine kinase. It can effectively reverse ABCG2-mediated resistance to topotecan and SN-38, notably enhancing the accumulation of topotecan specifically in cells expressing functional ABCG2[37]. While certain studies have also pointed out that imatinib also serves as a substrate for ABCG2 [38], the topic has remained a debate for a long period of time, until recently wherein structure of imatinib bound ABCG2 has emerged. The recent data points to a distinctive mode of interaction and action of imatinib compared to the conventional substrates of ABCG2 [13].
7	Ko143	Inhibitor	It is a nontoxic analog of fungal toxin fumitremorgin C and is a potent inhibitor of ABCG2, however recent studies have shown that it can also act as inhibitor for other ABC multidrug transporters [12]. Recent structural studies have shown that it binds to the central, inward-facing cavity of ABCG2, and blocks access for substrates and prevents conformational changes required for ATP hydrolysis [9].

Table 2

Molecular docking scores of compounds with ABCG2, and the crucial residues involve in the molecular binding of compounds at active site of ABCG2.
Compounds	Docking scores (kcal/mol)	Interacting active site residues of ABCG2
Compounds	Docking scores (kcal/mol)	Chain A	Chain B
STI (imatinib)	-9.97	Ile543, Asn436, Leu405, Val401, Ser440, Phe439, Thr435, Phe432, Phe431, Met549, Val546, Thr542	PHE432.B, PHE439.B, THR 435.B, VAL546.B, ASN 436.B, ILE 543.B, MET 549.B
Ko143	-8.08	PHE439.A, THR 538.A, LEU 539.A, THR 542.A, ILE 543.A, VAL 546.A, MET 549.A, LEU 555.A	GLN 398.B, VAL 401.B, LEU 405.B, PHE 431.B, PHE 432.B, THR 435.B, ASN 436.B, PHE 439.B, SER 440.B, VAL 442.B, THR 538.B, MET 549.B,
Tariquidar	-12.14	MET 549.A, VAL 546.A, THR 542.A, PHE 432.A, ASN 436.A, THR 435.A, PHE 439.A, ILE 543.A	GLN 398.B, SER 443.B, VAL 401.B, SER 440.B, MET 549.B, ARG 482.B, THR 542.B, LEU 539.B, ILE 543.B, VAL 546.B, VAL 442.B, PHE 439.B, ASN 436.B, PHE 432.B, THR 435.B
Ozanimod	-10.79	MET 549.A, PHE 432.A, ASN 436.A, THR 435.A, SER 440.A, PHE 439.A, THR 542.A, VAL 546.A, ILE 543.A	LEU 405.B, ASN 436.B, SER 440.B PHE 432.B, THR 435.B, MET 549.B, THR 542.B, ILE 543.B, VAL 546.B, PHE 439.B
Pitavastatin	-10.72	THR 542.A, PHE 439.A, VAL 546.A, MET 549.A,, PHE 431.A, PHE 432.A, THR 435.A, ASN 436.A, SER 440.A, VAL 442.A, THR 538.A	THR 538.B, PHE 439.B, THR 542.B, PHE 432.B, THR 435.B, MET 549.B, LEU 555.B, VAL 546.B, ILE 543.B, LEU 539.B
Docetaxel	-12.21	ILE 543.A, THR 538.A, LEU 539.A, THR 542.A, VAL 442.A, VAL 401.A, GLN 398.A, SER440.A, PHE 432.A, ASN 436.A, THR 435.A, PHE 439.A, MET 549.A, VAL 546.A	THR 538.B, VAL 442.B, THR 542.B, LEU 539.B, ILE 543.B, VAL 546.B, THR 435.B, PHE 432.B, ASN 436.B, SER 440.B, PHE 439.B
Tezacaftor	-12.04	MET 549.A, PHE 439.A, ASN 436.A, THR 542.A, VAL 442.A, SER 443.A, VAL 401.A, GLN 398.A, SER 440.A, PHE 432.A, THR 435.A,	THR 435.B, PHE 439.B, VAL 546.B, THR 38.B, THR 542.B, LEU 539.B, ILE 543.B, THR 435.B

Structural dynamics of protein-ligands

To identify the potential binding interactions of promising candidates, multiple MD simulations were performed on the ABCG2 docked complexes in bilayer lipid membranes of POPC, for the period of 200 ns. The average changes in the structural dynamics of docked complexes were examined by RMSD, Rg and SASA, as are shown in Table 3.

