Overcoming aminoglycoside antibiotic resistance in Mycobacterium tuberculosis by targeting Eis protein

Abstract

Tuberculosis (TB), a major global health concern, even after significant advancements in diagnosis and treatment, causing millions of deaths annually and severely impacting the healthcare systems of developing nations. Moreover, the rise of drug-resistant strains further diminishes the efforts made to control the infection and to overcome this scenario, highly effective drugs are required. Identifying new therapeutic uses of existing drugs through drug repurposing can significantly shorten the time and cost. In the current study, using a computational experimental approach, near about 3104 FDA-approved drugs and active pharmaceutical ingredients from Selleckchem database were screened against Enhanced intracellular survival (Eis) protein, responsible for causing drug resistance by inhibiting the aminoglycoside drug activity. Based on the three-level screening and Molecular Mechanics generalized Born surface area (MM/GBSA) scores, five drugs including Isavuconazonium sulfate, Cefotiam Hexetil Hydrochloride, Enzastaurin (LY317615), Salbutamol sulfate (Albuterol), and Osimertinib (AZD9291) were considered as potential Eis inhibitors. The 500 ns MD simulation results revealed that all these Eis-drug complexes are stable, with minor structural arrangements and stable binding patterns. The PCA and FEL analysis also confirmed the structural stability of the complexes. Overall, these drugs displayed promising results as Eis inhibitors, that can be regarded as suitable candidates for experimental validation.

Introduction

Tuberculosis is a significant public health concern, with approximately 10.6 million new incidents and 1.5 million fatalities reported worldwide in the year 2022 (Bagcchi 2023; Tuberculosis (TB) 2024). Despite the strenuous efforts by the medical community as well as various public health foundations, TB still remains an alarming medical problem, particularly in economically weaker countries (Bloom et al. 2017). The causative agent, Mycobacterium tuberculosis (Mtb), is a resilient pathogen that has evolved mechanisms to evade the body's immune defences and establish persistent infections (Chai et al. 2020). Clinically, TB is present in various forms, amongst which pulmonary tuberculosis is one of the most prevalent forms. However, the disease can also occur in the extrapulmonary sites such as the lymph nodes, bones, and central nervous system (Tobin and Tristram 2024).

The selection of antibiotics for treating bacterial infection has been challenging for the past few decades due to the increase in the case of drug resistibility (Ventola 2015). The arrival of multidrug-resistant (MDR-TB), as well as extensive drug-resistant (XDR-TB) forms of Mtb, has further complicated the treatment and control of the disease, underscoring the need for continued research in addition to the development of novel diagnostic tools, therapeutic strategies, and preventive measures (CDC 2024). Aminoglycosides are broad-spectrum antibiotics used against several Gram-negative bacterial infections including MDR-TB. These antibiotics target the bacterial ribosomes, thereby inhibiting protein synthesis and bacterial survival (Krause et al. 2016). However, Mtb has devised the aminoglycoside resistance mechanism by causing gene mutations, especially point mutations (Doi et al. 2016). This scenario is prominently seen in Mtb's N-acetyltransferase eis genes, leading to aminoglycoside drug resistance. The mutation in the promoter region of the N-acetyltransferase eis gene leads to increased expression of the Eis protein as a result, the responsible of aminoglycoside drugs that include streptomycin, amikacin and kanamycin (Sanz-García et al. 2019). These findings insinuate that Eis protein is a potential drug target and inhibition of the upregulated Eis protein has the potency to revive the aminoglycoside antibiotics activity in a drug-resistant Mtb strain (Zaunbrecher et al. 2009).

Mtb Eis protein has the unusual capability to multi-acetylate amino acids of aminoglycoside drugs. When the protein is upregulated the acetyl group present in the acetyl coenzyme A is transferred to various amino groups of the aminoglycoside drug (Kim et al. 2012). This versatility is enabled by a spacious binding pocket composed primarily of acidic and hydrophobic amino acid residues. The acetylation reaction follows a random sequential mechanism, where the binding of AcCoA or the aminoglycoside substrate with the enzyme causes the inactivation of the aminoglycoside drugs leading to the drug resistance condition (Chen et al. 2011). Two approaches have been considered to overcome the Eis-mediated resistance: (1) chemical modification of existing aminoglycoside drug structures or discovery of a new drug that would not surrender to the Eis-mediated acetylation process. (2) Develop potential inhibitors against Eis that can be used as adjuvants along with clinically employed drugs (Punetha et al. 2020). The first approach was not promising because the protein was versatile. Designing new drugs capable of withstanding the acetylation process can be time-consuming. Therefore, identifying and developing potential Eis inhibitors is more feasible than the first approach (Willby et al. 2016).

Recently, researchers have focused on drug repurposing to expedite the drug development process by re-employing the available FDA-approved drugs to test their efficacy to inhibit the Eis protein beyond their actual function (Kulkarni et al. 2023). Being FDA-approved, this strategy highly reduces the time consumed for examining their safety parameters. Recently, five FDA-approved drugs (Proguanil, Venlafaxine, chloroquine, azelastine and mefloquine) were identified as potential Eis inhibitors, which are under preclinical trials (Pang et al. 2023). However, no other FDA-approved drugs have been identified against the Eis protein. Therefore, this study aims to employs in silico approaches to identify new inhibitors of the Eis protein from the 3,104 FDA-approved drugs collected from Selleckchem database. Further, use the FDA-approved drugs to expedite the drug discovery process. This strategy is advantageous as FDA-approved drugs have already undergone extensive safety evaluations, thus significantly reducing the time required for safety assessments.

Methodology

Target protein structure selection

The experimentally derived mutated structure of the N-acetyltransferase Eis protein of Mtb bound with native inhibitor SGT388 was accessible in the global Protein Data Bank (PDB) archive (Berman et al. 2000). The PBD code for the Eis protein crystal structure was 6VUX (Punetha et al. 2020). The structure consists of a single chain composed of 422 amino acids and represents a mutated form with a resolution of 1.97 Å. Given that mutations in the eis gene are a primary contributor to aminoglycoside drug resistance, selecting this crystal structure is essential for reliable computational experiments. The retrieved Eis protein structure was bound with five small molecules or ligands: The first one is the native inhibitor SGT388 (Ligand ID-H66), the second is the sulfate ion, the third one is glycerol, the fourth one is dimethyl sulfoxide and the fifth one is sodium ion. The unwanted ligands and other elements of crystal structure were removed during the protein preparation phase.

Target protein preparation

The retrieved protein structure was initially visualised using the PyMol software and it was observed that certain amino acids were absent in some regions of the protein chain (DeLano 2002; Sadia et al. 2024). The Swiss-Model homology modelling server was utilised to resolve this issue observed in protein structure (Waterhouse et al. 2018). In the case of remodelling the target Eis protein structure, an appropriate structural template in the databank suitable for filling the gaps present in the protein sequence was used. Thereafter, the target protein sequence was compared with the template. Further, a model was built through which missing residues in the target structure were identified and the gaps were filled according to the sequence on the template structure (Waterhouse et al. 2018). The remodelled crystal structure was subsequently processed using the Protein Preparation Wizard tool provided in the Schrodinger software suite (Madhavi Sastry et al. 2013; Schrödinger Release 2023-1 2023a; Rehman et al. 2023). This tool identified missing hydrogen atoms in the protein structure and integrated them into the Eis protein for a more accurate representation. It removed co-crystallized water molecules and extraneous ligands beyond 5 Å from the binding pocket, and provided accurate metal ionization states for proper charge and force field treatment. Constrained optimization allowed unrestricted minimization of hydrogen atoms while adjusting heavy atoms to alleviate structural strains. The hydrogen bonding pattern was optimized to improve atomic interactions, followed by restrained minimization to converge heavy atoms to an RMSD of 0.3 Å, fully optimizing hydrogen atoms to refine their placement and geometry. This approach eased structural tensions and restored the hydrogen interaction framework.

