In-silico identification of phytochemical compounds from various medicinal plants as potent HIV-1 non-nucleoside reverse transcriptase inhibitors utilizing molecular docking and molecular dynamics simulations

Abstract

HIV is a global public health challenge. The Reverse Transcriptase (RT) enzyme facilitates an important step in HIV replication. Inhibition of this enzyme provides a critical target for HIV treatment. The aim of this study is to employ computational techniques to screen bioactive compounds from different medicinal plants toward identifying potent HIV-1 RT inhibitors better activity than the current ones. We conducted a literature review of HIV-1 RT inhibitors, and eighty-four (84) compounds, while target receptor (1REV) was retrieved from Protein Data Bank. The molecular docking and Molecular Mechanics/Generalized Born Surface Area (MM/GBSA) evaluations were performed using the Maestro Schrodinger software user interface. The drug-likeness and pharmacokinetic profile evaluation were carried out using SwissADME and ADMETlab3.0 web servers. Lastly, molecular dynamics simulation study was conducted using the Desmond tool of Schrodinger. The molecular docking study revealed that Rosmarinic acid (−13.265 kcal/mol), Evafirenz/standard drug (−12.175 kcal/mol), Arctigenin (−11.322 kcal/mol), Luteolin (−11.274 kcal/mol), Anolignan A (−11.157 kcal/mol), and Quercetin (−11.129 kcal/mol) can effectively bind with high affinity and low energy values to the HIV-1 RT enzyme. The relative binding free energies of Rosmarinic acid, Evafirenz, Arctigenin, Luteolin, Anolignan A, and Quercetin were −66.85, −66.53, −51.83, −49.77, −58.17, and −49.62 Δg bind, respectively. The ADMET profile of Arctigenin was similar to that of Efavirenz, and better than that of other top compounds. The molecular dynamics simulation study showed better stability of rosmarinic acid with the active site of HIV-1 NNRT than the cocrystalized ligand. Out of the top five compounds identified in this study, Rosmarinic acid, a current inhibitor of HIV-1 RT in vitro, showed the most promising prediction. However, further in vivo studies and human clinical trials are required to provide more concrete information regarding its efficacy as potent HIV-1 RT inhibitors.

Introduction

Human immunodeficiency virus infection (HIV) is a global public health challenge with transmission ongoing in all countries of the world and approximately 42.3 million deaths to date [1–4]. According to the World Health Organization (WHO), the African Region accounts for the highest prevalence of HIV, with an estimated 65% of the HIV-positive individuals as of the end of 2023 [1, 2, 5]. The WHO's 2022–2030 global health sector strategy on HIV seeks to lower HIV-related mortality from 680 000 in 2020 to <240 000 in 2030, as well as HIV infections from 1.5 million in 2020 to 335 000 by 2030 [6]. Highly active antiretroviral therapy (HAART), also referred to as “AIDS cocktail therapy,” has been instrumental in the management of HIV-1 infected patients as it helps to boost the immune system, improve the quality of life, and decrease morbidity and mortality rates [4, 7, 8]. However, the effectiveness of HAART has been limited by drug resistance, drug–drug interactions and adverse drug reactions [9]. HAART involves the coadministration of three or more antiretroviral agents primarily targeting the HIV-1 integrase, HIV-1 protease, and HIV-1 reverse transcriptase (HIV-1 RT) enzymes. These agents work by synergistic or additive mechanisms to suppress or block viral replication [4, 7, 8]. HIV-1 RT inhibitors remain a cornerstone of HAART regimen and are broadly classified into two: the nucleoside reverse transcriptase inhibitors (NRTIs) and non-nucleoside reverse transcriptase inhibitors (NNRTIs) [8, 10, 11]. NRTIs competitively bind to RT, causing the premature termination of DNA synthesis [9, 12]. On the other hand, NNRTIs bind to an allosteric, hydrophobic site in the p66 subunit of HIV-1 RT located away from the NRTI binding site, which causes conformational changes within RT and ultimately inhibits DNA polymerase [5, 8, 9].

HIV-1 RT is considered a crucial target in the HIV lifecycle and is the primary focus of many treatment regimens for several reasons [5, 13–15]. First, HIV-1 RT catalyzes reverse transcription, a process that involves the conversion of the single-stranded RNA genome into double-stranded DNA [5, 14]. Inhibiting this conversion at the very early stages makes RT an essential target for antiviral drugs. Second, RT possesses multiple enzymatic active sites, including the RNA-dependent DNA polymerase and ribonuclease H, which provide multiple points of interaction for antiviral drugs to block the activity of RT [13, 16]. Third, HIV-1 RT has no human homolog, making it a specific target for antiviral agents with limited potential to harm the human cell [14]. Furthermore, mutations of amino acid residues in the enzymatic domains of HIV-RT have led to multiple cases of drug resistance, hindering the ability of antiviral agents to bind to the active sites or block the activity of HIV-RT [15, 17]. This challenge has prompted a continued search for potent antiviral agents capable of completely inhibiting HIV-1 replication, thereby preventing drug resistance and reducing the virus level to the lowest. Similarly, there are reports of low virological outcomes and HIV resistance associated with HIV-1 integrase and HIV-1 protease inhibitors [18–20].

