Introduction

The current human population has been unfortunate enough to witness many deadly infectious diseases of diverse origins. Since ancient times, human civilizations have seen conditions like smallpox [1], plague [2], malaria [3], influenza [4], tuberculosis [5], HIV [6], cholera [7], rabies [8], pneumonia [9], Ebola [10], variant Creutzfeldt–Jakob disease [11], Marburg virus [12], Middle East respiratory syndrome (MERS) [13], dengue [14], yellow fever [15], hantaviruses [16], anthrax [17], MRSA “superbug” [18], pertussis [19], tetanus [20], meningitis [21], syphilis [22], SARS [23], leprosy [24], measles [25], and Zika [26]. Any onset of these infectious diseases has the potential to disrupt the health care systems mainly due to overburdening in the number of cases and limiting usual general activities, and also hampering the individual’s health. The current population was in the war course of dealing with various infectious diseases, when nCOVID-19 (also known as SARS-CoV-2) came into the battlefield to give a tough fight to humanity [27]. With cases rising daily since its emergence in Wuhan, Hubei Province of China, in December 2019, it was declared a pandemic by the World Health Organization (WHO). Currently, there are no antiviral drugs with proven efficacy against SARS-CoV-2, although several vaccines are now available. Despite vaccinations (11,864,214,773 doses [as of June 15, 2022]) being administered, there have been globally 533,816,957 confirmed cases of SARS-CoV-2, and the total number of deaths reported is 6,309,633 [28]. Immune-compromised individuals and patients with multi-morbidity such as those with human immunodeficiency virus (HIV) or tuberculosis (TB) seem more vulnerable to nCOVID-19 infection-causing devastating inflammatory tissue damage leading to ICU (ventilators) admission and death in most cases. Moreover, the emergence of various other strains (Delta, Omicron etc.) and various opportunistic infections also adds to the current viral load. The underlying fact seems that nCOVID-19’s pathogenicity may be amplified in HIV-positive/TB persons with weakened immune systems, as TB is the most common opportunistic infection among HIV patients [29]. Here comes this concept of co-infection which refers to the association of two or more clinical conditions or the co-existence of multiple chronic/acute conditions in a single individual at a rate higher than expected by chance. These related diseases may have distinct etiopathogenesis (if the etiology is unknown, with specific pathophysiology of the organ or system) and are present in the same person in a defined period. They usually require co-administration of various treatment regimens involving several medications simultaneously.

As per the WHO’s global tuberculosis report 2021, the incidence and mortality of TB patients has worsened since the onset of nCOVID-19 due to a rise in co-infection cases [30]. The primary site of infection in both pathogenic conditions is the respiratory tract system, particularly the lungs. The situation became grimmer with the onset of reports from HIV-infected patients as people with HIV are more susceptible to coronavirus infection because of their immunological state, which renders them tuberculosis-prone also.

As per the literature, there have been several reported cases of triple co-infections since the emergence of nCOVID-19 of which several instances had life-threatening consequences, including respiratory failure, shock, and organ failure. Among them, few have been discussed here: (i) Tolossa et al. [31] reported a case of triple co-infection of nCOVID-19, HIV, and TB in which the patient recovered after getting in-time medical attention. (ii) Ortiz-Martínez et al. [32] reported a fatal death case of a female patient with triple co-infection (HIV/TB/SARS-CoV-2). The nCOVID-19 infection puts TB and HIV responses in jeopardy. (iii) Rivas et al. [33] presented a case report of two patients who recovered from triple infection with TB, HIV, and SARS-CoV-2 by administration of antitubercular and antiretroviral therapies simultaneously. (iv) Farias et al. [34] reported two cases of co-infection. Both subjects had pulmonary TB and HIV and developed SARS-CoV-2 infection during the 2020 pandemic. As per the report published by Sarkar et al. [35], patients with TB and HIV have an increased mortality risk during a co-infection with SARS-CoV-2. Zhu et al. [36] described the recovery of an HIV-infected patient with coronavirus‐related pneumonia. Kumar et al. [37] reported a case of SARS-CoV-2 and TB co-infection in which the patient lost his life due to worsening respiratory parameters. These incidences justify the existence of co-infections that require urgent attention.

To add on, nCOVID-19 is already influencing TB and HIV control strategies. Globally in 2020, apart from 1.3 million deaths due to TB, there were additional 214,000 deaths among HIV-positive patients who were TB infected. HIV, TB, and HIV-TB mortality have been more severely impacted by the nCOVID-19 pandemic with a first 5.6% year-on-year increase in the deaths [30]. This has led to reversal in the years of global progress. The underlying facts seems to be the depletion of CD4 T cells in HIV and latent TB infection which destroys the integrity and design of TB granulomas in the lungs, allowing active TB to develop [38]. TB, too, creates a habitat that aids HIV multiplication through various mechanisms. In fact, after SARS-CoV-2 or TB, permanent improvements in lung architecture play a vital role in both SARS-CoV-2 and TB pathogenesis. Because triple pandemics are linked in the immune-pathological phase, co-infection with SARS-CoV-2, HIV, and TB might have negative implications in all SARS, HIV, and TB phases, forming a fatal loop [39]. Finally, due to the simultaneous use of antitubercular medications, antiviral therapies, and various nCOVID-19 therapy alternatives in a patient, there is a risk of drug–drug interactions and additive hepatotoxicity, which may further deteriorate the health condition of the patient. Hence in this pandemic era, one of the primary and chronic global health issues of the twenty-first century seems to be the threat due to nCOVID-19 in patients with TB or TB-HIV co-infection. These challenges demand for the identification of a single moiety with a broad spectrum of activity.