Ligand library assortment and preparation

A library consisting of 3104 FDA-approved drugs and active pharmaceutical ingredients (API), included in pharmacopoeia (designed for high throughput screening) was downloaded in the SDF format from the Selleckchem database (https://www.selleckchem.com/). The ligands were rapidly prepared using the LigPrep tool for the computational screening process (Schrödinger Release 2023-1, 2023b). Ligprep generates single, low-energy, 3D structures with correct chirality for each ligand. It also creates various compound structures from ligands in different ionization states, tautomeric forms, stereochemical forms, and ring conformations (Nguyen et al. 2021). The ionization states for each ligand were determined using the Epik software tool, which was employed to simulate the ligand properties within a pH range of 7.0 ± 2.0 (Schrödinger Release 2023-1 2023a, c; Johnston et al. 2023). Following this initial step, the OPLS-2005 force field was enforced to optimize the geometry and minimize the potential energy of the ligand structures, thereby refining the molecular models (Shivakumar et al. 2010). Further, these prepared ligands were again filtered by applying Lipinski’s Rule and QikProp filters for screening the compounds based on their drug-likeness and ADME properties prediction (Lipinski 2004; Schrödinger Release 2023-1 2023d). These processes removed the unnecessary examination of the designated failed compounds and improved the properties of identified compounds in order to increase their success rate during clinical trials.

Hierarchical virtual screening

Virtual screening of the prepared drug library was conducted using a hierarchical ranking approach, employing the Virtual Screening Workflow integrated with Glide docking from Schrödinger (Friesner et al. 2004; Schrödinger Release 2023-1 2023e). Before starting the virtual screening process, a receptor grid was constructed via the receptor grid generation module, with the grid box centered around the SGT388 native inhibitor binding site. The centroids were at 19.97 (X), − 11.08 (Y), and 37.66 (Z), with each dimension of the grid box set to 20 Å. The hierarchical virtual screening workflow with Glide involved three docking modes. First, the entire drug library was screened using Glide High Throughput Virtual Screening (HTVS) for speed and initial hit identification. The top 10% of hits from HTVS were then passed to Standard Precision (SP) Docking for balanced speed and accuracy. Finally, the top 10% of hits from SP were screened using Extra Precision (XP) Docking for high accuracy, reducing false positives. The top 10% of hits from XP were selected for continuing the experiment.

Compound selection via MM/GBSA rescoring

The Glide virtual screening generates highly accurate results based on the binding efficacy of small molecules with the receptor protein, yet it still requires validation. According to the literature analysis, the Glide docking score per the empirical scoring functions and the scores calculated do not necessarily support the experimental binding affinities. So, to validate this the Prime-MM/GBSA (Molecular Mechanics/Generalized Born Surface Area) calculation of ligands obtained after the XP docking was performed (Jacobson et al. 2002; Schrödinger Release 2023-1 2023f). The MM/GBSA method uses a force field (OPLS-2005) based algorithm to compute the relative binding enthalpy of the bound and unbound complex. The ranking of compounds based on the MM/GBSA score supports satisfactorily with experimental binding affinity and is better than the empirical score. So, the MM/GBSA method was used as a guiding parameter for the selection of the top ligands by rescoring the compounds screened through the XP docking mode. Based on the rescoring results top five compounds were selected and complexed with the Eis protein for further validation. For the comparative analysis, the co-crystallised inhibitor SGT388 docked with Eis protein was considered as the positive control for this investigation.

Classical molecular dynamics (MD) simulation and simulation trajectory analysis

MD simulation was conducted using the Desmond software component within the Schrodinger suite to investigate the dynamic stability and conformational relationship between the Eis protein and selected drug compounds (Bowers et al. 2006; Schrödinger Release 2023-1 2023g; Naveed et al. 2024). A solvated system containing the TIP4P water model was built by generating an orthorhombic box set at a distance of 10 Å by selecting the buffer as the parameter from the box size calculation method. Moreover, the built system was neutralized as first ions and salts within 20 Å around the ligand were omitted, and second, an adequate number of counter ions were incorporated in the system to restore the system charge balance. Furthermore, the solutions contained 0.15 M of the salt which will bring ions into the simulation box, therefore giving the physiological atmosphere. In addition, the atom type of the OPLS-2005 force field was applied to the whole system. Before the production run, its utilization was reduced by setting the maximum number of iterations to 5000, the convergence criterion to 1.0 kcal/mol Å and a minimum of 10 steps of steepest descent. This pressure was sustained by conforming the time of 0.002 ps for the anisotropic diagonal position scaling. Subsequently, for the production run, NPT was chosen as the ensemble class for the simulation, where the temperature was set to 300 K and pressure to 1.01 bar by maintaining the system density at 1 g/cm³. Finally, the simulation time was set to 500 ns with a recording interval of 10 ps and the relaxation time was set to 1000 ps. Further, the dynamic stability, flexibility and conformational changes experienced by each simulated Eis-drug complex along with major intermolecular interactions present in each complex were analysed using the information present in their simulation pathway.

Binding free energy estimation

To calculate the binding free energy, the conformers of the Eis-drug complex were extracted from the simulation trajectories, especially those formed by the end of simulation time (91–100 ns). For computing this, the molecular mechanics generalized Born surface area (MM/GBSA) method present in the Prime MM/GBSA module present in the Schrodinger platform was utilised during the investigation. Prime MM-GBSA provides energy properties for ligand, receptor, and complex structures (Jacobson et al. 2002; Schrödinger Release 2023-1 2023f). Herein, the energies and energy differences associated with strain and binding, analyzing the contributions from various components of the energy sources. The mathematical equation for binding free energy calculation as follows:

Here, “ΔG_Bind” is the binding free energy of the respective ligands.

Dimensionality reduction through principal component analysis

Principal component analysis (PCA) is one of the most prominent statistical techniques that reduce complex multidimensional datasets by converting the existing variables into new sets of orthogonal variables without losing the required set of information from the data. The principal components (PC) are the variables that describe the greatest amount of variation present in the dataset of the original form. This method aims at determining and quantitating the flexibility changes in the target protein to describe the dynamic behaviour of the complex. All these actions were done in Bio3D R package and scatter plot and scree plot both were obtained for further study (Grant et al. 2006).

Free energy landscape

To enhance comprehension of the changing characteristics of the complexes, an analysis of the FEL was plotted using the first and second PC values retrieved from the PCA results. To generate the FEL plots a Pymol-based plugin software known as Geo measures was employed (DeLano 2002; Kagami et al. 2020). Examining the FEL delivers a complete understanding of the energy distribution and the dynamic transitions between various states or conformations within the molecular complexes under study. This in-depth analysis visually represents the diverse energy states and their inherently dynamic nature, providing meaningful perspectives into the energetic properties and conformational dynamics of the studied systems.

Results

Selection of approved drugs based on virtual screening and MM/GBSA score

Virtual screening has been a successful computational program for predicting major ligand hits and helps in lead optimisation in structure-based drug discovery. In this computational experiment, the 3104 FDA-approved drugs have been screened by passing through, three major screening filters (HTVS, SP and XP). After the intense screening, 10 promising drug candidates were identified with Glide XP docking with values from − 13.19 to − 11.14 kcal/mol (Table 1). Despite the high accuracy rate of Glide screening, there remains a chance that the molecules may lack the desired stability. Therefore, to mitigate the false positives the free-binding energy of these drug molecules was estimated and on the basis of the MM/GBSA scores, the drugs were re-ranked. The top ten drug molecules displayed MM/GBSA scores spanning − 106.59 to − 30.16 kcal/mol (Table 1). Based on the free binding energies, the top five drugs with the highest MM/GBSA scores were selected for further investigation. The top five drug molecules are Isavuconazonium sulfate, Cefotiam Hexetil Hydrochloride, Enzastaurin (LY317615), Salbutamol sulfate (Albuterol), and Osimertinib (AZD9291) (Fig. 1), having MM/GBSA score of − 106.59 kcal/mol, − 69.20 kcal/mol, − 57.34 kcal/mol, − 51.82 kcal/mol and − 49.36 kcal/mol respectively.