In view of the current circumstances, it is pertinent to identify viable alternatives as anti-HIV drugs. Bioactive compounds from medicinal plants are promising anti-HIV-1 RT inhibitors, yet underexplored [21, 22]. Currently available anti-HIV-1 RT inhibitors approved by the United States of America Food and Drug Administration for the treatment of HIV between 1987 and 2024 are of synthetic origin [23, 24]. Several studies have reported the antiretroviral activity of crude plant extracts, secondary metabolites, and bioactive components from various medicinal plants [22, 25]. The plant extracts, often obtained through various techniques, are widely used as herbal therapy in many resource-constrained regions of the world with limited access to HAART [25–27]. A comprehensive literature review revealed that phytochemical compounds and extracts belonging to over 132 plant species possess anti-HIV-1 RT activity [21, 22]. These trials and reports imply that screening bioactive compounds from medicinal plants can lead to the discovery of novel anti-HIV-1 drug candidates. The in-silico method of drug discovery utilizes various computational tools and techniques for screening promising drug candidates [28–30]. This method has been successfully employed in the discovery of several FDA-approved drugs for various illnesses, including Captopril, an anti-hypertensive agent [30], Lapatinib, an anti-cancer drug [31], Cimetidine, an anti-ulcer drug [32], and Zanamivir, an anti-influenza agent [29]. Darunavir—HIV-1 protease inhibitor [30], and Lopinavir—an agent for the treatment of HIV that is resistant to other protease inhibitors [31]. Emerging insights into the mechanisms of polymerization, drug inhibition, and drug resistance from structure–activity relationship investigations may facilitate the development of novel plant-derived HIV-1 RT inhibitors capable of circumventing existing drug resistance mutation patterns [21, 24]. In this study, we utilized in-silico techniques, including molecular docking, Molecular Mechanics-Generalized Born Surface Area (MMGBSA), and molecular dynamics simulations, to screen bioactive compounds for identifying potent anti-HIV-1 RT drug candidates with better activity than existing ones.

Materials and methods

Study outline

We performed a comprehensive literature review of HIV-1 RT inhibitors [22], followed by filtering of identified compounds using the Lipinski rule of five [32]. Using the Maestro Schrodinger software user interface, we conducted the Molecular docking and Molecular Mechanics/Generalized Born Surface Area (MM/GBSA) calculations. Furthermore, we evaluated the drug-likeness and ADME properties of the top compounds through the SwissADME and ADMETLAB3.0 web servers [33, 34]. Lastly, we predicted stability of the top compound in the biological system by calculating the molecular dynamics (MD) simulation using the Desmond tool of Schrodinger. We performed all computational studies on an HP computer with 12th Gen Intel(R) Core (TM) i7-1265U 1.80 GHz with 16.0 GB installed RAM and a 64-bit operating system, with the exception of the MD simulations which was performed on Tesla V100-SXM2-16GB. Below is the workflow chart for this study (Fig. 1).

Figure 1.

Workflow of this study

Ligand selection and preparation

We retrieved a total of eighty-four (84) phytochemical compounds (Supplementary Table S1) with reported activities against the HIV-1 RT enzyme from published literature, including information on the plant family, plant distribution, and the specific part of the plant where each bioactive compounds were isolated [22]. These phytochemical compounds were screened using a two-step approach: first, by inputting the canonical Simplified Molecular Input Line Entry System (SMILES) into the SwissADME web server to ensure conformance with the Lipinski rule of five (RO5), and second, through expert-based selection considering chemical family, structural class, and molecular complexity. We selected only the thirty-four compounds that met all the Lipinski RO5 criteria (including hydrogen bond donor <5, hydrogen bond acceptor <10, molecular weight of <500 g/mol, and partition coefficient of <5) for the next stage of the screening protocol. We downloaded the three-dimensional structures (3D) of these compounds from the PubChem database in the SDF format and imported them into the Maestro Schrodinger software. The ligands were prepared using the LigPrep module of Maestro Schrodinger, using OPLS3 force field at a target pH of 7.0. LigPrep module functions by extending tautomeric and ionization states, ring conformations, and stereoisomers in accordance with the supplied data to comprehensively represent the pertinent molecular states in three dimensions.

Protein preparation

We retrieved the crystal structure of HIV-1 RT complexed with 4-chloro-8-methyl-7-(3-methyl-but-2-enyl)-6,7,8,9-tetrahydro-2h-2,7,9a-triaza-benzo[cd]azulene-1-thione (PDB: ID 1REV) from the Research Collaboratory for Structural Bioinformatics (RCSB) website (https://www.rcsb.org/) [35]. Subsequently, we imported the protein into the Maestro workspace and preprocessed it using the Protein Preparation Wizard. The protein consists of two chains, designated as chain A and chain B. Since only one chain was required for our analysis, we deleted chain B, leaving us with only chain A. Similarly, we removed water molecules and magnesium ions from chain A, and we assigned hydrogen bonds to optimize the protein using PROPKA at pH 7.5. Lastly, we subjected the protein to restrained minimization utilizing the OPLS3e force field.