In an attempt to find a multi-targeted ligand/single moiety here, we hypothesized to repurpose MbtA inhibitors (antimycobacterial agents) which are nucleotide analogues having proven affinity towards MbtA [40, 73. ]. These compounds being nucleotides gained our interest as we were in search for a chemical scaffold to target an appropriate vital protein in pathogens causing TB, HIV/AIDS, and nCOVID-19 infections/co-infections. Hence, we selected targets (HIV-1 reverse transcriptase (RT) and RNA-dependent RNA polymerase (RdRp) from SARS-CoV-2) for which the substrates were invariably nucleotides. Therefore, an appropriately designed nucleotide analogue (MbtA inhibitor) could have affinity towards all the three target proteins. We computationally validated this concept by performing molecular docking, molecular dynamics, post-MMGBSA analysis, and predictive in silico ADMET analysis. This study may rekindle the interest on the design and development of MbtA inhibitor and also pave a path for future multi-targeted ligand discovery against HIV-TB-nCOVID-19 co-infection. Avenues are open for the scientific community and researchers to explore this concept further.

Materials and methods

Molecular docking

Hardware and software employed

All simulation studies were performed on DELL workstation running Ubuntu 20.04.3 LTS (64-bit as OS, Intel® Core™ i7-11,800 CPU@2.30 GHz processor, 16 GB RAM, 4 GB GPU). AutoDock 4.2.6 and MGLTools 1.5.6 were employed for molecular docking simulations. Molecular dynamics simulations (MDSs) were carried out using the Desmond module of Schrodinger Suite developed by the D. E. Shaw Research group (academic license) [41]. All ligand structures were prepared using ChemOffice Suite 2019 by PerkinElmer Informatics. The 2D/3D protein–ligand interactions were visualized using BIOVIA Discovery Studio Visualizer (Molecular Graphics Environment; Dassault Systemes) [42]. All co-crystallized protein structures were obtained from the Protein Data Bank (PDB) [43] or AlphaFold Protein Structure Database [44].

Protein structure preparation

Two proteins were employed in this study: HIV-1 reverse transcriptase (PDB ID: 1RT2) [45] and RdRp from SARS-CoV-2 (PDB ID:7BV2) [46]. The X-ray crystal structures were downloaded from the PDB. The proteins were in complex with an inhibitor (internal ligand). The internal ligand was separated from the respective co-crystallized protein structures by using UCSF Chimera 1.16 [47]. The protein structure was then opened in the AutoDock 4.2.6 program [48]. The protein preparation steps involved (a) removal of water molecules, (b) addition of polar hydrogen, (c) designating AD4 atom type, and (d) addition of Gasteiger charges to the protein system. Also, selected flips to residues were applied and all-atom contacts were analyzed. The final protein structures were saved in.pdbqt format for further simulation studies. The energy minimization of these protein structures was performed using the YASARA server (http://www.yasara.org/minimizationserver.htm) [49].

Ligand preparation

The reported MbtA inhibitors have been used here as ligands for screening against HIV-1 RT (PDB ID: 1RT2) and RdRp from SARS-CoV-2 (PDB ID: 7BV2). They can be broadly classified into four categories concerning their structural scaffolds: (a) by linker modification (GV01–GV09), (b) by aromatic group modification (GV10–GV19), (c) sugar moiety modification (GV20–GV25), and (d) base modification (GV26–GV38) [40]. These ligand molecules are mentioned in Table 1 with their details. The ligand preparation steps involved (i) sketching individual structures in ChemDraw 19.1, (ii) energy minimization using the MM2 module present in Chem3D 19.1, and (iii) saving the final energy-minimized structure in.pdb format for protein–ligand docking.

Table 1 Tabular representation of 38 reported molecules (MbtA inhibitors) employed as ligands in the current study with their code, structure, and details
Full size table

Validation of docking procedure

Given the wide variety of docking and scoring functions available and the heterogeneity in their performance with different targets, it is evident that performing a docking validation study before commencing any virtual screening experiment/docking is essential [50]. The redocking was done to examine the docking procedure and efficiencies. The docking procedure was validated using method; viz., (i) the TNK inhibitor (internal ligand) from the HIV-1 RT and F58 from the RdRp of SARS-CoV-2 was extracted using UCSF Chimera 1.16 and redocked into the active site using AutoDock 4.2.6. Similar grid parameters were adopted. This exercise ensures the inhibitor’s exact binding in the active site. The deviation must be less compared to that of the co-crystallized complex available in the Protein Data Bank. The redocked complex was then superimposed onto the reference co-crystallized complex using AutoDock 4.2.6, and the root mean square deviation (RMSD) was calculated.