Table 3

Average structural order parameters values of ABCG2-ligand docked complexes and the average distance of ligand from the center of mass (COM) of ABCG2 active site during the evolution of MD simulation (200 ns).
Compounds	RMSD (Å)	Rg (Å)	RMSD (ligand) (Å)	Avg. Dist. of ligands with PHE 439 (Å)
STI (imatinib)	4.03 ± 0.46	37.00 ± 0.20	1.15 ± 0.29	4.27 ± 0.47
Ko143	5.35 ± 0.57	37.12 ± 0.15	1.64 ± 0.29	5.21 ± 0.26
Tariquidar	4.85 ± 0.59	37.39 ± 0.18	1.33 ± 0.64	4.90 ± 0.38
Ozanimod	5.03 ± 0.75	37.60 ± 0.21	0.49 ± 0.15	4.79 ± 0.36
Pitavastatin	5.04 ± 0.70	37.90 ± 0.19	1.27 ± 0.30	7.25 ± 0.82
Docetaxel	5.02 ± 0.78	37.46 ± 0.22	1.29 ± 0.28	5.53 ± 0.27
Tezacaftor	4.54 ± 0.60	37.15 ± 0.18	0.89 ± 0.17	4.75 ± 0.27

The backbone Cα-RMSD plots show that the structure of STI docked complex with ABCG2 optimized to RMSD ~ 3.50Å during the 0–25 ns, however, with the slight drift at ~ 30 ns, the RMSD shifted to ~ 4.0Å. But the structural dynamics remain consistent up to ~ 85 ns. The subsequent change in RMSD can be seen at ~ 90 ns which settles quickly, and the stable equilibrium is maintained up to 200 ns of simulation, at RMSD ~ 4.50 Å (Fig. 2).

The structure of Ko143 docked complex shows gradual increase in RMSD up to 70 ns, with the slight dropdown at ~ 70–75 ns, it attains equilibrium at ~ 75 ns. The stable equilibrium is retained around the RMSD 4.75 Å, till the simulation finished at 200 ns. The structure of ABCG2- Tariquidar achieves equilibrium at ~ 50 ns and observed stable up to 200 ns. ABCG2-Ozanimod gain equilibrium quickly in 0–10 ns, however, remains stable up to ~ 30 ns. Thereafter it shows gradual increase in RMSD up to 100 ns. The structure of ABCG2-Ozanimod optimized around RMSD ~ 5.0 Å and remains consistent up to 200 ns. The structure of ABCG2-Pitavastatin attains equilibrium at ~ 70 and the stable equilibrium is maintained throughout the remaining period of simulation ~ 70–200 ns. The structure of ABCG2-Docetaxel achieves equilibrium quickly in 0–25 ns and remains stable up to 50 ns. Due to slight drift of ~ 1.0 Å, during 50–60 ns, the RMSD shifted to 5.0 Å. It remains stable around RMSD 5.0 Å for the period of 60–125 ns. Further subsequent increase in RMSD seen at ~ 125 ns which gradually settled at ~ 150 ns and the RMSD trajectory remains stable up to 200 ns. The docked complex with Tezacaftor attains equilibrium at ~ 50 ns which is continued up to 125 ns. The drifts in trajectory can be seen during ~ 125–135 ns which settles at ~ 140 ns. After that the system gain stability and remained stable till 200 ns. Thus, the RMSD results suggest that the structures of ABCG2 docked will all seven compounds attains equilibrium during the simulation period of 200 ns. Although the different protein-ligand complexes achieved equilibrium states at the different time span, but the less deviation in RMSD values reveal that the binding of ligands at the active site, alters the structural stability and dynamic behaviour of ABCG2 minimally [39, 40].