Table 1.

List of top ten drugs ranked according to the MM/GBSA scores

Name of the drugs	Docking score (kcal/mol)	MMGBSA ∆G Bind (kcal/mol)
Isavuconazonium sulfate	− 11.479	− 106.59
Cefotiam Hexetil Hydrochloride	− 11.569	− 69.2
Enzastaurin (LY317615)	− 13.134	− 57.34
Salbutamol sulfate (Albuterol)	− 11.69	− 51.82
Osimertinib	− 12.105	− 49.36
Osimertinib mesylate	− 12.105	− 49.36
Lincomycin Hydrochloride Monohydrate	− 12.237	− 47.76
Naringin (Naringoside)	− 12.353	− 45.95
HS-10296	− 11.666	− 35.77
Rutin hydrate	− 13.198	− 30.16

Fig. 1.

The 2D and 3D structures of the selected FDA drug molecules a Eis-Isavuconazonium sulfate, b Eis-Cefotiam Hexetil Hydrochloride, c Eis-Enzastaurin, d Eis-Salbutamol sulfate, e Eis- Osimertinib.

Interestingly, these compounds were accurately bound to the target protein binding site after the docking process, indicating that these compounds may have the required binding stability for inhibiting the Eis protein. To comprehend the stability of individual protein-ligand complexes at the molecular level, the various types of intermolecular bonds, including hydrogen bonds, ionic associations, and van der Waals forces, established between the Eis protein and ligand atoms during the docking procedure were extensively analyzed and characterized.

Analysing the intermolecular bonds between Eis-drug complexes

Understanding the biological function of macromolecules requires examining the binding characteristics and patterns formed during interactions with themselves or other small molecules. The binding of proteins and small molecules with high affinity and specificity in the Eis-drug complex is a crucial aspect of studying their function at the molecular level. Understanding the molecular characteristics of the binding pattern, the identification and quantification of cooperative forces such as hydrogen bonds, hydrophobic bonds, and van der Waals force along with the presence and interactions formed by the water molecules are some of the crucial factors required to maintain the Eis-drug binding affinity.

This work analyse the Eis protein binding site residues and the 2D interaction between them and the chosen drugs at the molecular level. According to Fig. 2a, b, when the Eis protein complexed with Isavuconazonium sulfate drug, the Glu²⁰³ and another by Glu⁴⁰¹ each form hydrogen bonds with the atoms of the drug. Other than the hydrogen bonds these two residues participate in salt-bridge formation. Asp²⁶ was also seen to form the salt bridge with the ligand atoms. Moreover, the Phe²⁴ and the Trp³⁶ residues contributed to the pi-pi stacking bond formation. Similarly, in the Eis-Cefotiam Hexetil Hydrochloride complex, the presence of three hydrogen bonds formed by Leu⁸², Phe⁸⁴ and Phe⁴⁰², while the Trp³⁶ and Glu²⁰³ residues interacted with atoms to form pi-pi stacking and salt bride respectively (Fig. 2c, d). During the analysis of Fig. 2e, f, it was observed that Ile28 residue participates in two hydrogen bond formations in the Eis-Enzastaurin complex. Moreover, in this complex Phe²⁴, Asp²⁶ and Glu⁴⁰¹ residues formed salt bridges with the ligand and Phe⁸⁴ formed the pi-pi stacking interaction. In the Eis-Salbutamol sulfate complex, the amino acid residues Ile²⁸ and Trp³⁶ were involved in the formation of hydrogen bonding interactions. Furthermore, Trp³⁶ contributed to the establishment of pi-pi stacking along with the pi-cation interactions, which are crucial stabilizing forces within the complex (Fig. 2g, h). However, in the Eis-Osimertinib complex, there were a total of four hydrogen bonds formed by Asp²⁶, Ile²⁸, His¹¹⁹ and Glu⁴⁰¹. Also, in this complex Phe⁸⁴ and Phe⁴⁰² are engaged in pi-pi stacking and salt bridge formation respectively (Fig. 2i, j). The control complex, Eis-SGT388, indicated the formation of two hydrogen bonds involving the Phe⁸⁴ and Phe⁴⁰² residues. Additionally, multiple pi-pi stacking interactions were seen between the drug and the Trp³⁶ residue (Fig. 2k, l). In addition to the previously described molecular interactions, the binding interface displayed a combination of hydrophobic contacts and polar interactions displayed by the Eis-drug complexes (Table 2). The presence of weaker bonds including hydrogen and hydrophobic interactions are the major sources for stabilising the docked drugs in the binding pocket of the Eis protein. The presence of these bonds improves the binding affinity and enhances the drug efficacy during discovery. Considering the 2D protein-ligand interaction diagram of the docked complexes, the complex formed by the binding of Osimertinib with Eis protein revealed superior stability owing to the maximum amount of hydrogen bonds relative to other Eis-drug complexes and the control complex. Furthermore, all the docked complexes displayed at least two bonds of hydrogen along with hydrophobic interactions. Additionally, the formation of salt bridges in certain complexes contributed to the stabilization of the Eis-drug complex structure during the molecular docking procedure. Collectively, all the selected Eis-drug complexes demonstrated acceptable binding affinity and stability following the molecular docking process.

Fig. 2.

The 3D and 2D Eis-drug interaction representation of a Eis-Isavuconazonium sulfate, b Eis-Cefotiam Hexetil Hydrochloride, c Eis-Enzastaurin, d Eis-Salbutamol sulfate, e Eis-Osimertinib and f Eis- SGT388 (Positive control).

Table 2.

List of Amino acid residues involved in forming major interactions in Eis-drug complex

S. no.	Protein-ligand complex	Hydrogen bond	Hydrophobic interaction	Polar bond	Salt bridge	π-π stacking/π-cation*	Negative	Positive
1	Eis- Isavuconazonium sulfate	Glu203 Glu401	Trp13, Phe17, Phe24, Phe27, Ile28,Ala33, Trp36,Val40, Leu63, Met65, Phe84, Val85, Leu118, Tyr126, Phe400 Phe402	Ser83, Ser121, His119	Asp26 Glu203 Glu401	Phe24 Trp36	Asp26 Glu203 Glu401	Arg37
2	Eis-Cefotiam Hexetil Hydrochloride	Leu82, Phe84 Phe402	Trp13, Phe17, Phe24, Phe27, Ile28,Ala33, Trp36,Val40, Leu63, Met65, Leu82, Phe84, Val85, Leu118, Tyr126, Phe400, Phe402	Ser32, Ser83, His119, Ser121	Glu203	Trp36	Asp26 Glu203 Glu401	Arg37
3	Eis-Enzastaurin	Ile28	Trp13, Phe17, Phe24, Phe27, Ile28,Ala33, Trp36,Val40, Leu63, Met65, Phe84, Val85 Phe402	Ser23 Thr25 Ser32, Ser83 Ser121	Asp26, Glu401	*Phe24 Phe84	Asp26, Glu203 Glu401	Arg37,
4	Eis-Salbutamol sulfate	Ile28, Trp36	Phe27, Ile28,Ala33, Trp36, Leu63, Met65, Phe84, Phe402	Ser32	–	Trp36 *Trp36	Glu199, Glu203	–
5	Eis- Osimertinib	Asp26,Ile28, His119, Glu401	Trp13,Phe17,Ala20,Phe24 Phe27, Ile28,Ala33, Trp36,Val40, Leu63, Met65, Phe84, Leu118, Ala120, Phe402	Ser32, Ser83, His119, Ser121	Phe402	Phe84	Asp26 Glu401	Arg37
6	Eis-SGT388 (Positive control)	Phe84 Phe402	Trp13, Phe17, Phe24 Phe27, Ile28,Ala33, Trp36,Val40, Leu63, Met65, Leu82, Phe84, Val85, Phe402	Ser83, His119, Ser121	–	Trp36	Asp26	Arg37

