CC BY
0INTERNATIONAL SCIENTIFIC AND PRACTICAL CONFERENCE "STATUS AND DEVELOPMENT PROSPECTS OF FUNDAMENTAL AND APPLIED MICROBIOLOGY: THE VIEWPOINT OF YOUNG SCIENTISTS" _25-26 SEPTEMBER, 2024_
EXPLORING TARGET-BASED SCREENING, MOLECULAR DYNAMICS SIMULATION AND PRINCIPAL COMPONENT ANALYSIS FOR DRUG REPURPOSING IN NUT MIDLINE
CARCINOMA
1Ajijur Rehman, 2Mihir Deshnehare, 3Goldi Tiwari, 4Nagendra Kumar Pradhan, 5Ainyo
YL, 6Mohammad Kalim Ahmad Khan
1Assistant Professor, NIMS University Jaipur 2M. Sc. Student, NIMS University Jaipur 3M. Sc. Student, NIMS University Jaipur 4M. Sc Student, NIMS University Jaipur 5M. Sc. Student, NIMS University Jaipur 6Professor, Integral University Lucknow https://doi.org/10.5281/zenodo.13846921
Abstract. NUT Midline Carcinoma (NMC), a particularly malignant and invasive cancer resulting from the fusion of NUT and BRD4 genes, produces a fusion protein that influences cell growth and regulates transcription. Deregulation of this protein has been observed in numerous cancers. The BD1 domain of BRD4-containing proteins has been demonstrated binding to acetylated lysine residues on histone proteins, thereby impacting gene expression and chromatin remodeling. The DrugRep database was used for target-based screening (TBS) of FDA-approved drugs. CurPocket, an integrated tool in DrugRep, was employed to automatically identify potential binding pockets, and docking was performed using AutoDock Vina 1.1.2. SwissADME tools were used to assess pharmacokinetics and drug-likeness attributes. Additionally, the GROMACS and Galaxy platforms were employed for the MD simulations and principal component analysis, respectively. The TBS technique was employed to evaluate 100 drug candidates for human intestinal absorption and blood-brain barrier permeability using the Egan-Egg model. Eighty-four met the filtration criteria and the drug-likeness models narrowed the pool to 48. Toxicophoric parameters, including the PAINS and Brenk alerts, further refined the subset of potential drugs to 39. The lead-likeness criteria identified five drugs, which were evaluated based on binding free energy and hydrogen bond analysis. MD simulation and PCA facilitated more in-depth evaluations, ultimately leading to the selection of ataluren as a potential repurposed NMC drug.
1. Introduction
Dysregulation of BRD4 is increasingly observed in colorectal, breast, pancreatic, and lung
cancers, highlighting its role in cancer development. Tandem bromodomain 1 (BD1) of BRD4 is
critical for binding acetylated lysine residues on histone proteins, contributing to gene expression
and chromatin remodeling [1,2,3]. Bromodomains are integral to DNA replication, cell cycle control, DNA repair, chromatin remodeling, and transcriptional regulation, facilitating the assembly of nuclear macromolecular complexes at specific chromatin sites [4].
Bromodomain-containing proteins are key for the development of potent and specific BRD inhibitors. BETs significantly influence tumorigenesis by regulating the expression of genes essential for tumor growth and survival [5]. However, resistance to BET inhibitors remains a challenge, necessitating new strategies. This study examined the TBS of FDA-approved drugs by
INTERNATIONAL SCIENTIFIC AND PRACTICAL CONFERENCE "STATUS AND DEVELOPMENT PROSPECTS OF FUNDAMENTAL AND APPLIED MICROBIOLOGY: THE VIEWPOINT OF YOUNG SCIENTISTS" _25-26 SEPTEMBER, 2024_
using the DrugRep database (http://cao.labshare.cn/drugrep/). The selection criteria focused on their relevance to BRD4 inhibition and their potential therapeutic significance in NMC. We employed bioinformatics and computational tools, including molecular docking software, such as AutoDock Vina, and dynamic simulation tools, such as GROMACS, to analyze pharmacokinetics, drug-likeness features, molecular docking, and dynamic simulations. This comparative study aimed to evaluate the efficacy and safety of selected drugs targeting BRD4 for NMC treatment. Using these techniques, we thoroughly assessed the pharmacological and structural aspects of these compounds, thereby contributing to a comprehensive understanding of their potential therapeutic value in NMC
2. Methods and Materials
Preparation and Energy Minimization of Target Protein
We obtained the crystal structure of human BRD4 (PDB ID: 7REK) at a resolution of 1.20 A from the RCSB PDB database [6]. To prepare the protein for docking analysis, we extracted the apo form of the protein using SwissPDB viewer, excluding heteroatoms, ions, and additional molecules. Energy minimization was performed to improve the conformation and structural integrity of the target protein. This process involves refining the protein structure to reduce unfavorable interactions and stabilize the molecule. The aim was to obtain a more accurate representation of the native state of the protein for subsequent computational analyses [7,8,9].