Furthermore, to determine the structural compactness[41] of all seven ligands docked complexes of ABCG2, we also computed the radius of gyration (Fig. 3). Results show that all seven complexes gain structural stability around the average Rg value 37.0 Å. The Rg plot of ABCG2-STI complex shows that it attains equilibrium at ~ 50 ns and remained stable throughout the simulation period of 200 ns. The Rg plot of ABCG2-Ko143 settles quickly and observed less perturbed indicating ligand is well accommodated at the active site. The structure of ABCG2-Tariquidar also settles around ~ 50 ns and the stable equilibrium is continued up to 200 ns. The Rg trajectory of ABCG2-Ozanimod adjusted in initial ~ 20 ns and the obtained equilibrium around the average Rg 37.60 ± 0.21 Å is observed unperturbed up to 200 ns. The structure of ABCG2-Pitavastatin gains stability at ~ 20 ns and remained stable with the average Rg value 37.90 ± 0.19 Å till the simulation finished at 200 ns. Whereas the structure of ABCG2-Tezacaftor attains equilibrium around Rg value 37.15 ± 0.18 Å during the initial 0–25 ns and remained stable for 200 ns. The Rg analyses of protein–ligand complexes indicate that the binding of compounds (inhibitor) less alters the structural stability of ABCG2, thereby both substrate and inhibitors chosen here are well occupied at the active site of ABCG2. However, the structural dynamics analysis of protein-ligand complexes using RMSD and Rg provided the global notion of the structure during the evolution of simulation period [42, 43]. Thus, the molecular stability of ligands at binding site of ABCG2 was further checked by calculating the average distance between PHE 439, the crucial residue of active site and ligands (Fig. 4), and the average change RMSD of ligands with respect to initial frame (Fig. 4, Table 3). The results show that average change in RMSD restricted around ~ 1.50 Å, which is revealing the better occupancy of ligands at the active site of ABCG2. Among the seven compounds, Ozanimod shows least movement with average change in RMSD value 0.49 ± 0.15, whereas the higher RMSD (1.64 ± 0.29) is noted for Ko143 which is observed in the permissible range. The compound STI (imatinib) remains fix at the active site around average RMSD 1.15 ± 0.29 Å. It observed consistent during 0-140 ns. The drift of ~ 1.0 Å is seen at ~ 140–150 ns which settle down and the structure of STI was noticed to be stable for the period of 150–200 ns. The structure of Ko143 is optimized quickly around the RMSD 1.50 Å and remains stable up to 60 ns. With the slight drift of 0.50 Å the RMSD reaches to 2.0 Å, however, it remains consistent with RMSD 2.0 Å during 65–200 ns. The initial frames of Tariquidar were observed around the RMSD 0.50 Å which remained consistent up to 25 ns. With the drift of ~ 1.50 Å, RMSD was observed to rise to 2.0 Å, up to 75 ns. The gradual dropdown in trajectory is seen between 75–80 ns and the structure of ligand with RMSD 0.50 Å continued for the period of 80–140 ns. After that the orientation of Tariquidar again flipped and the RMSD shifted to ~ 2.0 Å which continued up to 200 ns. This observation suggests the two distinct postures of Tariquidar during the simulation which remains persistent for the longer time span > 50 ns. The structure of Ozanimod and Docetaxel observed fixed throughout the simulation, around the RMSD values ~ 0.50 Å and ~ 1.50 Å, respectively. Pitavastatin shows initial drifts which settles at ~ 5.0 ns and it remains consistent ~ 1.50 Å throughout the simulation period. Tezacaftor shows perturbation of ~ 0.50 Å during 0-125 ns, however, a steady equilibrium of ligand is observed around the RMSD value ~ 1.00 Å, during ~ 125–200 ns. These results denote that the binding of the ligand to the target protein is relatively stable, during the simulation period[40].