MD simulation trajectory analysis

Molecular dynamics (MD) simulation analysis offers detailed insights into drug-target interactions at the atomic level, aiding in the identification of potential drug candidates. It reveals the conformational flexibility of targets and compounds, crucial for understanding drug-induced structural changes. This computational technique allows for the analysis of simulation trajectories, showing the system's changes over time and studying protein-ligand interactions. MD simulations are essential in drug discovery, accelerating the identification of novel therapeutic targets. The simulation trajectory is used to compare system stability and conformational changes via RMSD measurements, while protein and ligand flexibility is assessed through RMSF measurements. Additionally, properties such as Radius of Gyration and Solvent Accessible Surface Area help determine conformational changes

Analysing the structural stability through RMSD calculation

The RMSD measures the structural deviation of the Eis-drug complex by measuring the spatial difference between the initial frame and all the subsequent frames. The RMSD values observed from the plot of RMSD define the convergence and structural stability throughout the simulation. In this computational experiment, MD simulation for 500 ns was performed and the simulation trajectories of each Eis-drug complex were examined against the Eis-SGT388 complex. The RMSD analysis of the Eis protein in the isavuconazonium sulfate complex system revealed a value of 2.5 Å, suggesting that the protein maintained structural stability and did not undergo significant conformational changes throughout the simulation. However, the Isavuconazonium sulfate ligand displayed a deviation of around 7.5 Å after 25 ns. It exhibited minor fluctuations without any significant change in the RMSD value till the end of the simulation. This suggests that the Isavuconazonium sulfate drug has converged during the simulation and remains in the stable binding pose without any large conformational changes throughout the simulation. The Eis-Isavuconazonium sulfate complex has reached an equilibrium state with no significant conformational changes and exhibited structural stability despite the relatively high ligand RMSD (Fig. 3a). Likewise, in the Eis-Cefotiam Hexetil Hydrochloride complex, the Eis protein displayed structural stability throughout the simulation, as its RMSD value was around 2.5 Å, while the Cefotiam Hexetil Hydrochloride showed the RMSD around 5 Å. This indicates that the Cefotiam Hexetil Hydrochloride maintain a stable interaction with the Eis protein regardless of minor, consistent fluctuations that reach up to 6 Å. This confirms that the Eis-Cefotiam Hexetil Hydrochloride complex exhibited effective binding affinity and stability over the span of 500 ns simulation (Fig. 3b). The Eis-Enzastaurin complex's RMSD analysis revealed that the Eis protein exhibited minor fluctuations at 150 ns and 350 ns, but by the conclusion of the simulation, the value levelled off at approximately 3 Å, suggesting the overall structural stability of the Eis protein within the complex. In contrast, the Enzastaurin drug exhibited more substantial deviations, with its RMSD value reaching 5–7 Å during the initial 50 ns and then gradually decreasing to 4–3.5 Å. Another notable deviation of 6 Å was observed between 200 and 250 ns, but after 250 ns, the RMSD value of the ligand stabilized around 3.5–4 Å until the end of the simulation. This suggests that during the initial phase, the drug might not have been binding in the active site, but as the simulation progressed, the RMSD value stabilized, confirming that the ligand had attained the required stability and the Eis-Enzastaurin complex had reached a stable structural conformation with no significant further changes (Fig. 3c). In the Eis-Salbutamol sulfate complex, the protein remained stable during whole simulation process displaying an RMSD value of 2.5 Å, while the Salbutamol sulfate drug showed insignificant fluctuations at 50 ns (6.5 Å), a considerable deviation of 7.5 Å at 150 ns and 6.5 Å at 175 ns. After, 175 ns the Salbutamol sulfate maintained a consistent deviation of 5.5 Å with no significant fluctuations. The Eis-Salbutamol sulfate complex achieved stability as the simulation reached convergence and maintained a stable binding pose throughout the simulation (Fig. 3d). The RMSD plot of the Eis-Osimertinib complex, showed that the Eis protein was structurally stable, as the protein RMSD value was 2.5 Å without any deviations. The Osimertinib ligand in this complex exhibited only minor deviations from its equilibrium state, maintaining a stable configuration with a relatively low RMSD value of approximately 5 Å, indicating a structural consistency throughout the analysis. This indicates that the ligand was present in the binding site throughout the simulation, confirming that the ligand attained a stable confirmation. Overall, the Eis-Osimertinib complex was stable and consistent, maintaining the equilibrium state and the simulation converging (Fig. 3e). Lastly, the Eis-SGT388 complex (control) exhibited structural stability. The protein maintained a consistent conformation, as evidenced by an RMSD value of 2.5 Å. Furthermore, the SGT388 ligand demonstrated only minor deviations, shifting from an initial displacement of 5.5 Å to approximately 4 Å by the end of the simulation. These observations indicate that the overall complex was structurally stable throughout the study (Fig. 3f). The analysis of the RMSD plots of each Eis-drug complex confirmed that during the 500ns simulation, all these docked complexes showed structural stability. The ligands converge to stable conformation and the Eis protein also remains stable state throughout the simulation. The binding of the selected drugs does not lead to any significant structural conformation changes in the protein indicating that the complexes have reached an acceptable equilibrium state during the simulation, making these drugs potential candidates for further studies.

Fig. 3.

The RMSD plot of a Eis-Isavuconazonium sulfate, b Eis-Cefotiam Hexetil Hydrochloride, c Eis-Enzastaurin, d Eis-Salbutamol sulfate, e Eis- Osimertinib and f Eis- SGT388 (Positive control).

Analysing the structural flexibility of protein and selected drugs through RMSF calculation

The RMSF analysis primarily focuses on calculating the flexibility of residues and atoms to characterize the changes in the protein and ligand at the local level. The deviation of each protein residue or ligand atom from its average position was calculated throughout the simulation. The RMSF analysis provides a deeper understanding of the structural stability of different regions of the biomolecule. In the Eis-Isavuconazonium sulfate complex, the Eis protein residues between 140–160 and 200–220 of the Eis protein showed insignificant fluctuations of 3.5 Å. The peaks in these regions indicate that they are flexible regions of the protein, but this flexibility does not lead to the potential instability of the protein structure and instead facilitates ligand binding within the binding site (Fig. 4a). Similarly, in the Eis-Cefotiam Hexetil Hydrochloride complex, residues between 40–60 and 200–220 fluctuated and displayed an RMSF value of 4.5 Å and 4.0 Å, respectively (Fig. 4b). In the case of the Eis-Enzastaurin complex, a significant peak of 4.0 Å was observed in the 200–220 residual regions of the Eis protein (Fig. 4c). Interestingly, the Eis protein in the Eis-Salbutamol sulfate complex showed significant fluctuations of 4.5 Å, 4.0 Å, and 5.0 Å by the residues present in 40–60, 140–160, and 200–220, respectively (Fig. 4d). Furthermore, in the Eis-Osimertinib complex, the Eis protein residues between 140 and 160 fluctuated up to 5.5 Å, and residues between 40–60 and 200–220 also fluctuated, showcasing an RMSF value of more than 3 Å (Fig. 4e). In the control complex, the Eis protein residues within the regions 40–60, 140–160, and 200–220 of the Eis protein exhibited flexibility across all simulated complexes. Notably, residues 140–160 showed fluctuations around 4 Å (Fig. 4f). The observed variations were considered negligible, as they did not significantly alter the protein's structural conformation. This finding corroborates the conclusion that the Eis protein exhibited consistent stability across all the simulated complexes.

Fig. 4.

The Eis protein RMSF plot of a Eis-Isavuconazonium sulfate, b Eis-Cefotiam Hexetil Hydrochloride, c Eis-Enzastaurin, d Eis-Salbutamol sulfate, e Eis-Osimertinib and f Eis- SGT388(Positive control).