Target-based Virtual Screening
We employed the DrugRep online platform (http://cao.labshare.cn/drugrep/) to acquire and process the compound libraries. This platform offers the ability to retrieve and dock compounds from three separate libraries: an approved drug library, an experimental drug library, and a traditional Chinese medicine library. Our primary emphasis was on the approved drug library, which we used for conducting TBS.
ADME Profiling of TBS compounds
The ligands identified by TBS were evaluated based on several ADME parameters, including assessments of Human Intestinal Absorption (HIA) and Blood-Brain Barrier (BBB) permeability [10]. Moreover, drug likeness was evaluated using criteria such as Lipinski's Rule of Five (RO5), Ghose, Veber, Egan, and Muegge parameters [11,12,13,14]. Medicinal chemistry attributes, including PAINS and Brenk alerts, as well as lead-likeness parameters assessed using the SwissADME web tool, were taken into consideration [15,16].
Shortlisting of Drug Candidates using AG and Hydrogen Bond Criteria
We emphasized the role of hydrogen bonds in molecular recognition and binding specificity by utilizing metrics such as AG and hydrogen bond formation to identify potential candidates for ADME-sifted drugs [17].
MD Simulation of BRD4 Docked Complexes with Top Drug Candidates
Molecular dynamics simulations were carried out on complexes formed by the human BRD4 protein, with the top four drugs exhibiting the lowest AG and at least one hydrogen bond. Simulations were executed at 300 K using the MM level within the GROMACS package, version 5.1.2 [18].
Principal Component Analysis
To explore the complex interactions and conformational changes that occur between proteins and drugs, we utilized the analytical tool of Principal Component Analysis (PCA) within the Galaxy platform (https://usegalaxy.eu/). By employing PCA, we were able to identify and
INTERNATIONAL SCIENTIFIC AND PRACTICAL CONFERENCE "STATUS AND DEVELOPMENT PROSPECTS OF FUNDAMENTAL AND APPLIED MICROBIOLOGY: THE VIEWPOINT OF YOUNG SCIENTISTS" 25-26 SEPTEMBER, 2024
highlight statistically significant conformations within the 10 ns trajectory of human BRD4 and selected drug molecules [19,20].
Results and Discussion
Target-Based Virtual Screening Analysis
Using human BRD4 as the template in the DrugRep database, a list of the top 100 compounds based on their binding propensities was generated. This carefully curated dataset was derived from an in-house compilation of FDA-approved libraries comprising a total of 2,315 drugs.
Evaluation of Druglikeness in Compounds Meeting HIA and BBB Permeation Criteria
The SwissADME tool utilizes five distinct sets of pharmaceutical- and biotechnology-based rules, each consisting of various filters that provide qualitative predictions regarding a molecule's potential as an oral drug candidate in terms of bioavailability. These rules encompass Pfizer's Rule of Five (RO5) criteria (molecular weight < 500, hydrogen bond donors < 5, hydrogen bond acceptors < 10, log P < 5) [11], Amgen's Ghose filters (160 < molecular weight < 480, -0.4 < WLogP < 5.6, 40 < molecular refractivity < 130, 20 < atom count < 70) [12], GlaxoSmithKline's Veber filters (rotatable bonds < 10, topological polar surface area < 140) [13], Pharmacia's Egan parameters (WLogP < 5.88, topological polar surface area < 131.6), and Bayer's Muegge filters (200 < molecular weight < 600, -2 < XLogP3 < 5, topological polar surface area < 150, number of carbons > 4, number of heteroatoms > 1, rotatable bonds < 15, hydrogen bond acceptors < 10, hydrogen bond donors < 5) [14]. Among the drugs meeting both the HIA and BBB criteria (84 in total), 48 exhibited no violation of the aforementioned models and rules. Furthermore, the reference molecule (R6S) exhibited one breach in the RO5 criterion (MW>500), four breaches in the Ghose filter (MW>480, WLOGP>5.6, MR>130, atom counts >70), and one breach each in the Egan (WLOGP>5.88) and Muegge (XLOGP3>5) filters.