Receptor grid generation

We generated a receptor grid from the minimized protein using the Glide module to identify the active site of the 1REV for ligand-receptor docking. This was achieved by selecting the cocrystalized ligand [4-chloro-8-methyl-7-(3-methyl-but-2-enyl)-6,7,8,9-tetrahydro-2h-2,7,9a-triaza-benzo[cd]azulene-1-thione] at the active site of 1REV to produce a cubic grid box with three-dimensional coordinates X, Y, and Z having values of 2.26 Å, −37.14 Å, and 22.51 Å, respectively. We further utilized the following parameters to develop the receptor grid: a 1.0 Van der Waals radius scaling factor, a 0.25 partial charge cutoff, and default settings for site constraints, rotatable groups, and excluded volumes.

Molecular docking

We performed molecular docking analysis using the Glide module within the Maestro interface of Schrödinger 11.8, employing three hierarchical docking activities with increasing precision: high-throughput virtual screening (HTVS), standard precision (SP), and extra precision (XP). The 34 ligands that have been previously prepared through the LigPrep module, and the standard drug (Evafirenz), as well as the receptor grid obtained from the receptor grid module, were utilized for docking. We screened the prepared ligands using HTVS precision, and a docking threshold of <−6.0 kcal/mol was used, followed by SP precision docking, after which a docking threshold of <−8.0 kcal/mol was used to filter the ligands. Lastly, we performed XP docking using a docking threshold of <−10.0 kcal/mol to screen the ligands. We adopted the default scaling factor of 0.8 and partial charge cutoff of 0.15 to soften the potential for nonpolar parts of the ligand. Lastly, the default setting samples nitrogen inversions and sample ring conformations for the ligand sampling, and the addition of Epik state penalties to the docking score was left unchanged.

Estimation of binding energy using MMGBSA

To further validate the results obtained from the XP docking, we estimated the free binding energy using Molecular Mechanics-Generalized Born Surface Area (MM-GBSA) of the Maestro’s Prime module. We evaluated the biological response of the complexes by estimating the free binding energy using Molecular Mechanics-Generalized Born Surface Area (MM-GBSA) of the Maestro’s Prime module [36]. This analysis was carried out using the top 4 ligands from the XP precision docking, along with the standard ligand (Evafirenz), utilizing a default OPLS3 force field and the VSGB solvation model.

ADME-Tox and drug-likeness prediction

We transformed the structures of the top 5 compounds, and the standard drug into their respective canonical simplified molecular-input line-entry system. This was followed by analysis of the drug likeness property and toxicity profile using SwissADME (http://www.swissadme.ch/index.php) and ADMETLAB 3.0 (https://admetlab3.scbdd.com/). We evaluated the following key pharmacokinetic and physicochemical parameters: topological polar surface area, bioavailability score, Lipinski rule of five, gastrointestinal absorption, CYP1A2 inhibitor, CYP3A4 Inhibitor, P-gp inhibitor, plasma protein binding, volume of distribution, clearance, half-life, carcinogenicity, hepatotoxicity, nephrotoxicity, and genotoxicity.

Molecular dynamics simulation (MD simulation)

Using the Desmond tool in Schrodinger, we subjected the top compound from the XP precision docking and the cocrystallized ligand to an MD simulation of 100 ns. The simulation activity was performed under the following conditions: 1.01325 bar, 300 K, and the addition of Na+ and Cl− to maintain neutrality in an explicit solvent system within a physiological environment, using an orthorhombic box with a 10 × 10 × 10 Å buffer and 0.15 M salt. The force field utilized was transferable intermolecular potential-4-point (TIP4P). Following the MD simulation, we visualize the trajectories for Root Mean Square Fluctuation (RMSF), Root Mean Square Deviation (RMSD), protein–ligand interactions, radius of gyration, and hydrogen bond distribution through the Desmond-Schrodinger suite’s interaction diagram module.

Results and discussion

Molecular docking analysis of HIV-1 non-nucleoside reverse transcriptase inhibitors

The co-crystallized ligand, located at the active site of HIV-1 RT, was extracted and re-docked in the same binding pocket to validate the docking protocol. The native and redocked co-crystallized ligands were superimposed, yielding a low root mean square deviation (RMSD) of 2.8734 Å (Supplementary Fig. S1, see online supplementary material for a color version of this figure), confirming the reliability of the docking protocol [37]. The results of the analysis were presented in a table and graphical format (Table 1, Figs 2 and 3). The top five compounds (Rosmarinic acid, Arctigenin, Luteolin, Anolignan A, and Quercetin) demonstrated docking scores ranging from −11.129 to −13.265 kcal/mol. Rosmarinic acid has a higher binding affinity (−13.265 kcal/mol) than the standard drug, Evafirenz (−12.175 kcal/mol) and other compounds. The binding free energies (in kcal/mol) provide insight into each compound’s relative binding energy, with larger negative values suggesting stronger binding affinity [38].