The RMSF analysis of each drug, following the protein RMSF analysis, provided the information about the flexibility and stability of the ligand structures. Isavuconazonium sulfate exhibited an overall RMSF of approximately 3 Å, with atoms 21–24 fluctuating up to 4 Å. This suggests a stable conformation during the simulation, indicating strong binding to the Eis protein (Fig. 5a). Cefotiam Hexetil Hydrochloride showed fluctuations reaching 3.5 Å beyond atom 24, signifying flexibility that likely contributes to its adaptation and strong binding interaction with the protein (Fig. 5b). Enzastaurin displayed minimal fluctuations, with atoms up to position 32 exhibiting an RMSF of 2 Å and atoms 33–38 showing a minor fluctuation of 3 Å, indicating a stable conformation (Fig. 5c). Salbutamol sulfate also remained within acceptable fluctuation ranges, with atoms 9–12 deviating up to 3.8 Å, suggesting a stable conformational state and strong binding affinity to the Eis proteins (Fig. 5d). Osimertinib exhibited minor fluctuations reaching up to 3.5 Å (Fig. 5f). In contrast, the control ligand SGT388 showed significant fluctuations, with atoms 10–16 fluctuating between 3 and 4.5 Å (Fig. 5f). These findings collectively indicate that the selected drug compounds displayed relatively modest atomic fluctuations, in contrast to the ligand SGT388, suggesting greater stability and potentially enhanced binding strength for the Eis protein. This suggests that the selected drugs may be more effective and reliable therapeutic candidates for inhibiting the Eis protein.

Fig. 5.

The ligand RMSF plot of a Isavuconazonium sulfate, b Cefotiam Hexetil Hydrochloride, c Enzastaurin, d Salbutamol sulfate, e Osimertinib and f SGT388(Positive control).

Analysing the Solvent accessible surface area (SASA) of the ligands

The SASA represents a critical metric for evaluating the solvent-accessible surface area of a drug molecule. This metric provides valuable insight into the degree of ligand exposure to the solvent and the extent to which the ligand is enclosed within the binding site of the protein. In the case of the drug Isavuconazonium sulfate, the SASA value was around 420 Å² until 200 ns of the simulation. Subsequently, the value decreased to a range of 300–250 Å² from 210 to 410 ns, and during the last 50 ns, it remained around 400 Å² (Fig. 6a). For Cefotiam Hexetil Hydrochloride, the SASA value was approximately 300 Å² until 200 ns, which reduced to less than 200 Å² (Fig. 6b). In the last 100–50 ns, the SASA value increased from 250 to 350 Å². The SASA analysis of Enzastaurin showed a consistent value of 200 Å² (Fig. 6c). Salbutamol sulfate bound to the Eis protein had a SASA exceeding 200 Å² at 50 ns, subsequently decreasing to less than 100 Ų by the end of the simulation (Fig. 6d). The SASA for Osimertinib was 300 Å² until 350 ns, then decreased to less than 200 Ų (Fig. 6e). For the control compound, the SASA was less than 300 Ų during the first 100 ns and then reduced to less than 200 Ų till the end (Fig. 6f). The SASA plot analysis revealed that Salbutamol sulfate had the lowest solvent exposure among the drugs examined, comparable to the control ligand. A significant portion of the drug was found to be buried within the binding site of the target protein. The Enzastaurin compound displayed diminished solvent exposure and maintained its adherence with the Eis protein over the course of the simulation duration, which is likely attributable to its stable and consistent solvent interaction profile. The Osimertinib drug also remained bounded with Eis protein with insignificant solvent exposure. However, Isavuconazonium sulfate and Cefotiam Hexetil Hydrochloride drugs were significantly exposed to the solvent compared to the other three drugs and control molecule, but they remained bounded to Eis protein in an acceptable binding pattern.

Fig. 6.

The SASA plot of a Isavuconazonium sulfate, b Cefotiam Hexetil Hydrochloride, c Enzastaurin, d Salbutamol sulfate, e Osimertinib and f SGT388(Positive control).

Analysing the compactness of ligand

The potential binding efficacy and the compactness of the ligand in the simulated complex are also assessed by estimating the radius of gyration (rGyr). This provides a foresight into the spatial distribution and compactness of the atoms from the ligand’s centre mass. In this investigation, the rGyr plots of each selected drug were studied and compared with the control ligand to understand the compactness and structural changes of each drug when bonded with Eis protein. Herein, the Isavuconazonium sulfate ligand significant fluctuation of more than 6.0 Å was observed at various time points (150 ns, 200 ns, and 500 ns). Apart from these major fluctuations, the minor fluctuations were below 6 Å. This indicates that the drug was less compact and has undergone conformational flexibility to improve the binding efficacy of the ligand (Fig. 7a). Similarly, the Cefotiam Hexetil Hydrochloride displayed minor fluctuations and remained consistent with a rGyr value of 6.25 Å till 350 ns. However, after 350 ns, the rGyr value further reduced to 6 Å until the simulation was completed, indicating that the drug was in a compact state with no significant conformational flexibility (Fig. 7b). The Enzastaurin drug also showed minor fluctuations, and the rGyr value was less than 5.4 Å, indicating that the drug maintains a consistent conformation and compactness (Fig. 7c). Interestingly, the rGyr value of Salbutamol sulfate was around 3.8 Å, suggesting it is a highly compact structure (Fig. 7d). The Osimertinib drug demonstrated a rGyr of less than 5.25 Å up to 325 ns of the simulation, with the value further decreasing to 4.75 Å by the end of the simulation. This finding suggests that the drug exhibited an acceptable level of compactness and maintained a stable conformation throughout the simulated timeframe (Fig. 7e). The control compound SGT388 demonstrated a notable rGyr value exceeding 6 Å between 50 and 100 nanoseconds, but this decreased to 4.5 Å after 110 nanoseconds and maintained a stable, compact structural configuration for the remainder of the simulation period (Fig. 7f). Based on the radius of gyration (rGyr) analysis, it was observed that Salbutamol sulfate exhibited the highest level of compactness among the selected drugs. Additionally, it maintained a stable conformation throughout the study, outperforming both the control and other selected drugs in terms of consistency.

Fig. 7.

The radius of gyration plot of a Isavuconazonium sulfate, b Cefotiam Hexetil Hydrochloride, c Enzastaurin, d Salbutamol sulfate, e Osimertinib and f SGT388(Positive control).