Theoretical concept involved

The AutoDock 4.2.6 scoring function is evaluated by using the experimentally observed protein–ligand complex as a positive control and calculating the RMSD of the other docked ligands with respect to this bound ligand [51]. Through the following equation, the RMSD compared the average distance between atoms of two ligands:

$$\mathrm{RMSD}\;\left(a,b\right)=\sqrt{\frac{1}{n}\sum\limits_{i=1}^{n}{({a}_{ix}-{b}_{ix})}^{2}+{({a}_{iy}-{b}_{iy})}^{2}+{({a}_{iz}-{b}_{iz})}^{2}},$$

where a ~ i ~ refers to the atoms of molecule 1 and b ~ i ~ to the atoms of molecule 2, respectively; the subscripts x, y, and z denote the x–y–z coordinates for every atom.

Overlay methods

Herein, the ligand’s docked conformation was overlaid with its bioactive crystallized conformation.

Chemical resemblance

Herein, the process involves mimicking the specific ligand binding process within the active binding site of the receptor as it occurs in the biological cellular system in terms of interacting residues [51].

Binding site identification and grid box generation

The binding site/active site for both proteins was identified by using the position of the internal ligand and the interacting residues of the individual macromolecules with the help of the AutoDock 4.2.6 program. The identified binding site was further utilized to specify the grid parameter points required to generate a 3D grid box. In both proteins, the grid box was located by considering the internal ligand as a center and wrapping each of the macromolecular residues interacting with the ligand to ensure that every possible ligand conformation falls inside the grid box [52]. The grid box dimensions used are mentioned in Table 2. The AutoGrid 4.2 was executed by providing the AutoGrid executable, and.gpf files as input and were converted to the grid log file (.glg). It generated map files for different types of atoms present in the ligand as well as the receptor. In this study, A C Cl F Br I HD N NA OA SA, etc., map files were generated by AutoGrid 4.2 [53]. The generated map files were utilized by the AutoDock 4.2.6 program for carrying out molecular docking simulations.

Table 2 Details of grid parameters used in the current study
Full size table

Protein–ligand docking

All molecular docking simulation studies were performed using the AutoDock 4.2.6 program (ADP) [54]. After the successful execution of AutoGrid 4.2, the genetic algorithm was set to default. Lamarckian genetic algorithm (LGA) produces a trail population of various ligand conformations followed by mutational conformations and swapping different parameters relating to successive generations of biological evaluations for the final selection of a bioactive confirmation with the lowest binding energy. In addition to it, individual and selective conformational search for their local conformational space, and identification of local minima, is an additional characteristic of the Lamarckian algorithm [53]. LGA was used for search parameters [55]. For particular proteins, the semi-empirical force field was used to predict the ligand’s binding energy. The machine-generated docking parameter file (DPF) contains the numerous parameters required for each ligand docking in the protein’s active site [52]. Parameters were as follows: the number of genetic algorithm (GA) runs was set to 50, with 2,500,000 evaluations and population size of 150. AutoDock 4.2 executable was used to run molecular docking of each ligand, and.dpf files as input were converted to the docking log file (.dlg). The final .dlg file offers binding energies for every run and inhibition constant along with a clustering histogram. From the histogram, low-energy conformer from the largest cluster was selected. This low-energy conformer of top-scoring molecules was then taken for further analysis. Complex files for binding mode analysis and for further MD simulations were then generated using the data from the .dlg file and the protein.pdbqt files. The protein–ligand complex interactions were then visualized in 2D/3D using UCSF Chimera 1.16 [47] and BIOVIA Discovery Studio Visualizer program [42].

Molecular dynamics simulation study

MDS helps study the protein–ligand complex’s structural stability and flexibility. In this study, MDS was performed for the top hit compound to legitimize the protein–ligand complex (PLC) and measure the ligand-binding constancy in the active site of the selected target. MDS was carried out using the Desmond module of Schrodinger Suite developed by the D. E. Shaw Research group (academic license) [41]. Through the system’s builder panel, the orthorhombic simulation box was prepared with the simple point-charge (SPC) explicit water model. A minimum distance of 10 Å was maintained between the protein and the solvent surface. Protein–ligand docked complexes were solvated using the cubic SPC water model [56]. The solvated system was then neutralized with counter ions and physiological salt (0.15 M). The receptor–ligand complex system was designated with the OPLS AA force field [57]. A hybrid energy minimization algorithm with 1000 steps of the steepest descent followed by conjugate gradient algorithms was utilized for energy minimization of the PLC. The reversible reference system propagator algorithm (RESPA) integrator [58], the Nosé–Hoover chain thermostat [59], and the Martyna–Tobias–Klein barostat were used with two ps relaxation times. The equilibrated PLC system (1RT2-GV17 and 7BV2-GV17) was used for the final production of the MD simulation for 100 ns at 310.15 K temperatures at 1.0 bar pressure with NPT (isothermal–isobaric ensemble [60], i.e., constant temperature, constant pressure, constant number of particles) ensemble while using default settings for relaxation before simulation. The trajectory files were written. The _out.cms file was imported to view the trajectories for further exploration. To understand the stability of the complex during MD simulation, the protein backbone frames were aligned to the backbone of the initial frame. Finally, the simulated interaction diagram was analyzed by loading the _out.cms file and selected RMSD to obtain the mentioned plots below [61]. To perform the post-simulation MM-GBSA analysis of GV17 with both proteins, the thermal_MMGBSA.py script of the Prime/Desmond module of the Schrodinger Suite was used (Schrödinger; institute license) [62]. The post-simulation MM-GBSA analysis of free binding energy calculation was carried out with the generation of 0–1000 frames. A total of 200 frames were processed and analyzed throughout the MM-GBSA calculation of 100-ns MDS data. The binding energy calculation was performed on the basis of this parameter: MM-GBSA ΔG bind, the binding energy of the receptor and ligand as calculated by the prime energy, a molecular mechanics + implicit solvent energy function (kcal/mol).