Table 1.

Docking, MMGBSA scores and interactions modes of the top five compounds, and the standard drug.

Name of compounds	Docking score (XP) (kcal/mol)	MMGBSA (Δg bind)	Interactions
			H-bond	Hydrophobic	Other interactions
Rosmarinic acid	−13.265	−66.85	LYS101, LYS103, HIS235	PHE227, TRP229, TYR183, TYR181, VAL179, TYR188, PRO95, LEU100, TYR318, LEU234, PRO236, VAL106	Pi-Pi Stacking: TYR188
Evafirenz (standard)	−12.175	−66.53	LYS101 (2), LYS103	PRO95, TYR181, VAL179, TYR188, LEU100, TYR318, VAL106, PRO236, LEU234, PHE227, TRP229	None
Arctigenin	−11.322	−51.83	LYS101	TRP229, PHE227, PRO95, LEU234, PRO236, PR097, LEU100, TYR318, VAL106, VAL179, VAL179, ILE180, TYR181, TYR188, VAL189	Pi-Pi Stacking: TYR188
Luteolin	−11.274	−49.77	LYS101	PRO95, TRP229, PHE227, TYR183, TYR181, VAL179, TYR188, LEU234, PRO236, VAL106, TYR318, LEU100	Pi-Pi Stacking: TYR 188
Anolignan A	−11.157	−58.17	LYS101 (2)	TRP229, PHE227, TYR188, TYR181, ILE180, VAL179, VAL106, LEU234, TYR318, PRO236, LEU100	Pi-Pi Stacking: TYR188 (2), TRP229 (2)
Quercetin	−11.129	−49.62	LYS101	TRP229, PHE227, PRO95, TYR183, TYR181, TYR188, VAL179, VAL106, LEU234, PRO236, LEU100, TYR318	Pi-Pi Stacking: TYR188

Figure 2.

The chemical structure of the top five compounds and the standard drug

Figure 3.

Graphical representation of the binding free energy (docking score) and prime/MM-GBSA (Δg bind) binding energy for the top five hits and the standard drug

The 2D and 3D Molecular interactions of amino acid residues of HIV-1 Reverse Transcriptase with top compounds and standard drugs are shown in Figs 4 and 5, respectively. All the compounds demonstrated a typical hydrogen bond interaction via LYS101 at the active site of 1REV, except for trachelogenin. Hydrogen bonds play a crucial role in protein-ligand interactions, influencing molecular recognition, binding affinity, and overall stability of the complex. They contribute to the specificity of these phytochemical compounds and are vital for biological processes. Out of the top five compounds, only rosmarinic acid showed hydrogen bond interaction with three different amino acid residues (LYS101, LYS103, HIS235), which could have contributed to its higher docking score. Furthermore, most of the compounds showed interaction with hydrophobic amino acids, including TRP 229, TYR 183, TYR 181, TYR 188, PRO 95, and VAL 179.

Figure 4.

2D-Molecular interactions of amino-acid residues of HIV-1 reverse transcriptase with top compounds and standard drugs. (a) rosmarinic acid, (b) evafirenz, (c) arctigenin, (d) luteolin, (e) anolignan A, and (f) quercetin

Figure 5.

3D-Molecular interactions of amino-acid residues of HIV-1 reverse transcriptase with top compounds and standard drugs. (a) rosmarinic acid, (b) evafirenz, (c) arctigenin, (d) luteolin, (e) anolignan A, and (f) quercetin

The NNRTIs, also known as allosteric inhibitors, bind to a hydrophobic pocket (non-nucleoside inhibitor-binding pocket; NNIBP) distal to the active site within RT. Binding to the NNIBP alters the shape of the RT active site, thereby inhibiting viral DNA synthesis. Overall, the NNIBP is predominantly hydrophobic and comprises two subunits, the p66 subunit having the LEU 100, LYS 101, LYS 103, VAL 106, THR 107, VAL 108, VAL 179, TYR 181, TYR 188, VAL 189, GLY 190, PHE 227, LEU 234, TRP 229, TYR 234, and TYR 318 amino acid residues, and the p51 subunit having only one amino acid residue, GLU 138 [28, 29, 39, 40]. The p66 subunit of HIV-1 RT enzyme contains both the polymerase domain (responsible for DNA synthesis) and the RNase H domain (involved in degrading the RNA template). Therefore, the interaction of phytochemical compounds with the amino acid residues of the p66 subunit is critical for inhibiting viral DNA synthesis [28, 29, 39, 40]. The top two compounds, Rosmarinic acid and arctigenin, were observed to possess similar hydrophobic interactions as the standard drug with the following residues: PRO 95, TYR 181, VAL 179, TYR 188, LEU 100, TYR 318, VAL 106, PRO 236, LEU 234, PHE 227, and TRP 229, contributing to their higher binding affinity at the allosteric site of HIV-1 RT enzyme. Thereby, inducing conformational changes in the active site of the enzyme. Additionally, this interaction provides a strong basis for their anti-HIV-1 inhibitory activity.