Free Energy of Binding and Hydrogen Bonding Analysis
Five lead-like compounds, namely estrone, ataluren, perampanel, ergometrine, and exemestane, displayed zero lead-likeness violations. Ataluren exhibited a AG value of -9.5 kcal/mol and engaged with 17 residues through five distinct binding interactions (Vdw, 1 hydrogen bond, CHB, Pi-Alkyl, and Pi-Pi T-shaped). Similarly, estrone showed a AG value of -9.0 kcal/mol, interacting with 12 residues through Vdw, one hydrogen bond, Alkyl, and Pi-Alkyl interactions. Perampanel, ergometrine, and exemestane demonstrated AG values of -8.5 kcal/mol, 8.5 kcal/mol, and 8.2 kcal/mol, respectively. They interacted with 14, 14, and 11 residues, involving Vdw, 1 hydrogen bond, CHB, Pi-Alkyl, and Pi-Sigma interactions for perampanel, five interactions including Vdw, 2 hydrogen bond, CHB, Alkyl, and Pi-Alkyl for ergometrine, and two interactions (Vdw and Alkyl) for exemestane. Furthermore, the control molecule R6S engaged with 18 residues via seven distinct molecular forces having a AG value of -8.1 kcal/mol.
Evaluation of Docked Complex Stability through MD Simulation
The interaction between exemestane and the target protein human BRD4 was found to lack hydrogen bonding. Subsequently, the four remaining drugs, ataluren, estrone, perampanel, and ergometrine, were selected for further molecular dynamics simulation studies, as they exhibited hydrogen bonding during docking. The stability of these complexes was assessed through a 10 ns molecular dynamics simulation using the GROMACS package, with RMSD, RMSF, SASA, AGSolv, Rg, and HBs graphs plotted for analysis [17,18,20] (Figure 1A-F)
INTERNATIONAL SCIENTIFIC AND PRACTICAL CONFERENCE "STATUS AND DEVELOPMENT PROSPECTS OF FUNDAMENTAL AND APPLIED MICROBIOLOGY: THE VIEWPOINT OF YOUNG SCIENTISTS" 25-26 SEPTEMBER, 2024
B
E
F
Figure 1A-F: MD simulation plots for selected parameters: A) RMSD, B) RMSF, C) SASA, D) AGSolv, E) Rg, F) HBs. The magenta, blue, red, and green curves represent BRD4 in complex with ataluren, estrone, perampanel, and ergometrine, respectively.
Principal Component Analysis
Principal component analysis (PCA) was utilized to evaluate and compare the trajectories of the docked complexes involving ataluren, estrone, perampanel, and ergometrine with human BRD4 [19,20]. PCA analysis of ligand-protein docked complexes involves several mathematical steps. Initially, the data is centered by subtracting the mean of each feature from all data points:
Where X represents the original data matrix, and X~ is the mean of each feature. This centered data matrix is then used to calculate the covariance matrix:
Where n is the number of samples and T typically represents the transpose operation. Eigenvalue decomposition is performed on this covariance matrix, yielding eigenvectors V and eigenvalues A:
INTERNATIONAL SCIENTIFIC AND PRACTICAL CONFERENCE "STATUS AND DEVELOPMENT PROSPECTS OF FUNDAMENTAL AND APPLIED MICROBIOLOGY: THE VIEWPOINT OF YOUNG SCIENTISTS" 25-26 SEPTEMBER, 2024
These eigenvectors represent the principal components, which are directions of maximum variance in the data. By projecting the centered data onto these eigenvectors, principal components are obtained:
The proportion of variance explained by PC1 for ataluren (19.
estrone (24%),
perampanel (25.6%), and ergometrine (24.9%) was indicative of the trajectories associated with each compound. PCA plots depicting the principal components are presented in Figures 2A-D. A lower proportion of variance for PC1 suggested a more constrained or stable trajectory. Thus, ataluren, with the lowest proportion of variance (19.6%), appeared to exhibit a relatively more stable interaction compared to estrone, perampanel, and ergometrine, which had higher proportions of variance in PC1.