ADME and toxicity prediction

Absorption, distribution, metabolism, and excretion (ADME) is a pharmacokinetic/pharmacodynamic process that defines how the body reacts to a drug. In silico ADME data supports the drug development process as it helps in lead optimization [63]. In this study, based on molecular docking and dynamics results, the identified potential hit molecule was subjected to predictive ADME evaluation using SwissADME (an online web server developed and maintained by the Swiss Institute of Bioinformatics (SIB) (https://www.swissadme.ch)) [64]. The structures of ligands were drawn individually in the Marvin JS input panel provided on the website (http://swissadme.ch/index.php), or the SMILES format can be uploaded. After the final run, the server predicted in silico ADME. It is necessary to assess a drug’s safety profile to predict the in silico toxicity. Despite determining the harmful levels in animals, they also help reduce the number of animal tests. In the present study, we have used the pkCSM web server to predict the pharmacokinetic properties of our small molecules using graph-based signatures [65]. This server database provides the following toxicity details: AMES toxicity, maximum tolerated dose, hepatotoxicity, skin sensitization, and hERG I and II inhibition.

Results

Molecular docking studies

Validation of docking procedure

The validation/redocking studies on the crystal structure of HIV-1 RT (PDB ID: 1RT2) and RdRp of SARS-CoV-2 (PDB ID:7BV2) revealed the binding energy of − 11.62 kcal/mol and − 8.70 kcal/mol with a Ki value of 3.04 nM and 422.25 nM and the reference RMSD of 0.54 Å and 1.24 Å. These minor RMSD fluctuations are acceptable for small molecules (0–3 Å). The overlay conformations of both internal ligands concerning their crystallized conformation are presented in Fig. 1.

Fig. 1
figure 1

The superimposed overlay conformation of the docked internal ligand a TNK concerning its crystallized conformation obtained from the co-crystallized complex structure (PDB: 1RT2) and b F56 concerning its crystallized conformation obtained from the co-crystallized complex structure (PDB: 7BV2)

Full size image

Docking-based virtual screening

All ligands’ docking investigation with RdRp from SARS-CoV-2 main protease and HIV-1 RT revealed favorable binding energies and inhibition constants. Also, as an observation, it was seen that nearly 30% of the ligands showed a better binding affinity than the reference/internal ligand; this shows how well the ligands fitted into the sub-pockets of RdRp-SARS-CoV-2 and HIV-1 RT. Considering the top four scoring molecules/top hits (Figs. 2, 3, 4, 5, 6, 7, 8, and 9), only GV17 was found to bind effectively with both the proteins (1RT2 and 7BV2) with binding energies of − 12.64 kcal/mol and − 9.44 kcal/mol and inhibition constants (Ki) of 546.25 pM and 121.22 nM, respectively. The binding energies/docking scores and inhibition constants of all molecules are presented in Table 3. Tables 4 and 5 show the interaction details of the top four ligands with HIV-1 RT (PDB ID: 1RT2) and RdRp-SARS-CoV-2 (PDB ID: 7BV2).

Fig. 2
figure 2

Docking interaction of GV16 (MbtA inhibitor) in the binding pocket of 1RT2 showing various interactions within the active site

Full size image
Fig. 3
figure 3

Docking interaction of GV09 (MbtA inhibitor) in the binding pocket of 1RT2 showing various interactions within the active site

Full size image
Fig. 4
figure 4

Docking interaction of GV17 (MbtA inhibitor) in the binding pocket of 1RT2 showing various interactions within the active site

Full size image
Fig. 5
figure 5

Docking interaction of GV19 (MbtA inhibitor) in the binding pocket of 1RT2 showing various interactions within the active site

Full size image
Fig. 6
figure 6

Docking interaction of GV35 (MbtA inhibitor) in the binding pocket of 7BV2 showing various interactions within the active site

Full size image
Fig. 7
figure 7

Docking interaction of GV29 (MbtA inhibitor) in the binding pocket of 7BV2 showing various interactions within the active site

Full size image
Fig. 8
figure 8

Docking interaction of GV17 (MbtA inhibitor) in the binding pocket of 7BV2 showing various interactions within the active site

Full size image
Fig. 9
figure 9

Docking interaction of GV30 (MbtA inhibitor) in the binding pocket of 7BV2 showing various interactions within the active site

Full size image
Table 3 Results of molecular docking study of 38 analogues (MbtA inhibitors) against HIV-1 reverse transcriptase (PDB ID: 1RT2) and RNA-dependent RNA polymerase from SARS-CoV-2 (PDB ID: 7BV2) arranged in order of best score
Full size table
Table 4 Details of the docking interaction of top-scoring analogues (MbtA inhibitors) with the interacting residues in the binding pocket of HIV-1 reverse transcriptase (PDB ID: 1RT2) along with H bond length (Å)
Full size table
Table 5 Details of the docking interaction of top-scoring analogues (MbtA inhibitors) with the interacting residues in the binding pocket of nCOVID-19 RdRp (PDB ID: 7BV2) along with H bond length (Å)
Full size table

Molecular docking interaction analysis

Since we are focusing on finding a triple co-infection inhibitor, it is crucial to have a detailed study on the interactions observed to develop a proof of concept. Among the top dock score molecules, the detailed visual pose view analysis has been performed for GV17 as it was seen to interact with both the proteins (1RT2 and 7BV2) efficiently.