Analysing the protein-ligand interaction

Examination of the segmented column chart depicting the Eis-drug interactions offered valuable insights into the binding affinity and durability of the selected pharmacological agents when interacting with the protein over the course of the simulation. The segmented column chart predominantly represented the various molecular forces present between the Eis protein and the selected drugs during the simulation. Each bar segment represents specific interactions that include hydrogen bonds, hydrophobic interaction, ionic bonds and water bridges, corresponding to the protein residues responsible for these bond formations. The segmented column chart study of the Eis-Isavuconazonium sulfate complex revealed that the Glu¹²² and Glu⁴⁰¹ residues played a significant role in hydrogen bond formation, with the Glu¹²² residue involved for 90% and the Glu⁴⁰¹ residue involved for 65% of the total simulation period. Also, in this complex, Ser²⁴ for 95%, Trp³⁶ for 75% and Phe⁸⁴ for 60% of the total simulation time predominantly participated in hydrophobic bond formation, while, Phe⁴⁰² residues contacted with the ligand to form the water bridge for 100% of the overall simulation duration (Fig. 8a). Likewise, in the case of the Eis-Cefotiam Hexetil Hydrochloride complex, Leu⁸², Val⁸⁵ and Glu²² displayed hydrogen bonding for 75%, 95% and 85% of the 500 ns simulation, respectively. Moreover, the Phe⁴⁰² and Ser⁸³ residues interacted with atoms via ionic bonds for 90% and 70% of the total simulation period. The Phe⁴⁰² residue also contributed to the water bridge and maintained the interaction for 100 % of the entire 500 ns simulation (Fig. 8b). Similarly, in the Eis-Enzastaurin complex, the Ile²⁸ residue formed the hydrogen bond for 98% of the simulation time, whereas the Asp²⁶ and Phe⁴⁰² residues contributed to the water bridge (60% of the interaction fraction) and ionic bond (35 % of the interaction fraction) formations, respectively. Moreover, the Phe²⁴, Trp³⁶ and Phe⁸⁴ residues were involved in the hydrophobic interactions for 100%, 75% and 80% of the 500 ns run time (Fig. 8c). In the segmented column chart of the Eis-Salbutamol sulfate complex, significant hydrogen bonds were established by the Glu¹⁹⁹, Glu²⁰³ and Glu⁴⁰¹ for 30%, 55% and 35% of the simulation time. Additionally, Phe²⁴, Phe²⁷, Trp³⁶ and Phe⁸⁴ residues interact with atoms through hydrophobic interaction for 75%, 55%, 90% and 78% of the total simulation run time (Fig. 8d). Lastly, in the Eis- Osimertinib hydrophobic bonds were predominately present and residues Phe²⁴ for 55%, Phe²⁷ of 75%, Trp³⁶ for 100 % and Phe⁸⁴ for 85% of the 500 ns run time. Other than the hydrophobic bond, Glu¹⁹⁹ (25 %), Glu⁴⁰¹ and Phe⁴⁰² (50%) residues participated in the water bridge formation (Fig. 8e). The stacked column chart of the Eis-SGT388 control complex also demonstrated a substantial presence of hydrophobic bonds involving the Phe²⁴, Trp³⁶, and Phe⁸⁴, which were observed for approximately 80% of the total simulation duration. The Glu⁴⁰¹ and Phe⁴⁰² residues participated in the water bridge formation for 90% and 40% of the simulation time (Fig. 8f). The segmented column chart evaluation of protein-ligand interactions demonstrated that the stability of the Eis-drug complexes was primarily maintained by the establishment of hydrogen interaction along with the development of hydrophobic bonds throughout the simulation. Relative to the Eis-SGT388 complex, the Eis-drug complexes exhibited significantly greater stability, which was attributed to the presence of hydrogen bonds and hydrophobic bonds involving multiple residues over extended interaction periods.

Fig. 8.

The segmented bar chart of a Eis-Isavuconazonium sulfate, b Eis-Cefotiam Hexetil Hydrochloride, c Eis-Enzastaurin, d Eis-Salbutamol sulfate, e Eis-Osimertinib and f Eis- SGT388(Positive control).

Binding free energy estimation through MM/GBSA calculations

Estimating binding free energy provided insight into the thermodynamical stability and distribution of binding free energy in the biomolecular complex. The MM/GBSA calculation aids in understanding the binding affinity of the drug molecules when bound to the target site, which is necessary for assessing drug efficacy. Moreover, the total binding energy can be decomposed into various components including van der Waals interaction, electrostatic interactions and solvation effects. The MM/GBSA calculation divulged that the Eis-Isavuconazonium complex displayed the highest binding affinity having a ΔG_Bind value of − 94.89 ± 5.91 kcal/mol. Likewise, the Eis-Cefotiam Hexetil Hydrochloride complexes also exhibited a strong binding affinity with ΔG_Bind of − 93.36 ± 8.47 kcal/mol. Moreover, these values were significantly lower than the ΔG_Bind of the Eis-SGT388, which was 86.53 ± 3.71 kcal/mol. Although, the Eis-Enzastaurin complex (ΔG_Bind = − 80.78 ± 4.14 kcal/mol), and Eis-Osimertinib complex (ΔG_Bind = − 66.25 ± 4.17 kcal/mol) had higher binding free energy than the Eis-SGT388 (control), the obtained binding free energy were within the admissible range for the stable Eis-drug interactions. However, the Eis-Salbutamol Sulfate complex exhibited the lowest binding free energy with a ΔG_Bind of − 43.42 ± 4.09 kcal/mol, indicating a comparatively weaker binding affinity. Based on the values obtained, the Eis-Isavuconazonium and Eis-Cefotiam Hexetil Hydrochloride complexes exhibited strong binding affinity compared to the other drugs. Also, the Eis-Enzastaurin and Eis-Osimertinib complexes showed acceptable binding energy with stable binding affinities. Meanwhile, the Eis-Salbutamol Sulfate complex, with its higher binding free energy, may be less effective in forming a stable complex with the target protein. Additionally, the energy distribution components analysis displayed that the ΔG_{Bind Lipo} (Lipophilic) and ΔG_{Bind vdW} (van der Waals interaction) made the most significant contribution to the stabilization of the Eis-drug complexes (Table 3 and Fig. 9).

Table 3.

Energy was obtained for each component after MM/GBSA calculation.

Energy distribution components	Eis- Isavuconazonium sulfate	Eis- Cefotiam Hexetil Hydrochloride	Eis- Enzastaurin	Eis-Salbutamol sulfate	Eis- Osimertinib	Eis-SGT388 (positive control)
ΔGBind	− 94.89 ± 5.91	− 93.36 ± 8.47	− 80.78 ± 4.14	− 43.42 ± 4.09	− 75.19 ± 3.59	− 86.53 ± 3.71
ΔGBind Coulomb	− 23.04 ± 38.40	− 47.04 ± 11.45	− 18.69 ± 11.68	2.19 ± 13.46	− 23.99 ± 6.76	− 18.01 ± 10.23
ΔGBind Covalent	1.06 ± 1.44	0.04 ± 3.57	2.20 ± 1.11	0.99 ± 0.93	2.91 ± 0.67	3.52 ± 0.82
ΔGBind Hbond	− 1.59 ± 0.49	− 2.51 ± 0.40	− 0.5 ± 0.02	− 0.85 ± 0.19	− 0.56 ± 0.36	− 1.16 ± 0.21
ΔGBind Lipo	− 26.97 ± 2.01	− 28.48 ± 2.46	− 26.03 ± 1.79	− 17.78 ± 1.80	− 27.81 ± 1.28	− 31.93 ± 1.68
ΔGBind Packing	− 6.07 ± 0.47	− 0.65 ± 0.60	− 6.41 ± 1.02	− 0.52 ± 0.16	− 12.60 ± 1.16	− 9.38 ± 1.40
ΔGBind Solv GB	− 23.15 ± 37.66	54.65 ± 10.25	31.44 ± 11.10	4.44 ± 13.23	49.95 ± 6.26	26.24 ± 10.18
ΔGBind vdW	− 61.20 ± 3.50	− 68.89 ± 5.45	− 62.75 ± 3.64	− 31.88 ± 2.02	− 63.07 ± 2.73	− 55.81 ± 2.21

Fig. 9.

The energy distribution plot of a Eis-Isavuconazonium sulfate, b Eis-Cefotiam Hexetil Hydrochloride, c Eis-Enzastaurin, d Eis-Salbutamol sulfate, e Eis-Osimertinib and f Eis- SGT388 (Positive control) generated through MM/GBSA calculation.

Principal component analysis (PCA)

PCA is an assessment tool used to decrease the number of variables in the dataset obtained from the simulation trajectory while retaining the maximum number of variables. The covariance matrix of the atomic fluctuations experienced by the protein-ligand complex was computed, wherein the matrix accumulates the correlated atomic movements in the system. Further, the eigenvalues and eigenvectors from the matric are calculated, subsequently, the eigenvectors are arranged according to their eigenvalues in descending order. The highest eigenvectors represent the principal components (PCs), the new set of variables containing the most significant information in the dimensionally reduced state. The PCA data was visualized by generating the scree plot, scatter plots and free energy landscape.