Furthermore, to confirm the spacial binding of ligands at the binding pocket of ABCG2, we also measured the average distance of ligands with reference to the crucial active site residue PHE 439 (Fig. 5, Table 3). The time evolution plot of average distance shows that all compounds stayed encompassed to the active site within 5.0 ± 0.50 Å, except Pitavastatin which shows average distance 7.25 ± 0.82 Å. The reason behind is the structure of Pitavastatin which has two dominant scaffold, one end is dominated by hydroxy group attached with aliphatic chain of enolic acid connected with bulky hydrophobic, fluorophenyl group. Furthermore, the active site of ABCG2 is submerged at the transmembrane regions which is formed with interaction of two monomer chains of ABCG2. The orientation of the active site is elongated, and the upper and lower moieties are governed by the hydrophobic residues. Remarkably the best docked score compounds, Tariquidar and Tezacoftor show stable interactions, having the average distance ~ 4.50 ± 0.50 Å. Moreover, the time evolution plots of average distance for all seven compounds provide evidence that all compounds spatially occupied at the active site of ABCG2, during the simulation period.

Hydrogen bond interactions

To investigate the molecular interaction of ligands at the active site of ABCG2, we also calculated the hydrogen bond (H-bond) interaction of ligands at active site (Fig. 6). The H-bond interactions play a very important role in the determination and specification of ligands as drugs [44, 45]. The time evolution plots of H-bonds show the maximum occupancy of two H-bonds interactions of STI with ABCG2, however, one H-bonds observed consistent throughout the simulation time whereas another H-bonds appeared intermittently. Ko143 shows maximum four H-bonds which is continued up to ~ 150 ns, however, during the last 50 ns of simulation only H-bonds observed consistent. Tariquidar shows maximum occupancy of four H-bonds, out of which three H-bonds observed intact throughout the simulation period 0-200 ns. Ozanimod shows maximum four H-bonds interactions with ABCG2, among them three H-bonds observed consistent throughout the simulation period whereas one is appeared and vanished infrequently till the simulation finished at 200 ns. Pitavastatin shows two H-bond interactions with protein, only two H-bonds interactions remain consistent during the simulation period. Docetaxel shows maximum occupancy of H-bond interactions with ABCG2, among all the ligands. It shows five H-bond interactions, out of that three H-bond interactions observed intact up to ~ 125 ns, however, during the last ~ 75 ns only two H-bond interactions remain stable. Tezacaftor shows four H-bond interactions at the active site of ABCG2. During the initial 0–25 ns, it shows two H-bond interactions which increased up to four H-bond interactions at ~ 25 ns. However, only three H-bond interactions are continued up to 200 ns. The stable H-bond interactions with of ligand observed during the simulation indicates the preferential improved binding of compounds at the active site of ABCG2.

Binding free energy

To determine the molecular stability of compounds at the binding site of ABCG2, the binding free energy of all seven ligands were estimated using MM-GBSA[24]. The binding of compounds at active site of protein is stabilized by the bonded (polar) and non-bonded (van der Waals and electrostatic) molecular interactions. The binding free energy (ΔG_Bind) scores obtained from MM-GBSA include various energy components, including the bonded (polar) and van der Waals (ΔE_vdW) interactions, electrostatic (ΔE_EEL), the energetic term ∆G_SURF define contributing SASA and solvation free energy (ΔG_GB) for ligand interactions at the active site of ABCG2 (Table 4).

Table 4

Binding free energy estimation of compounds using MM-GBSA
Compounds	ΔE_vdW	ΔE_EEL	ΔE_GB	ΔE_SURF	ΔG_BIND
STI (Imatinib)	-73.34 ± 3.40	-18.38 ± 3.91	52.12 ± 3.37	-8.76 ± 0.27	-48.36 ± 3.67
Ko143	-63.41 ± 2.98	-23.04 ± 3.11	43.98 ± 2.52	-7.48 ± 0.27	-49.96 ± 3.17
Tariquidar	-84.29 ± 3.74	-297.01 ± 6.1	332.70 ± 5.16	-10.11 ± 0.28	-59.72 ± 4.20
Ozanimod	-54.07 ± 2.79	-10.02 ± 3.22	29.78 ± 2.53	-6.65 ± 0.30	-40.97 ± 2.83
Pitavastatin	-49.10 ± 2.95	-289.5 ± 12.09	306.96 ± 9.86	-6.28 ± 0.36	-37.93 ± 5.41
Docetaxel	-77.76 ± 3.42	-20.70 ± 3.19	52.12 ± 3.32	-9.32 ± 0.32	-55.67 ± 3.37
Tezacaftor	-66.54 ± 3.30	-22.4 ± 5.17	50.78 ± 2.92	-8.54 ± 0.25	-46.7 ± 4.11