HIV-1 reverse transcriptase (PDB ID: 1RT2)–GV17 complex

Nucleoside reverse transcriptase inhibitors (NRTIs) and non-nucleoside reverse transcriptase inhibitors (NNRTIs) are the two basic types of reverse transcriptase inhibitors (RTIs). The former act as chain terminators, and the latter act by impeding DNA synthesis. Nucleotide RTIs also contribute to the inhibition program as their mode of action is the same as that of NRTIs. The reported MbtA inhibitors are modified nucleotides and hence interact well with 1RT2. Moreover, the docking score of GV17 was more than that of the reference compound TNK 651. Due to the apparent critical function of reverse transcriptase in HIV replication, suppressing this enzyme is one of the most promising targets for AIDS treatment. GV17 showed a significant interaction with active site amino acid residues and H bond interactions with the critical amino acid residues. This binding pocket has the advantage of high potency, selectivity, specificity, and low toxicity. The active site pocket is mainly a hydrophobic pocket comprising side chains of aromatic amino acid residues (Tyr181, Trp229, Tyr188, Phe227, and Tyr318) and hydrophobic amino acid residues (Pro95, Leu100, Val106, Val108, Val179, Leu234, and Pro236 from the p66 subunit). The HIV protein binding site also contains the following essential residues: Lys101A, Tyr 101A, Leu 100A, Val 106A, Tyr 188A, Tyr 181A, Val 179A, and Glu 138B. GV17 was favorably embedded in the hydrophobic pocket surrounded by the side chains of Leu 100A, Val 106A, Val 189A, Val 179, Leu 234A, Pro 236A, Ser 105A, Gly 190A, Pro 226A, Asp 237A, and Lys102A. It also had four hydrogen bond interactions: Leu 234A with an amino group (-NH2) present in the purine ring and Pro 236A, Lys103A, and Leu 100A with the hydroxy groups (-OH) present in tetrahydrofuran-3,4-diol. The presence of a more significant number of H bonds leads to the effective binding of the ligand in the active site. Regarding the internal ligand, the π–π stacking interactions (attractive, non-covalent interactions between aromatic rings) were conserved in GV17: Trp 229A and Tyr 188A with the quinolin-4(1H)-one ring and His 235A with the 9H-purine ring. It helps in understanding the intrinsic nature of the molecular assembly. The repositioning of Tyr 181A to form favorable van der Waals interactions is essential. The Pro 236A loops flex to optimize contacts with the substrate, which may suffice for the stability of the ligand (GV17). The loops move partly as rigid bodies leaving a residual pocket partially occupied by an electron-dense group that appears to form a hydrogen bond with the hydroxyl oxygen of the residue Lys103A. These interactions may be responsible for the binding affinity of the molecule, as indicated by the docking scores of − 12.47 kcal/mol compared to the reference ligand Tnk 651. Hence, the total volume of the active site is exploited, and the conformation is the stabilized presence of a compact inhibitor (GV17). Thus, from the binding mode analysis and docking studies, it can be concluded that GV17 with a tetrahydrofuran moiety flanked with purine moiety and quinolinone ring, which have been substituted with electron-donating and electron-withdrawing groups, showed a significant affinity towards HIV-1 reverse transcriptase compared to the reference drug TNK 651. Thus, this type of nucleotide scaffold could be exploited to develop novel HIV-1 RT inhibitors, which can facilitate better patient adherence and inhibit resistant strains of HIV.