Scatter plot analysis

The scatter plot visualises the conformational space of the complex during the MD simulation. Three main scatter plots are generated during the PCA; PC1 Vs PC2, PC2 Vs PC3 and PC1 vs PC3. Three distinct clusters were present in the scatter plot. The red colour cluster represents the initial conformational state of the protein-ligand complex followed by the white colour clusters representing the intermediate state and the blue colour cluster represents the final state. The PC1 vs PC2 scatter plot analysis of the Eis-Isavuconazonium sulfate showed separation between the clusters indicating significant conformational changes along these principal components (Fig. 10a1). In contrast, the PC2 vs PC3 plot, exhibited overlapping clusters, suggesting the conformers in these clusters do not undergo significant conformational changes along these components (Fig. 10a2). However, the PC1 vs PC3 plot displayed separated clusters, signifying conformational changes among these principal components (Fig. 10a3). The overall analysis of the scatter plots suggests that the binding of Isavuconazonium sulfate to the Eis protein may have induced structural rearrangements in the complex to attain dynamic stability during the molecular dynamics simulation. In the case of Eis-Cefotiam Hexetil Hydrochloride, the scatter plot PC1 vs PC2, the initial conformers (red) were overlapping with some of the intermediate conformers (white), however, a transition state was observed between these conformers. Moreover, overlapping between the intermediate and final conformers (blue) was observed in the plot. This indicates a significant conformational change in the protein-ligand complex among these components (Fig. 10b1). A similar scenario was observed in the plots PC3 vs PC2 and PC1 vs PC2, where the final conformers are spread to the essential space indicating the presence of conformational change. The scatter plot of Eis-Cefotiam Hexetil Hydrochloride suggests that the complex may have undergone distinct conformational changes, potentially corresponding to different binding modes for stabilization (Fig. 10b2, b3). Likewise, while analysing the Eis-Enzastaurin complex scatter plot, the PC1 vs PC2 plot showed separated clusters suggesting the presence of conformational change among these components (Fig. 10c1). However, in PC3 vs PC2, plot overlapping was observed indicating that conformers with no structural rearrangements were present within these components (Fig. 10c2). The PC1 vs PC3 plot also showed separated clusters denoting the presence of conformational changes (Fig. 10c2). The examination of all the scatterplots of Eis- Enzastaurin indicated a flexible interaction between the receptor and drug, enabling them to achieve a stable conformational state during the simulation. Analysis of the PC1 vs PC2 plot for the Eis-Salbutamol sulfate revealed segregated clusters, indicating the presence of substantial structural changes among these components (Fig. 10d1). On the other hand, the PC3 vs PC2 plot exhibited overlapping of the clusters indicating the existence of multiple minor conformations (Fig. 10d2). The scatter plot of PC1 vs PC3 also showed scatter clusters (Fig. 10d3). These results confirm that the Eis-Salbutamol sulfate complex also has undergone structural rearrangement to reach the stable conformation. Likewise, in the PC1 vs PC2 plot of the Eis-Osimertinib, the cluster of conformers generated at the initial state overlapped partially with the conformers from the intermediate state and some of the conformers from the intermediate stage overlapped with the conformers generated in the final stage (Fig. 10e1). According to this scatter plot a transition between conformers was observed among these components, signifying the presence of structural rearrangement among the conformers. Whereas in the PC3 vs PC2 plot, all the conformers in every stage were overlapped signifying the presence of similar conformers with no major structural rearrangement upon ligand binding among these components (Fig. 10e2). In the PC1 vs PC3 plot, a similar pattern of the clusters as seen in the PC1 vs PC2 was observed confirming the structural rearrangement of conformers present among these components (Fig. 10e3). Altogether in this scatter plot of this complex also, a flexible nature of the interaction was observed during the simulation. The relative evaluation of the Eis-SGT388 complex was subsequently explored through a scatter plot analysis. The results revealed that while the clusters were distinct in the PC1 vs PC2 plot (Fig. 10f1), they exhibited partial overlap in the PC3 vs PC2 and PC1 vs PC3 plots (Fig. 10f2, f3). These observations indicate that the control complex also exhibited minor structural conformational changes to remain in a dynamically stable state. The overall scatter plot analysis of all the complexes indicated that to attain the required stability in the Eis-drug complex a smooth transition between the structural conformations was observed similar to the control complex Eis-SGT388 and this also suggests that upon binding the drugs stabilizes specific conformations of the protein to attain maximum stability.

Fig. 10.

The Scatter plot (1–3) and Scree plot (4) of a Eis-Isavuconazonium sulfate, b Eis-Cefotiam Hexetil Hydrochloride, c Eis-Enzastaurin, d Eis-Salbutamol sulfate, e Eis-Osimertinib and f Eis- SGT388 (Positive control).

Scree plot analysis

Further, the obtained eigenvector and their corresponding eigenvalues were plotted to generate the scree plot and these scree plots were analysed to understand the behaviour of the complexes. During the analysis, it was observed that the “elbow” in the scree plot seems to appear after the four PCs in all the complexes. This suggests that the first three to four PCs capture the majority of the variance in the dataset of each complex. So, the principal components from the initial three dimensions (PC1, PC2 and PC3) were considered in order to determine the total variance accounted for by each complex. So, the total proportion of variance obtained after summing up all the PCs (PC1+ PC2 + PC3) was 27.7% for Eis-Isavuconazonium sulfate, 43.1% for Eis-Cefotiam Hexetil Hydrochloride, 36.2% for Eis-Enzastaurin, 29.8% for Eis-Salbutamol sulfate, 34.7% for Eis-Osimertinib, and 33.7% for Eis-SGT388 (Fig. 10a4, b4, c4, d4, e4, f4). Herein, the first three PCs of Eis-Cefotiam Hexetil Hydrochloride captured the maximum variance, followed by Eis-Enzastaurin, Eis-Osimertinib, Eis-SGT388, Eis-Salbutamol sulfate and Eis-Isavuconazonium sulfate. The higher percentage of variance in Eis-Cefotiam Hexetil Hydrochloride indicates that a large portion of the dynamical conformation of the protein-ligand complex can be described using just these three principal components. The PCs of Eis-Isavuconazonium sulfate and Eis-Salbutamol sulfate captured the least variance due to their low percentage compared to other complexes. This indicates that these complexes have high structural complexity or conformational flexibility. The other two complexes (Eis-Enzastaurin, and Eis-Osimertinib) including the control complex (Eis-SGT388) contain an acceptable amount of information and conformational states.

Analysing the conformational state using free energy landscape (FEL)

The FEL is a type of heat map used for the visualisation and understanding of the conformational landscape and transitions between different states of the complex. FEL provides the visual representation of the energy minima and the conformational pathways. The PC1 and PC2 are used to calculate the free energy and then the 2D plot with PC1 on the X-axis and PC2 on the Y-axis were subsequently created. The colour gradients present in the plot are major components for visualising different free energy levels and aid in the identification of stable and unstable conformation states of the receptor-ligand complex. In the heat plot, low-energy and stable conformational states are depicted by the darker colours, while the lighter colours represent higher-energy and less stable states. The regions with relatively low free energies are designated as basins, indicating the stable conformational states of the Eis-drug complex. The Eis-drug complexes examined through the FEL plot revealed that the most stable conformations possessed Gibbs free energies ranging from 0 to 2 kJ/mol. In contrast, the control complex exhibited a Gibbs free energy between 0 and 1.5 kJ/mol. The basin analysis of the Eis-Isavuconazonium sulfate contains a single basin, however, it was not deeper and the pathway connecting the basins showed that the light blue regions represented the presence of conformers with higher free energy are more spread in the landscape. This indicates that only a few conformers were able to transfer from this barrier to attain maximum stability. So, the plot suggests the complex has experienced substantial structural rearrangements, as demonstrated by the high energy barrier observed (Fig. 11a). Likewise, in the Eis-Cefotiam Hexetil Hydrochloride, three basins were present, and among these three, one showed maximum depth indicating the presence of a more stable conformational state. However, a transition pathway connecting to the dark basin indicated that certain conforms are stable but could not reach the lowest energy point. So, this complex was dynamically stable (Fig. 11b). Similarly, in the Eis-Enzastaurin complex, two relatively low-energy basins in one were deep and the width of the energy barrier was less indicating that the system was likely able to move into a different conformational with much stability. This analysis confirmed that the complex attained dynamic stability during the simulation (Fig. 11c). The FEL plot of the Eis-Salbutamol sulfate complex revealed the presence of two distinct basins, accompanied by a wider transition pathway. This observation suggests that only a limited number of conformers were able to reach the global minimum energy state, indicating that the complex underwent conformational changes (Fig. 11d). In the case of Eis-Osimertinib also one deeper basin and one small basin were observed two basins were present and the transition pathway connecting each basin was present indicating the conformational shift to attain the minimum energy level (Fig. 11e). The free energy landscape analysis of the Eis-SGT388 control complex revealed the presence of two deep energy minima and a metastable transition state within its conformational space (Fig. 11f). This suggests the existence of stable conformational states with global minimum free energy values. Comparative analysis of scree plots, scatter plots, and free energy landscape plots indicated that the Eis-drug complexes underwent conformational changes during the simulation to attain a stable binding state, in contrast to the control complex.