Results show the major contribution of hydrophobic interactions in stabilizing the ligands at binding pocket of ABCG2. Although the compounds, Tariquidar and Pitavastatin show highest energetic contribution of electrostatic interactions, -297.01 ± 6.1 and − 289.5 ± 12.09 kcal/mol. However, the larger value of solvation energies of Tariquidar (ΔE_GB = 332.70 ± 5.16 kcal/mol) and Pitavastatin (ΔE_GB = 306.96 ± 9.86 kcal/mol) diminished the effect of electrostatic interactions. Among the seven compounds, Tariquidar and Docetaxel showed the highest binding free energy (ΔG_BIND) against ABCG2, -59.72 ± 4.20 kcal/mol and − 55.67 ± 3.37 kcal/mol, respectively. The ABCG2 inhibitors, imatinib and Ko143 show marginal differences in total binding energy, having ΔG_BIND values − 48.36 ± 3.67 kcal/mol and − 49.96 ± 3.17 kcal/mol. Results show that van der Waals interaction is the major energy contributor term (-73.34 ± 3.40 kcal/mol) as compared to electrostatic term ΔE_EEL = -18.38 ± 3.91 for the binding stability of STI at the active site of ACBG2. The stabilization of Ko143 at the active site, shows slightly larger contribution of electrostatic term (-23.04 ± 3.11 kcal/mol) and less contribution of van der Waals (-63.41 ± 2.98) as compared to STI. Docetaxel shows larger van der Waals interactions with ΔE_vdW = -77.76 ± 3.42 and the higher ΔG_BIND = -55.67 ± 3.37, plausibly acting as a better substrate against ABCG2. As shown in Fig. 7 the energy contribution terms of active site residues of ABCG2, which provide elegance evidence of the major contribution of PHE 439 in stabilising the ligand during the simulation. Thus, MM-GBSA results suggest that improvising the electrostatic interaction along with van der Waals interactions may enhance the molecular stability of compounds at the active site of ABCG2.

Given the exacerbation of drug resistance phenotypes in cancer cells due to the overexpression of BCRP/ABCG2, there is an imperative need to look for drug molecules that are least plausible to act as substrates or inhibitor molecules that can interact with the protein and lead to conformational changes hat block substrate efflux. Both the requirements can be fulfilled if there is a holistic understanding of molecular interactions that classify substrates and inhibitors. In that aim, we performed an exhaustive assessment of ABCG2 substrates and their interaction capabilities as well as the nature of interactions. Amongst all the screened substrates, we could identify 4 notable substrates, viz., Docetaxel, Ozanimod, Pitavastatin and Tezacaftor to possess the strongest affinity for the substrate binding site. Importantly, structural details of interactions of these compounds also are hitherto unknown. As a result, we provide an essential reference for molecular interactions that result in export of these compounds by ABCG2. Furthermore, we were also able to contrast the interaction of these compounds with known inhibitors of ABCG2. Overall, this study serves as an excellent platform for further experimental investigations to completely characterize the transport and inhibition mechanism of ABCG2 and lays the foundation for pharmacological exploitation.

Conflict of interest:

The authors declare no competing interests.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Author Contribution

Pawan Kumar: Methodology, Software, Formal analysis, Investigation, Visualization, Writing & Editing; Indu Kumari: Methodology, Investigation, Writing & Editing; Rajendra Prasad: Conceptualization, Resources; Shashikant Ray: Writing- Reviewing and Editing; Atanu Banerjee: Conceptualization, Methodology, Validation, Resources, Writing- Reviewing and Editing, Supervision; Amresh Prakash: Conceptualization, Methodology, Validation, Resources, Writing- Original draft preparation, Writing- Reviewing and Editing, and Supervision.