RNA-dependent RNA polymerase from SARS-CoV-2 (PDB ID: 7BV2)–GV17 complex

Identifying potential molecules that can disrupt the functionalities of critical proteins of the SARS-CoV-2 machinery can be used as a line of defense against nCOVID-19. Apart from 3CL main protease, RdRp is a target protein in SARS-CoV-2 that has been validated and extensively studied for drug development in nCOVID-19 as it shares a high degree of homology between SARS-CoV and MERS-CoV. It has highly conserved active catalytic motifs. Due to the crucial role of RdRp in viral replication, it is considered an essential target for designing, developing, and repurposing antiviral compounds against nCOVID-19. Nucleotide drugs like ribavirin, favilavir, and remdesivir can inhibit SARS-CoV-2 in vitro [66]. GV17, a nucleotide analogue, could be a promising therapeutic moiety as nucleotide analogues have shown promising activity in various earlier studies [67, 68]. The docking score of GV17 (− 9.44 kcal/mol) was more than that of the reference compound F58 (− 8.70 kcal/mol). GV17 interacted with RdRp amino acid residues via a predominant metal coordination bond and hydrogen bonding with the active site. It revealed conserved interactions with the active site as that of remdesivir. It had formed four hydrogen bonds with RdRp pocket residues: Arg 555A and POP 1003A with hydroxy groups (-OH) present in tetrahydrofuran-3,4-diol, Ser 682A with an amino group (-NH2) present in purine ring, U 10 T (uridine base) with an amino group (-NH2) present in purine ring, and the N heteroatom in purine ring. The hydrogen bonding with the uridine base leads to a stable complex formation, as has been observed with remdesivir. The two extra hydrogen bonds (Arg 555A and Ser 682A) may explain the apparent higher potency of GV17 in inhibiting SARS-CoV-2 replication as it stabilizes the incoming nucleotide in the correct position for catalysis. Further, GV17 was covalently linked to the primary strand in protein structure to the pyrophosphate moiety and three magnesium as catalytic ions, as observed in remdesivir monophosphate. With reference to remdesivir, the π–π stacking interactions (attractive, non-covalent interactions between aromatic rings) were conserved in GV17: U 20P (uridine base) and A 11 T with the 9H-purine ring. It helps in understanding the intrinsic nature of molecular assembly [69].

Molecular dynamics simulations and post-MM-GBSA analysis

Molecular dynamics simulation studies were carried out for GV17-HIV-1 RT and GV17-RdRp-SARS-COV-2 to test the constancy of the ligand binding in the active site of the selected targets. MD studies are implemented in many drug discovery applications to study the nature of macromolecules or to interpret mechanisms of drug resistance [70]. The obtained simulation findings are discussed below.

HIV-1 reverse transcriptase (PDB ID: 1RT2)-GV17 complex

In this target protein, the conformations revealed significant RMSD values of 4.2 Å, indicating that the protein–ligand complexes were maintained constantly throughout the simulation time. RMSD explains the structural confirmations throughout the simulation. From Fig. 10, it can be interpreted that the protein–ligand complex revealed maximum stability after a 22-ns simulation with a combined RMSD of 4.2 Å (protein) and 4.2 Å (ligand). For small biomolecules, this fluctuation was acceptable. Because possible effective inhibitors should be able to bind strongly to the enzyme and create stable non-dynamic complexes, the dynamicity property provides a valid criterion to evaluate the efficiency of a proposed inhibitor.

Fig. 10
figure 10

Root mean square deviation (RMSD) of the protein–ligand complex of 1RT2 with the lowest binding energy compound GV17

Full size image

As illustrated in Fig. 11, the hydrogen bond interactions were maximum during the simulation. The amino acid residues Lys-103, Leu-234, Tyr-318, Glu-138, and Pro-236 exhibited hydrogen bond contact with GV17; interestingly, the same amino acid residue was observed in the protein’s docked pose with the ligand (Lys-103, Leu-234, Pro236, and Leu-100) (Fig. 4). Figure 12 shows the detailed atomic interactions of ligand GV17 with the protein residues of PDB-1RT2. It shows the amino acid residues involved in the bond formation with GV17 in the active site of 1RT2. This suggests that the protein–ligand complex remained stable throughout the simulation, and the system’s backbone fluctuations were modest. The simulation demonstrated more hydrophobic contacts and water-mediated linkages with GV17; the MD simulations also revealed amino acid–mediated water bridges. Figure 13 represents the specific contacts made by the protein with the ligand throughout the MD simulation. The darker the color, the greater the number of linkages with the amino acid.

Fig. 11
figure 11

Plot (stacked bar charts) of protein interactions with the ligand supervised throughout the molecular dynamics simulation of the 1RT2-GV17 complex system

Full size image
Fig. 12
figure 12

Detailed ligand GV17 atomic interactions with the protein residues of PDB-1RT2

Full size image
Fig. 13
figure 13

Specific contacts made by the protein with the ligand throughout the trajectory. (Dark color indicates more specific contact with the ligand)

Full size image

RNA-dependent RNA polymerase from SARS-CoV-2 (PDB ID: 7BV2)–GV17 complex

Any viral polymerase enzyme’s function is to replicate the virus genome or polyproteins, necessitating much flexibility in the active site to accommodate both the template and the replicate [71]. The polymerases described so far are relatively dynamic and have a large active site [72]. In this study, the computed RMSD of the protein was 2.40 Å, respectively, showing the SARS-COV-2 RNA-dependent RNA polymerase’s less dynamic characteristics. Also, because possible effective inhibitors should be able to bind strongly to the enzyme and create stable non-dynamic complexes, this dynamicity provides a valid metric to evaluate the efficacy of a potential inhibitor. It is well evident from Fig. 14 that the protein–ligand complex was stable after the 10-ns simulation with a combined RMSD of 2.4 Å (protein) and 4.5 Å (ligand).