The molecular docking results obtained in this study are consistent with those from previous literature. For instance, the results obtained from the paper titled “Identification of the novel drug target for the treatment of HIV by using the Bioinformatics approach” showed that out of the six bioactive compounds isolated from the aerial parts of Salvia Officinalis, Origanum vulgare and Thymus vulgaris, rosmarinic acid exhibited the highest binding affinity across six target proteins (3LII, 1HD2, 1QS4, 5CDQ, 3MNG and 2IOK) [41]. Similarly, of the 15 ligands evaluated for antiviral activity in this study titled “HIV-1-RT inhibition activity of Satureja spicigera (C.KOCH) BOISS. Aqueous extract and docking studies of phenolic compounds identified by RP-HPLC-DAD,” rosmarinic acid was among the top three ligands with the highest docking scores [42]. These reports highlight the therapeutic potential of rosmarinic acid as a promising candidate for the discovery of antiviral drugs.

Estimation of binding energy using MMGBSA

Molecular Mechanics-Generalized Born Surface Area (MM-GBSA) provides a method for estimating the binding affinity of a ligand to a protein target by calculating the free energy change upon binding, which is achieved by decomposing the energy into different components (e.g. van der Waals, electrostatic, and solvation) and summing them up. By comparing the calculated binding free energies of different ligands, the most promising candidates for further investigation can be identified [43]. The MMGSA scores of the five top compounds, the standard drug, and the co-crystallized ligand range from −49.62 to −66.85 ΔG bind, and are consistent with the docking results (Table 1 and Fig. 3). Rosmarinic acid has the highest docking and MMGBSA scores (−13.265 kcal/mol, −66.85 ΔG bind), followed by Evafirenz (−12.175 kcal/mol, −66.53 ΔG bind). However, the MMGBSA scores of Anolignan A (−58.17 Δg bind) and Arctigenin (−51.83 Δg bind) were found to be higher than those of Luteolin (−49.770 Δg bind) and Quercetin (−49.62 Δg bind). Overall, all the top compounds exhibit good interactions with the receptor, making them promising candidates for HIV-1 RT inhibition.

Drug-likeness and ADME-Tox prediction

Table 3 presents the results of the drug-likeness properties and bioavailability scores for the top compounds, the standard drug, and the co-crystallized ligands. The assessment was based on Linpiski’s “Rule of 5” (RO5), which predicts the bioavailability, solubility, and permeability of proposed drug candidates. According to the rule, compounds are more likely to have poor oral absorption or poor biological barrier penetration if they have hydrogen bond donors >5, hydrogen bond acceptors >10, a molecular weight >500, and a partition coefficient (LogP) value >5 [32]. All the top compounds, and the standard drug showed good physicochemical properties and did not violate Linpiski’s RO5.

Table 3.

Absorption, distribution, metabolism, excretion and toxicity profiles of the top five compounds, and the standard drug.

Compounds	GI absorption	CYP1A2 inhibitor	CYP3A4 inhibitor	P-gb inhibitor	PPB	VD	CLplasma	T1/2	Carcinogenicity	Hepatotoxicity	Nephrotoxicity	Genotoxicity
Rosmarinic acid	Low	No	No	No	77.3%	0.451	13.238	1.906	Yes	No	Yes	No
Evafirenz	High	Yes	No	Yes	99.3%	0.907	1.615	1.412	No	No	No	No
Arctigenin	High	No	Yes	Yes	95.8%	0.432	5.071	2.222	No	No	No	No
Luteolin	High	Yes	Yes	Yes	97.6%	0.243	8.482	1.373	No	Yes	Yes	No
Anolignan A	High	Yes	No	Yes	97.7%	4.536	3.024	1.417	No	Yes	Yes	No
Quercetin	High	Yes	Yes	Yes	98.7%	0.132	8.289	1.586	Yes	No	No	Yes

The Total Polar Surface Area (TPSA) is a crucial descriptor that offers valuable insights into a molecule’s absorption, distribution, polarity, and lipid solubility profiles [44]. The TPSA values of rosmarinic acid (144.52 Ų) was slightly higher than the normal range (20–130 Ų), which could negatively affect their permeability across biological membranes. Furthermore, the bioavailability scores of all top compounds, and efavirenz were between 0.55 and 0.56 (Table 2), implying a favorable bioavailability score, as compounds with bioavailability scores between 0.00 and 0.55 are considered pharmaceutically active.

Table 2.

In-silico drug-likeness prediction of the top five compounds, and the standard drug.