Fig. 11.

The 3D and 2D heatmap of a Eis-Isavuconazonium sulfate, b Eis-Cefotiam Hexetil Hydrochloride, c Eis-Enzastaurin, d Eis-Salbutamol sulfate, e Eis-Osimertinib and f Eis- SGT388(Positive control) generated through FEL analysis

Discussion

With the increase in multi-drug resistance cases, TB management has faced a serious downfall. Also, with a longer treatment period, the pathogen attains resistance towards drugs engaged in treatment. Therefore, the administration of a new drug or combination of drugs was considered to be an effective method to enhance the bioactivity against drug-resistant strains (Navisha et al. 2022). Drug repurposing has been a successful process for exploring the efficacy and usage of existing drugs in a new disease arena. This process has been very successful in pharmaceutical research and development industries, as it reduces the cost of production and allows the researcher to bypass the clinical phase I (Singh et al. 2023). The implementation of computational drug discovery techniques systematically and logically has minimised the risk of failure during drug repurposing (Ekeomodi et al. 2023). Recently, researchers have identified niclosamide and tribromsalan exhibit inhibitory properties against M. tuberculosis and M.abscessus using computational drug repositioning (Yang et al. 2024). Similarly, Atosiban and Rutin were two significant compounds identified using a computational drug repurposing strategy to inhibit the HemD enzyme of Mtb (Sharma et al. 2023). These recent research findings confirm that computational procedure for drug repurposing has become the core framework for TB drug discovery.

Through this computational investigation, we have identified Isavuconazonium sulfate, Cefotiam Hexetil Hydrochloride, Enzastaurin, Salbutamol sulfate, and Osimertinib as promising candidates using advanced computational drug discovery methods to potentially inhibit the Eis protein, which causes aminoglycoside drug resistance in Mtb. The literature analysis confirmed that some identified drugs were significantly repurposed by an in-silico approach for finding therapeutic solutions for other diseases. For instance, Isovuconazonium and Enzastaurin was identified as one of the potential inhibitors of SARS-CoV2 protease (Mendoza-Martinez and Rodriguez-Lezama 2020; Achilonu et al. 2020; Bembenek 2020). Enzastaurin was also identified as an inhibitor of two major biomarkers (CASP3 and MCL1) of colorectal cancer. (Somadder et al. 2023). These findings significantly shed light on the futuristic scope of these drugs as one of the promising therapeutic agents for new treatment for which they are not designated.

In this study, the hierarchical virtual screening of 3104 drug candidates, followed by filtration using MM/GBSA calculation of these drugs provided the top ten drug candidates with significant binding affinity with the Eis protein. Further scrutinising based on the obtained MM/GBSA scores, the top five drug candidates (Isavuconazonium sulfate, Cefotiam Hexetil Hydrochloride, Enzastaurin, Salbutamol sulfate, and Osimertinib) were selected for in-depth validation. Interestingly, this screening protocol has been used by various researchers for the identification of potential lead molecules (Chouhan et al. 2024; Bajrai et al. 2023; Kumar et al. 2023). Recently, five natural compounds (CNP0187003, CNP0176690, CNP0136537, CNP0398701 and CNP0043390) were screened out as suitable inhibitors of Eis protein using this screening protocol (Kumar et al. 2024). The virtual screening analysis results showed that Isavuconazonium sulfate has the strongest binding affinity with the Eis protein having an MM/GBSA score of − 106.59 kcal/mol, higher than previously identified natural compound inhibitors of Eis, whose highest MM/GBSA score was − 96.14 kcal/mol (Kumar et al. 2024). The maximum glide docking scores exhibited by these drugs were between − 13.6 to − 11.4 kcal/mol, higher than quercetin (− 8.39 kcal/mol) a known Eis inhibitor (Radhakrishnan et al. 2024). Based on this screening result analysis, it is revealed that the selected drug molecules bind more strongly to the target protein compared to the Eis inhibitors.

The protein-ligand interaction analysis of the Eis-drug complex showed that hydrogen and hydrophobic bonds provided the maximum binding affinity to maintain the stability of the complex. A similar pattern of bond formation was observed in the previously identified natural product-based Eis inhibitor. It was observed that Phe²⁴, Trp³⁶, and Phe⁸⁴ are the key contributors to bonds and these residues were also involved in bond formation in the Eis-natural product and Eis-alkaloid complexes (Kumar et al. 2024; Swain et al. 2023). Literature studies indicate that longer simulations offer a more comprehensive understanding of the druggable conformation adopted by the selected drug when complexed with the receptor, in contrast to shorter simulation durations (Ahmed et al. 2023). In the computational investigation conducted by Swain and the team, the simulation run time for the Eis-alkaloid complex was only 100 ns (Swain et al. 2023), whereas in our investigation a 500 ns MD simulation was performed, and the trajectory analysis showed that all the Eis-drug complexes showed acceptable dynamic stability. The simulation was further validated by the PCA and FEL analysis. These all suggest that these five identified drugs may have the potential with which new anti-TB drugs can be developed. These drugs could be considered for future experimental validation including enzymatic inhibition activity for the development of the new drug.

Conclusion

Drug repurposing has proven to be one of the most cost-effective techniques to accelerate the process of identification and development of new anti-TB agents. The drug resistance condition especially the aminoglycoside drug resistance still needs more attention by the scientific community. So, finding potential drugs through computational drug repurposing can accelerate the process of potential Eis inhibitors. Therefore, in this investigation, after rigorous screening and binding affinity evaluation through MM/GBSA resocuring five promising drugs (Isavuconazonium sulfate, Cefotiam Hexetil Hydrochloride, Enzastaurin, Salbutamol sulfate, and Osimertinib) were identified from 3109 FDA-approved drugs. For the drugs that interacted with the Eis protein, it found that both hydrogen and hydrophobic interactions greatly enhance the binding some protein with drug candidate. The MD simulation trajectory examination revealed that the complexes attained a state of dynamic equilibrium without undergoing any significant structural alterations during the simulation period. These findings were supported by the PCA and FEL plot analysis where the complexes showed evidence of acceptable structural arrangement to attain the final conformational stability. In conclusion, these repurposed drugs are a newer addition to the list of Eis inhibitors that can help to prevent the aminoglycoside drug-resistant condition. However, future studies including the experimental investigation of these findings need to be focused on for further validation.

Author contributions

G.S.K., K.M., R.S., E.I.A., V.D.D., and S.A. contributed towards conceptualisation. G.S.K. contributed towards data curation and performing experiments. V.D.D. and S.A. contributed towards supervision. G.S.K., K.M., R.S., E.I.A., V.D.D., and S.A. contributed towards visualisation and validation. G.S.K. and V.D.D. contributed towards writing—original draft preparation. G.S.K., K.M., R.S., E.I.A., V.D.D., and S.A. contributed towards writing—review and editing.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.