Fig. 14
figure 14

Root mean square deviation (RMSD) of the protein–ligand complex of 7BV2 with the lowest binding energy compound GV17

Full size image

As illustrated in Fig. 15, the hydrogen bond interactions were maximum during the simulation. The amino acid residues Tyr-619, Cys-622, Thr-680, Ser-681, Asn-691, and Asp-760 exhibited hydrogen bond contact with GV17. Figure 16 shows the detailed atomic interactions of ligand GV17 with the protein residues of PDB-7BV2. This suggests that the protein–ligand complex remained stable throughout the simulation, and the system’s backbone fluctuations were modest. The simulation demonstrated more hydrophobic contacts and water-mediated linkages with GV17; the MD simulations also revealed amino acid–mediated water bridges. Figure 17 represents the specific contacts made by the protein with the ligand throughout the MD simulation. The darker the color, the greater the number of linkages with the amino acid.

Fig. 15
figure 15

Plot (stacked bar charts) of protein interactions with the ligand supervised throughout the molecular dynamics simulation of the 7BV2-GV17 complex system

Full size image
Fig. 16
figure 16

Detailed ligand GV17 atomic interactions with the protein residues of PDB-7BV2

Full size image
Fig. 17
figure 17

Specific contacts made by the protein with the ligand throughout the trajectory. (Dark color indicates more specific contact with the ligand)

Full size image

The robust binding capacity of GV17 and the numerous contacts created between GV17 and its target proteins under study may account for its possible inhibitory activity. Thus, our MD studies revealed the complexes’ exceptional stability for both proteins. Furthermore, the H bond study demonstrates that the H bonds remained stable throughout the simulation and would likely play a substantial role in complex stabilization.

The calculated binding free energy, ΔG average of the molecule GV17, was found to be − 72.30 ± 7.85 kcal/mol and − 65.40 ± 7.25 kcal/mol for 1RT2 and 7BV2, respectively. The more negative binding energy indicates a stronger affinity of GV17 towards both the receptors.

Thus, based on the above interaction analysis of GV17 with HIV-1 RT and RdRp from SARS-CoV-2 main protease and the MM-GBSA data, it is clear that the MbtA inhibitor (GV17) could serve the purpose of solving triple co-infection cases. GV17, being a modified nucleotide analogue, interacts much more efficiently with both the proteins. To sum up, GV17 could have good efficacy against RdRp from SARS-CoV-2 and HIV-1 RT when tested experimentally. Significantly, it has the potential to block the critical residues of both the receptors, as discussed above.

ADME and toxicity prediction

Results of drug-likeness, bioavailability, synthetic feasibility, and alerts for PAINS and Brenk filters

GV17 was subjected to ADME predictive evaluation using SwissADME. Drug-likeness predicts the possibility of a molecule transforming into an oral drug. In our study, five filters were employed to calculate the drug-likeness. GV17 exhibited moderate violation of drug-likeness and had a bioavailability score (55%). The Abbot Bioavailability Score predicts the fate of a molecule for 10% oral bioavailability (in rats) or quantifiable Caco-2 cell line permeability experiment. It may be defined by a feasibility score of 11%, 17%, 56%, and 85%. GV17 exhibited a score of 55%, suggesting a good bioavailability. PAINS and Brenk methods were employed to recognize the possible uncertain fragments that yield false-positive biological output. The studies indicated that GV17 did not violate any of the criteria. The synthetic accessibility of GV17 showed a moderate level of toughness as per protocols (on a scale of 1 (easy) to 10 (extremely tough)). Detailed analysis is shown in Table 6.

In silico evaluation for pharmacokinetics compliance

The fate of a molecule in the human body is evaluated in terms of its ADME properties. The ADME parameters of GV17 were estimated by calculating the different physicochemical and biopharmaceutical parameters. The physicochemical features of GV17 were analyzed.

The results indicated that the molar refractivity, which accounts for the overall polarity of the molecules, was 115; this value was in the acceptable range of the standard values (30–140). The topological polar surface area (TPSA) was 206.72 Å2. These data suggest that the molecules cannot cross the blood–brain barrier (BBB). Solubility class lipophilicity refers to the capacity of a molecule to dissolve itself into a lipophilic medium and correlates to various representations of drug properties that affect ADMET, including permeability; absorption, distribution, metabolism, and excretion; solubility; plasma protein binding; and toxicity. Results of iLOGP and Silicos-it suggested that the iLOGP value 1.20 was in the acceptable range (− 0.4 to + 5.6), while the Silicos-it value − 1.18 was in the favorable range. This lead molecule had an excellent intestinal absorption profile (48%). Water solubility is an important parameter affecting a drug’s absorption and distribution. Log S calculations represent the molecule’s solubility in water at 25 °C. For adequate solubility, the calculated log S values through the ESOL model should not exceed 6. GV17 showed a log S value of − 2.96, accounting for good solubility. The data above suggests that GV17 had a good balance between permeability and solubility and might show a good bioavailability upon oral drug administration. The predicted GI absorption was low.

Permeability predictions help understand the outcomes of ADMET and the cell-based bioassays. Results showed that the permeability over human skin was − 9.09 cm/s which was in the acceptable range. GV17 did not show the properties to cross the BBB, as discussed earlier. Metabolism can sometimes lead to drug–drug interaction and affect the bioavailability of drugs. Only the free form of the drug can bind with drug-metabolizing enzymes. To study the metabolic behavior of our lead compounds, it is vital to study their interaction with cytochrome P450 enzymes (CYPs), as they are the most notable class of metabolizing enzymes. The lead compounds found were assessed for their CYPs’ (CYPs of human liver microsomes (HLMs)) inhibitory activity. Detailed analyses are given in Table 7.