Compounds	Molecular weight (g/mol)	Num. H bond donor	Num. H bond acceptor	TPSA (A2)	LogP	Bioavailability score	Linpiski (RO5)
Rosmarinic acid	360.08	5	8	144.52	2.005	0.56	Yes; 0 violation
Efavirenz	315.03	1	3	38.33	3.32	0.55	Yes; 0 violation
Arctigenin	372.16	1	6	74.22	2.043	0.55	Yes; 0 violation
Luteolin	286.05	4	6	111.13	2.247	0.55	Yes; 0 violation
Anolignan A	310.12	2	4	58.92	3.242	0.55	Yes; 0 violation
Quercetin	302.04	5	7	131.36	1.448	0.55	Yes; 0 violation

The pharmacokinetic properties, including absorption, distribution, metabolism, and excretion profiles, are presented in Table 3 and Fig. 6. The gastrointestinal (GI) absorption of rosmarinic acid was found to be low, but all other reported compounds have high GI absorption. Evafirenz, luteolin, anolignan A, and quercetin are potential inhibitors of cytochrome (CYP) 1A2 (CYP1A2), while others are non-inhibitors of this isoenzyme. Similarly, the inhibitors of CYP3A4 include arctigenin, luteoline, and quercetin, while other compounds showed no potential for inhibition. Both CYP1A2 and CYP3A4 isoenzymes belong to a superfamily of drug-metabolizing enzymes called CYP450, which are primarily located in the liver and are part of six predominant isoenzymes responsible for 90% of drug metabolism [44]. Evafirenz, arctigenin, luteoline, anolignan A, and quercetin are not P-gb substrates, hence have high oral absorption. On the other hand, rosmarinic acid is a P-gp substrate, a factor which could be responsible for its low oral absorption. P-glycoprotein (P-gp) is a member of the ATP-binding cassette (ABC) superfamily that pumps drugs out of the cell. P-gp inhibitors suppress this efflux activity, thereby enhancing drug absorption and increasing bioavailability [45].

Figure 6.

Predicted ADMET properties of the top compounds and standard drug

The Plasma Protein Binding (PPB) of rosmarinic acid, efavirenz, arctigenin, luteolin, anolignan A, and quercetin were 77.3%, 99.3%, 95.8%, 97.6%, 97.7%, 98.7%, and 51.3%, respectively. The optimal PPB of drugs is < 90%. High plasma protein-bound drugs may have a low therapeutic index. Rosmarinic acid, arctigenin, luteolin, and quercetin have moderate plasma clearance, while anolignan A and efavirenz have low clearance. The top compounds exhibited a short half-life. The toxicity profile showed that Rosmarinic acid and quercetin have carcinogenic potential, whereas Luteolin and anolignans A are hepatotoxic. Furthermore, rosmarinic acid, luteolin, and anolignan A were predicted to have nephrotoxic potential. Lastly, only quercetin showed genotoxic potential. Interestingly, arctigenin and efavirenz exhibited a similar toxicity profile, with no hepatotoxic, nephrotoxic, genotoxic, or carcinogenic potential (Fig. 6).

Molecular dynamics simulation

Molecular dynamics (MD) simulation is a computational method used to understand the intrinsic interactions between selected ligands and the protein structure. MD tends to further refine and confirm the docking results of protein-inhibitor complexes. Additionally, MD simulation elucidates the dynamic features of the top-ranking small molecules obtained from docking procedures on a nanosecond scale. The simulation results obtained are analyzed based on statistical parameters, including the root mean square deviation (RMSD), root mean square fluctuation (RMSF), and the radius of gyration (Rg) [46]. To better understand the interactions involved in the inhibition of HIV-1 reverse transcriptase (HIV-1 RT), we performed an MD simulation for 100 ns, and subsequently analyzed the results.

Root mean square fluctuation

RMSF provides in-depth insight into the flexibility of amino acid residues of a protein. High RMSF values for amino acid residues suggest mobility and instability in the receptor. Low RMSF values of amino acid residues imply a stable and stiff receptor [47]. The RMSF plot exhibits varying fluctuation, depicting the various levels of the protein’s flexibility throughout the simulation time. The extent of fluctuation of each protein residue is reflected by the height of the peak and is utilized for determining which protein residue is most stable. From the graphical illustration depicted in Fig. 7, the fluctuation pattern observed between HIV-1 reverse transcriptase and rosmarinic acid, as well as between HIV-1 reverse transcriptase and crystallized rosmarinic acid, during the simulation time of 100 ns, is very dissimilar. For the rosmarinic acid complex, high RMSF peaks were observed for the amino acid residues ASP 67, ASN 136, ASN 137, GLU 138, LYS 287, ALA 288, LEU 289, THR 290, GLU 91, and HIS 539, with RMSF values ranging from 4.7 to 6.0 Å.

Figure 7.

Line representation of the evolution of root-mean-square fluctuation of HIV-1 reverse transcriptase Cα during the simulation. (a) rosmarinic acid and (b) cocrystallized ligand

Similarly, the following amino acid residues PRO 25, LYS 65, LYS 66, ASN 136, ASN 137, GLU 138, THR 470, ALA 538, and HIS 539, having RMSF values between 4.35 and 8.59 Å, were found to exhibit more fluctuations in the cocrystalized ligand complex. A similar pattern of fluctuations was also observed across some amino acid residues found in both complexes, including ASN_136, ASN_137, GLU_138, and HIS_539, indicating a similar binding of both rosmarinic acid and the cocrystalized ligand to the receptor. It is worthy of note that none of the amino acids residues (the LEU 100, LYS 101, LYS 103, VAL 106, THR 107, VAL 108, VAL 179, TYR 181, TYR 188, VAL 189, GLY 190, PHE 227, LEU 234, TRP 229, TYR 234, and TYR 318) at the allosteric site of HIV-RT exhibit a high level of fluctuations, showcasing the stability of the amino acid residues at the allosteric site, and fluctuation of those at the distal part of the receptor. An implication of the result is the existence of a balance between the flexibility and stability observed across the binding site residues interacting with these compounds.