Rees DC, Johnson E, Lewinson O (2009) ABC transporters: the power to change. Nat Rev Mol Cell Biol 10(3):218–227
Begicevic RR, Falasca M (2017) ABC Transporters in Cancer Stem Cells: Beyond Chemoresistance. Int J Mol Sci 18(11)
Banerjee A, Pata J, Chaptal V, Boumendjel A, Falson P, Prasad R (2023) Structure, function, and inhibition of catalytically asymmetric ABC transporters: Lessons from the PDR subfamily. Drug Resist Updat 71:100992
Thomas C, Aller SG, Beis K, Carpenter EP, Chang G, Chen L, Dassa E, Dean M, Van Duong F, Ekiert D et al (2020) Structural and functional diversity calls for a new classification of ABC transporters. FEBS Lett 594(23):3767–3775
Natarajan K, Xie Y, Baer MR, Ross DD (2012) Role of breast cancer resistance protein (BCRP/ABCG2) in cancer drug resistance. Biochem Pharmacol 83(8):1084–1103
Vaidya FU, Sufiyan Chhipa A, Mishra V, Gupta VK, Rawat SG, Kumar A, Pathak C (2022) Molecular and cellular paradigms of multidrug resistance in cancer. Cancer Rep (Hoboken) 5(12):e1291
Taylor NMI, Manolaridis I, Jackson SM, Kowal J, Stahlberg H, Locher KP (2017) Structure of the human multidrug transporter ABCG2. Nature 546(7659):504–509
Yu Q, Ni D, Kowal J, Manolaridis I, Jackson SM, Stahlberg H, Locher KP (2021) Structures of ABCG2 under turnover conditions reveal a key step in the drug transport mechanism. Nat Commun 12(1):4376
Jackson SM, Manolaridis I, Kowal J, Zechner M, Taylor NMI, Bause M, Bauer S, Bartholomaeus R, Bernhardt G, Koenig B et al (2018) Structural basis of small-molecule inhibition of human multidrug transporter ABCG2. Nat Struct Mol Biol 25(4):333–340
Irobalieva RN, Manolaridis I, Jackson SM, Ni D, Pardon E, Stahlberg H, Steyaert J, Locher KP (2023) Structural Basis of the Allosteric Inhibition of Human ABCG2 by Nanobodies. J Mol Biol 435(19):168234
Kannan P, Telu S, Shukla S, Ambudkar SV, Pike VW, Halldin C, Gottesman MM, Innis RB, Hall MD (2011) The specific P-glycoprotein inhibitor Tariquidar is also a substrate and an inhibitor for breast cancer resistance protein (BCRP/ABCG2). ACS Chem Neurosci 2(2):82–89
Weidner LD, Zoghbi SS, Lu S, Shukla S, Ambudkar SV, Pike VW, Mulder J, Gottesman MM, Innis RB, Hall MD (2015) The Inhibitor Ko143 Is Not Specific for ABCG2. J Pharmacol Exp Ther 354(3):384–393
Orlando BJ, Liao M (2020) ABCG2 transports anticancer drugs via a closed-to-open switch. Nat Commun 11(1):2264
Jumper J, Evans R, Pritzel A, Green T, Figurnov M, Ronneberger O, Tunyasuvunakool K, Bates R, Žídek A, Potapenko A et al (2021) Highly accurate protein structure prediction with AlphaFold. Nature 596(7873):583–589
Paquet E, Viktor HL (2013) Macromolecular structure comparison and docking: an algorithmic review. Curr Pharm Des 19(12):2183–2193
Sastry GM, Adzhigirey M, Day T, Annabhimoju R, Sherman W (2013) Protein and ligand preparation: parameters, protocols, and influence on virtual screening enrichments. J Comput Aided Mol Des 27(3):221–234
Pettersen EF, Goddard TD, Huang CC, Couch GS, Greenblatt DM, Meng EC, Ferrin TE (2004) UCSF Chimera–a visualization system for exploratory research and analysis. J Comput Chem 25(13):1605–1612
Laskowski RA, Swindells MB (2011) LigPlot+: multiple ligand-protein interaction diagrams for drug discovery. J Chem Inf Model 51(10):2778–2786
Mikhailovskii O, Izmailov SA, Xue Y, Case DA, Skrynnikov NR (2023) X-ray Crystallography Module in MD Simulation Program Amber 2023. Refining the Models of Protein Crystals. J Chem Inf Model
Case DA, Cheatham TE 3rd, Darden T, Gohlke H, Luo R, Merz KM Jr., Onufriev A, Simmerling C, Wang B, Woods RJ (2005) The Amber biomolecular simulation programs. J Comput Chem 26(16):1668–1688
Singh R, Kumar A, Rane JS, Khan R, Tripathi G, Ajay AK, Prakash A, Ray S (2022) Arylcoumarin perturbs SARS-CoV-2 pathogenesis by targeting the S-protein/ACE2 interaction. Sci Rep 12(1):17038
Mishra CB, Pandey P, Sharma RD, Malik MZ, Mongre RK, Lynn AM, Prasad R, Jeon R, Prakash A (2021) Identifying the natural polyphenol catechin as a multi-targeted agent against SARS-CoV-2 for the plausible therapy of COVID-19: an integrated computational approach. Brief Bioinform 22(2):1346–1360