Table 6 Tabular representation of different drug-likeness rules, bioavailability, synthetic accessibility, and alerts for PAINS and Brenk
Full size table

Toxicity prediction

Analogue GV17 was studied in detail for its in silico toxicity profile. AMES test, which helps identify the mutagenic potential of a chemical compound using bacteria, showed favorable results as it showed no AMES toxicity. Results suggested that the maximum tolerated dose (human) was 0.326 log mg/kg/day, representing a moderate dosage level as per protocols. Results revealed that these compounds showed no hERGI (human ether-a-go-go-related gene) inhibition, negating the probability of ventricular arrhythmia. The oral rat acute toxicity (LD50) value was 2.401 mol/kg, while the oral rat chronic toxicity (LOAEL) value was 3.037, indicating a good safety profile. It showed neither hepatotoxicity nor skin sensitization. The predicted toxicity results of analogues GV17 are mentioned in Table 8.

Table 7 Details of the in silico ADMET profile of GV17 using the SwissADME online server
Full size table
Table 8 Tabular representation data of predicted toxicity-identified leads
Full size table

Conclusion

The co-existence of various diseases poses a significant threat to humanity. The rationale for performing this study was to repurpose the antitubercular drugs (MbtA inhibitors) for their effectiveness against HIV-RT and RdRp from SARS-COV-2. This also aimed to prevent the side effects of excessive medication in triple co-infection cases due to the administration of multiple drugs and multiple-dose regimens. The primary objective was to find a multi-targeted ligand/inhibitor that would show strong affinity with both the protein targets. The fact that these MbtA inhibitors were nucleotides gained our attention as we are in search for a chemical scaffold to target an appropriate vital protein in pathogens causing tuberculosis, HIV/AIDS, and nCOVID-19 infections/co-infections. The substrates for HIV-1 RT and RdRp of nCOVID-19 were invariably nucleotides, and hence, an appropriately designed nucleotide analogue could have affinity towards all the three target proteins. In the anticipation to combat these triple co-infections by finding a multi-targeted ligand herein, we had employed the concepts of structure-based virtual screening (SBVS) widely employed in drug discovery and repurposing. With regard to SBVS approaches, molecular docking results suggested that GV17 was the only molecule that interacted strongly with the active site residues of both proteins and successfully established various interactions, mostly hydrogen bonding. GV17 could mimic most of the interactions, as in the case of internal ligands in both proteins. Moreover, the binding energy achieved by GV17 was more than that of internal ligands, which indicates a very strong binding in the active site pocket. To validate molecular docking results, molecular dynamic simulations (100 ns) of both the PLCs (1RT2-GV17 and 7BV2-GV17) were run to evaluate and improve our design concept. The results suggested that GV17 had the potential to bind effectively in the active site of both proteins. Both PLCs were found to be stable, as is evident from the observed RMSD profile. The overall simulated structure does not exhibit any significant conformation changes and remains close to the experimental structures. Both the bioactive molecules under study stayed in the binding pocket during the simulation and formed many stable hydrophobic, polar, and H bond interactions, as discussed. This confirms the stability of GV17 in the binding pocket of both proteins. Herein, we have supported the MD simulation with the binding free energy calculations as it reflects the amount of the energy released during complex formation. A relatively stable protein–ligand complex displays more negative binding energy, indicating a stronger ligand affinity towards its receptor. We observed a similar pattern of ΔG for both our PLCs. The predicted in silico ADMET profile of GV17 was satisfactory. However, there remains scope for improving the pharmacokinetic profile of GV17 by an active analogue approach. However, we faced severe limitations while comparing three different sets of diseases belonging to different pathological situations. Also they lack phylogenetic correlation. The research on this concept is at a very nascent stage, and hence, lack of literature on this very idea made things more complex but, at the same time allowed us to explore and generate specific hypotheses, which we further tried to validate through in silico studies. The goal of this study was to not only find a promising triple co-infection inhibitor/multi-targeted ligand but also to pave a pathway for future nCOVID-19, TB, and HIV-1 RT drug development. Our findings can open a new avenue to fight against nCOVID-19-HIV-TB infection simultaneously using a single drug moiety.

Future vision

This study aimed to identify a potential triple co-infection inhibitor/multi-targeted ligand for the deadly TB, nCOVID-19, and HIV co-infection through a structure-based drug design approach. Although we have identified a possible multi-targeted inhibitor for RdRp of SARS-CoV-2 and HIV-1 RT, through the in silico approach, much work is to be done before the identified HIT (GV17) reaches clinical trials. However, we have identified the crucial residues and types of bonds required to block the RdRp of SARS-CoV-2 and HIV-1 RT, the same needs to be validated through in vitro studies. It is indeed worth thinking about improving the ADMET profile and drug-likeness. Based on the findings of this investigation, we will pursue more structure-based drug design methodologies in the future to develop a lead chemical suitable for clinical trials for the treatment of nCOVID-19 also; hence, it could be concluded that the future scope of this study depends on the in vitro response of GV17 against HIV-1 RT and RdRp of SARS-CoV-2.