Root-mean-square deviation

RMSD is used to evaluate the stability of the receptor-ligand complex [35, 48]. To assess the stability of ligand-protein complexes during simulation, the RMSD of the HIV-1 RT backbone atoms was plotted alongside the primary crystal protein across 100 ns (Fig. 8). Figure 8a shows the ligand RMSD and protein RMSD plots for rosmarinic acid–HIV-1 RT complex during the 100 ns simulation period. During the early phase of the trajectory, the protein RMSD (left Y-axis) for rosmarinic acid ranges between 0.8 and 6.4 Å from 0 to 20 ns, demonstrating noticeable conformational changes as the simulation begins. This type of fluctuation is often encountered at the beginning of the simulation as the system adjusts to attain stability. Interestingly, the system exhibited a similar pattern of fluctuation between 20 ns and 100 ns, with RMSD values ranging from 3.8 to 5.8 Å. The protein RMSD value obtained at the end of the simulation was within the standard, acceptable range (1–3 Å), indicating that the system had converged by the end of the simulation. The Ligand RMSD (right Y-axis) of rosmarinic acid showed a very similar pattern of fluctuation with the protein RMSD. During the early trajectory, there was an observable fluctuation in the system, with a value of 1–7.8 Å, between 0 and 20 ns. As the simulation continued, there was an increase in stability within the system, with a range of 4.3–7 Å observed between 20 and 50 ns. Toward the end of the simulation time (50–100 ns), the system equilibrates, with RMSD values of 3.5–6 Å, demonstrating a stable orientation of rosmarinic acid within the binding pocket of the HIV-1 RT enzyme.

Figure 8.

The RMSD values for alpha carbon (Cα) atoms (blue line) of HIV-1 reverse transcriptase and ligand fit on Prot (red line). (a) Rosmarinic acid and (b) cocrystallized ligand were plotted to a 100 ns simulation period

Figure 8b shows the ligand RMSD and protein RMSD plots for the cocrystallized ligand. The protein RMSD plots (left Y-axis) revealed cycles of fluctuations followed by stabilization at specific intervals. A significant fluctuation occurred during the first 50 ns simulation period, with an RMSD ranging from 2.8 to 7.2 Å. Afterward, it stabilized between 51 and 100 ns, with protein RMSDs of 4.8 and 7.2 Å. The Ligand RMSD (right Y-axis) revealed noticeable fluctuation as the simulation began, with the cocrystallized ligand having an RMSD of 0–4.5 Å between 0 and 20 ns, demonstrating the tendency of the ligand to diffuse from the active site of the HIV-1 RT enzyme. This was followed by appreciable stability with values 3.2–5.0 Å from 21–60 ns. After this period, further fluctuations were observed, with values ranging from 4.0 to 7.2 Å from 61 to 100 ns. This implies that the RMSD of the Ligand is still increasing and decreasing toward the end of the simulation period without stabilizing at a fixed point.

Protein–ligand contact

Molecular dynamics simulation sheds light on the possible interactions that occur between a protein and small molecules, usually referred to as protein-ligand contacts. This interaction is typically referred to as protein-ligand contacts. It is illustrated with the aid of a histogram, as shown in Fig. 9. Protein-ligand contacts are categorized into four types: hydrogen bonds (H-bonds), Hydrophobic Interactions, ionic interactions, and Water Bridges. The presence of Hydrogen bonds between the protein and ligand is key in maintaining binding specificity to the protein’s active site [48]. The MD simulation result of rosmarinic acid revealed that, during the 100 ns simulation, it maintained the H-bond it formed with LYS 101 during the docking studies and also formed a new one with ASP 186 and GLN 91.

Figure 9.

Protein–ligand interaction mapping of (a) rosmarinic acid and (b) cocrystallized ligand

While the H-Bond interaction formed with LYS 101 lasted for 20% of the simulation time, the contact time with ASP 186 and GLN 91 accounted for 5% of the simulation time. Further analysis revealed that LEU 100, LYS 103, VAL 179, TYR 181, TYR 188, TYR 229, and LEU 234 exhibited hydrophobic interactions during the simulation process. Similarly, GLN 91, LYS 101, ILE 180, VAL 189, and TYR 318 formed water bridges during the simulation. In comparison with the protein-ligand contact formed during the molecular docking study, rosmarinic acid exhibits similar H-Bond contact with LYS101 and LYS103. Additionally, the hydrophobic interactions with amino acid residues (PRO 95, TYR 181, VAL 179, TYR 188, VAL 106, LEU 234, and TRP 229) at the active site of HIV-RT, as identified during the docking study, were also maintained.