Khater S, Kumar P, Dasgupta N, Das G, Ray S, Prakash A (2021) Combining SARS-CoV-2 Proofreading Exonuclease and RNA-Dependent RNA Polymerase Inhibitors as a Strategy to Combat COVID-19: A High-Throughput in silico Screening. Front Microbiol 12:647693
Genheden S, Ryde U (2015) The MM/PBSA and MM/GBSA methods to estimate ligand-binding affinities. Expert Opin Drug Discov 10(5):449–461
Hughes JP, Rees S, Kalindjian SB, Philpott KL (2011) Principles of early drug discovery. Br J Pharmacol 162(6):1239–1249
Anderson AC (2012) Structure-based functional design of drugs: from target to lead compound. Methods Mol Biol 823:359–366
Hospital A, Goñi JR, Orozco M, Gelpí JL (2015) Molecular dynamics simulations: advances and applications. Adv Appl Bioinform Chem 8:37–47
Du Q, Jiang G, Li S, Liu Y, Huang Z (2018) Docetaxel increases the risk of severe infections in the treatment of non-small cell lung cancer: a meta-analysis. Oncoscience 5(7–8):220–238
Ostwal V, Bose S, Sirohi B, Poladia B, Sahu A, Bhargava P, Doshi V, Dusane R, Nashikkar C, Shrikhande SV et al (2018) Docetaxel/Oxaliplatin/Capecitabine (TEX) triplet followed by continuation monotherapy in advanced gastric cancer. Indian J Cancer 55(1):88–93
Cree BA, Selmaj KW, Steinman L, Comi G, Bar-Or A, Arnold DL, Hartung HP, Montalbán X, Havrdová EK, Sheffield JK et al (2022) Long-term safety and efficacy of ozanimod in relapsing multiple sclerosis: Up to 5 years of follow-up in the DAYBREAK open-label extension trial. Mult Scler 28(12):1944–1962
Sands BE, Schreiber S, Blumenstein I, Chiorean MV, Ungaro RC, Rubin DT (2023) Clinician's Guide to Using Ozanimod for the Treatment of Ulcerative Colitis. J Crohns Colitis
Baker D, Forte E, Pryce G, Kang AS, James LK, Giovannoni G, Schmierer K (2023) The impact of sphingosine-1-phosphate receptor modulators on COVID-19 and SARS-CoV-2 vaccination. Mult Scler Relat Disord 69:104425
Hoy SM (2017) Pitavastatin: A Review in Hypercholesterolemia. Am J Cardiovasc Drugs 17(2):157–168
Hirano M, Maeda K, Matsushima S, Nozaki Y, Kusuhara H, Sugiyama Y (2005) Involvement of BCRP (ABCG2) in the biliary excretion of pitavastatin. Mol Pharmacol 68(3):800–807
Ridley K, Condren M (2020) Elexacaftor-Tezacaftor-Ivacaftor: The First Triple-Combination Cystic Fibrosis Transmembrane Conductance Regulator Modulating Therapy. J Pediatr Pharmacol Ther 25(3):192–197
Kowal J, Ni D, Jackson SM, Manolaridis I, Stahlberg H, Locher KP (2021) Structural Basis of Drug Recognition by the Multidrug Transporter ABCG2. J Mol Biol 433(13):166980
Houghton PJ, Germain GS, Harwood FC, Schuetz JD, Stewart CF, Buchdunger E, Traxler P (2004) Imatinib mesylate is a potent inhibitor of the ABCG2 (BCRP) transporter and reverses resistance to topotecan and SN-38 in vitro. Cancer Res 64(7):2333–2337
Nakanishi T, Shiozawa K, Hassel BA, Ross DD (2006) Complex interaction of BCRP/ABCG2 and imatinib in BCR-ABL-expressing cells: BCRP-mediated resistance to imatinib is attenuated by imatinib-induced reduction of BCRP expression. Blood 108(2):678–684
Kufareva I, Abagyan R (2012) Methods of protein structure comparison. Methods Mol Biol 857:231–257
Liu X, Shi D, Zhou S, Liu H, Liu H, Yao X (2018) Molecular dynamics simulations and novel drug discovery. Expert Opin Drug Discov 13(1):23–37
Lobanov M, Bogatyreva NS, Galzitskaia OV (2008) [Radius of gyration is indicator of compactness of protein structure]. Mol Biol (Mosk) 42(4):701–706
Hollingsworth SA, Dror RO (2018) Molecular Dynamics Simulation for All. Neuron 99(6):1129–1143
Prakash A, Dixit G, Meena NK, Singh R, Vishwakarma P, Mishra S, Lynn AM (2018) Elucidation of stable intermediates in urea-induced unfolding pathway of human carbonic anhydrase IX. J Biomol Struct Dyn 36(9):2391–2406
Wade RC, Goodford PJ (1989) The role of hydrogen-bonds in drug binding. Prog Clin Biol Res 289:433–444
Chen D, Oezguen N, Urvil P, Ferguson C, Dann SM, Savidge TC (2016) Regulation of protein-ligand binding affinity by hydrogen bond pairing. Sci Adv 2(3):e1501240