Figure 2A: PCA plot for ataluren and BRD4 docked complex.
PCI (19.6%) Eigenvalue Rank
Figure 2B: PCA plot for estrone and BRD4 docked complex.
INTERNATIONAL SCIENTIFIC AND PRACTICAL CONFERENCE "STATUS AND DEVELOPMENT PROSPECTS OF FUNDAMENTAL AND APPLIED MICROBIOLOGY: THE VIEWPOINT OF YOUNG SCIENTISTS" 25-26 SEPTEMBER, 2024
PCI (25.591») Eigenvalue Rank
Figure 2C: PCA plot for perampanel and BRD4 docked complex.
Figure 2D: PCA plot for ergometrine and BRD4 docked complex. The investigation demonstrated that ataluren exhibited superior drug-like properties compared to the other compounds examined in the study. It displayed high permeability across the
INTERNATIONAL SCIENTIFIC AND PRACTICAL CONFERENCE "STATUS AND DEVELOPMENT PROSPECTS OF FUNDAMENTAL AND APPLIED MICROBIOLOGY: THE VIEWPOINT OF YOUNG SCIENTISTS" 25-26 SEPTEMBER, 2024
gastrointestinal tract and blood-brain barrier, fulfilled drug-likeness and lead-likeness criteria, and formed stable noncovalent interactions with key amino acid residues. Molecular dynamics simulations corroborated its stability, with favorable RMSD, RMSF, and SASA values, as well as promising Gibbs Free Energy and PCA results. Although some deviations were observed in unsaturation, ataluren met all other critical parameters, positioning it as a potential therapeutic candidate for NMC.
Conclusion
The extensive analysis conducted in this study consistently demonstrated the favorable characteristics of ataluren across a broad range of parameters, thereby establishing its potential as a promising and stable ligand for human BRD4. These findings strongly suggest the need for further research regarding the incorporation of ataluren into drug development initiatives, specifically those targeting human BRD4 and its relevance to NMC. Although compounds such as estrone, perampanel, and ergometrine also exhibit favorable characteristics, the inherent variability in their stability metrics underscores the importance of employing a multifaceted approach when evaluating ligands for target proteins. It is noteworthy that while ataluren has emerged as a promising repurposed drug candidate against NMC based on in silico findings, the necessity for wet-lab validation remains paramount prior to its practical application.
REFERENCES
1. Liu Z, Tian B, Chen H, Wang P, Brasier AR, Zhou J. Discovery of potent and selective BRD4 inhibitors capable of blocking TLR3-induced acute airway inflammation. Eur J Med Chem. 2018 May;151:450-61.
2. Hao J, Yang Z, Wang L, Zhang Y, Shu Y, Jiang L, et al. Downregulation of BRD4 inhibits gallbladder cancer proliferation and metastasis and induces apoptosis via PI3K/AKT pathway. Int J Oncol. 2017 Sep;51(3):823-31.
3. Dhalluin C, Carlson JE, Zeng L, He C, Aggarwal AK, Zhou MM, et al. Structure and ligand of a histone acetyltransferase bromodomain. Nature. 1999 Jun;399(6735):491-6.
4. Sanchez R, Zhou MM. The role of human bromodomains in chromatin biology and gene transcription. Curr Opin Drug Discov Devel. 2009 Sep;12(5):659-65.
5. Picaud S, Wells C, Felletar I, Brotherton D, Martin S, Savitsky P, Diez-Dacal B, Philpott M, Bountra C, Lingard H, Fedorov O. RVX-208, an inhibitor of BET transcriptional regulators with selectivity for the second bromodomain. Proceedings of the National Academy of Sciences. 2013 Dec 3;110(49):19754-9.