No competing interests reported.

Graphicalabstract.jpg
Graphical abstract

Download PDF

Editorial decision: Revision requested
27 Sep, 2024
Reviews received at journal
26 Sep, 2024
Reviewers agreed at journal
26 Sep, 2024
Reviewers agreed at journal
25 Sep, 2024
Reviews received at journal
21 Sep, 2024
Reviewers agreed at journal
19 Sep, 2024
Reviewers agreed at journal
19 Sep, 2024
Reviews received at journal
05 Sep, 2024
Reviewers agreed at journal
03 Sep, 2024
Reviewers agreed at journal
03 Sep, 2024
Reviewers agreed at journal
02 Sep, 2024
Reviewers invited by journal
29 Aug, 2024
Editor assigned by journal
22 Aug, 2024
Submission checks completed at journal
22 Aug, 2024
First submitted to journal
21 Aug, 2024

You are reading this latest preprint version

Elucidating the binding specificity of interactive compounds targeting ATP binding cassette subfamily G member 2 (ABCG2)

Status:

Version 1

Abstract

Figures

Introduction

Materials and methods

Chemical library preparation and virtual screening

Molecular Dynamics Simulation

Binding Free-energy estimation using MM-GBSA

Results and discussion

Virtual screening

Molecular docking

Structural dynamics of protein-ligands

Hydrogen bond interactions

Binding free energy

Conclusions

Declarations

Funding

Author Contribution

References

Additional Declarations

Supplementary Files

Status:

Version 1