6. Ibrahim MAA, Abdelhamid MMH, Abdeljawaad KAA, Abdelrahman AHM, Mekhemer GAH, Sidhom PA, et al. Pyronaridine as a Bromodomain-Containing Protein 4-N-Terminal Bromodomain (BRD4-BD1) Inhibitor: In Silico Database Mining, Molecular Docking, and Molecular Dynamics Simulation. Molecules. 2023;28(15):5713.
7. Morris GM, Goodsell DS, Halliday RS, Huey R, Hart WE, Belew RK, et al. Automated docking using a Lamarckian genetic algorithm and an empirical binding free energy function. J Comput Chem. 1998 Nov 15;19(14):1639-62.
8. Khan FI, Lai D, Anwer R, Azim I, Khan MKA. Identifying novel sphingosine kinase 1 inhibitors as therapeutics against breast cancer. J Enzyme Inhib Med Chem. 2020 Jan 1;35(1):172—86.
INTERNATIONAL SCIENTIFIC AND PRACTICAL CONFERENCE "STATUS AND DEVELOPMENT PROSPECTS OF FUNDAMENTAL AND APPLIED MICROBIOLOGY: THE VIEWPOINT OF YOUNG SCIENTISTS" 25-26 SEPTEMBER, 2024
9. Ahmad KMK, Salman A, Al-Khodairy Salman F, Al-Marshad Feras M, Alshahrani Abdulrahman M, Arif Jamal M. Computational Exploration of Dibenzo [a,l] Pyrene Interaction to DNA and its Bases: Possible Implications to Human Health. Biointerface Res Appl Chem. 2020 Dec 10;11(4):11272-83.
10. Daina A, Michielin O, Zoete V. SwissADME: a free web tool to evaluate pharmacokinetics, drug-likeness and medicinal chemistry friendliness of small molecules. Sci Rep. 2017 May 12;7(1):42717.
11. Lipinski CA. Lead- and drug-like compounds: the rule-of-five revolution. Drug Discov Today Technol. 2004 Dec;1(4):337-41.
12. Ghose AK, Viswanadhan VN, Wendoloski JJ. A Knowledge-Based Approach in Designing Combinatorial or Medicinal Chemistry Libraries for Drug Discovery. 1. A Qualitative and Quantitative Characterization of Known Drug Databases. J Comb Chem. 1999 Jan 12;1(1):55—68.
13. Veber DF, Johnson SR, Cheng HY, Smith BR, Ward KW, Kopple KD. Molecular Properties That Influence the Oral Bioavailability of Drug Candidates. J Med Chem. 2002 Jun 1;45(12):2615-23.
14. Muegge I, Heald SL, Brittelli D. Simple Selection Criteria for Drug-like Chemical Matter. J Med Chem. 2001 Jun 1;44(12):1841-6.
15. Brenk R, Schipani A, James D, Krasowski A, Gilbert IH, Frearson J, et al. Lessons Learnt from Assembling Screening Libraries for Drug Discovery for Neglected Diseases. ChemMedChem. 2008 Mar 14;3(3):435-44.
16. Baell JB, Holloway GA. New Substructure Filters for Removal of Pan Assay Interference Compounds (PAINS) from Screening Libraries and for Their Exclusion in Bioassays. J Med Chem. 2010 Apr 8;53(7):2719-40.
17. Khan MKA, Alouffi S, Ahmad S. Identifying potential inhibitors of C-X-C motif chemokine ligand10 against vitiligo: structure-based virtual screening, molecular dynamics simulation, and principal component analysis. J Biomol Struct Dyn. 2023 Aug 11;1—18.
18. Van Der Spoel D, Lindahl E, Hess B, Groenhof G, Mark AE, Berendsen HJC. GROMACS: Fast, flexible, and free. J Comput Chem. 2005 Dec;26(16):1701-18.
19. Grant BJ, Skj^rven L, Yao XQ. Comparative protein structure analysis with Bio3D-web. Structural Bioinformatics: Methods and Protocols. 2020:15-28.
20. Khan MKA, Ahmad S, Rabbani G, Shahab U, Khan MS. Target-based virtual screening, computational multiscoring docking and molecular dynamics simulation of small molecules as promising drug candidate affecting kinesin-like protein KIFCl. Cell Biochem Funct. 2022 Jul 27;40(5):451-72.