For the Cocrystallized ligand complex, the following residues were observed to form H-bonds with the binding site of the target, including: LYS 101, VAL 179, TYR 188, and VAL 189. Although other residues exhibit <10% of the contact time, residue TYR 188 was observed to have the highest interaction time, accounting for 10% of the simulation time. Worthy of note is the fact that although hydrogen bonds are key to ligand binding specificity, the stability of the HIV-1 RT–ligand complex is due to the presence of hydrophobic bonds, which are more prevalent in the core regions of the protein [49]. In lieu of this, the MD result of the Cocrystallized ligand complex was assessed for the presence of hydrophobic interactions, with nine residues exhibiting holistic hydrophobic contacts. The residues include: ILE 94, PRO 95, PRO 97, LEU 100, VAL 106, TRP 229, MET 230, TYR 232, and LEU 234. Each residue is observed to show varying contact time, as is evident with ILE 94 forming a hydrophobic bond for 18% of the simulation time, TRP 229 having the maximum interaction period of over 40%, and every other residue being below 15% of the simulation time.

By analyzing the interactions that occur in more than 30.0% of the simulation, rosmarinic was found to interact with more protein residues (LEU 100, LYS 101, VAL 179, ILE 180, TYR 181, TYR 188, TRP 229, and LEU 234) than the cocrystalized ligand, which only interacted with two (2) protein residues (TYR 181 and TRP 229). Additionally, rosmarinic acid and the cocrystalized ligand displayed interaction fractions of 1.2 and 0.4, respectively, with the protein residues. This suggests that rosmarinic acid exhibits higher binding affinity and stability with HIV-1 RT than the cocrystalized ligand.

Radius of gyration analysis

Radius of gyration (Rg) is a parameter used to assess the overall compactness, stability, and folding of the protein, which may be affected due to the presence of a ligand [50]. Estimating the Rg provides valuable insights into the structural dynamics of a protein during the simulation. The lower the value of Rg, the more stable the protein-ligand complex. Figure 10 shows the Rg of (a) Rosmarinic acid and (b) cocrystalized ligand. The rosmarinic acid-protein complex showed significant stability of 34 Å between 0 and 5000 ps. However, the Rg peaks to 55 Å at several intervals. Toward the end of the simulation, a lower Rg of 38 Å was observed at 100000 ps. In comparison, the Rg of the cocrystalized ligand-protein complex was 39 Å 10000ps at the start of the simulation, but it reduced to 35 Å at 100000ps simulation time. Average Rg values of HIV-1 RT: rosmarinic acid (38.29 Å) was close to the cocrystalized ligand (35.19 Å). Overall, HIV-RT showed similar compactness and stability when bound with both rosmarinic acid and the cocrystalized ligand.

Figure 10.

Radius of gyration (Rg) of (a) rosmarinic acid and (b) cocrystalized ligand

Hydrogen bonds analysis

The number of intermolecular hydrogen bonds formed is vital for understanding the protein-ligand stability. A strong inhibitor tends to form more hydrogen bonds with its target protein. As the number of hydrogen bonds increases and the length of hydrogen bonds decreases, the stability and anchoring of the complex get tighter [51]. The intermolecular hydrogen bonds for (a) Rosmarinic acid and (b) cocrystalized ligand are depicted in Fig. 11. In the case of rosmarinic acid (a), a maximum of seven (7) hydrogen bonds were observed, while 5 or 6 bonds were noticed during most of the simulation period. With respect to the cocrystalized ligand (b), a maximum of 2 hydrogen bonds was recorded, while 1 hydrogen bond was observed during most of the simulation. Comparatively, rosmarinic acid exhibit more hydrogen bond interactions with the target protein than the cocrystalized ligand suggesting a stronger binding affinity and stability between rosmarinic acid and HIV-1 RT complex than the cocrystalized ligand complex.

Figure 11.

Intermolecular hydrogen bonds for (a) rosmarinic acid and (b) co-crystalized ligand

Conclusion

In this study, we screened bioactive compounds obtained from published in vitro studies and identified novel inhibitors of HIV-1 RT enzyme. Rosmarinic acid was found to exhibit a higher binding affinity and binding energy than all compounds and the standard drug (Efavirenz). All the top compounds demonstrated potent activity against the HIV-1 RT enzyme and can be explored as lead compounds for the development of potent, effective anti-HIV-1 drugs. While the in vitro and in silico efficacy of these bioactive compounds have been established, further in vivo studies and human clinical trials are required to provide more concrete information on the efficacy of the top compound (rosmarinic acid) as potent HIV-1 RT inhibitors.

Author contributions

Suleiman Danladi (Conceptualization [equal], Data curation [equal], Investigation [equal], Methodology [equal], Project administration [equal], Resources [equal], Software [equal], Supervision [equal], Validation [equal], Writing—review & editing [equal]), Ayinde Abdulwahab Adeniyi (Conceptualization [equal], Data curation [equal], Formal analysis [equal], Investigation [equal], Methodology [equal], Project administration [equal], Resources [equal], Software [equal], Validation [equal], Visualization [equal], Writing—original draft [equal], Writing—review & editing [equal]), Zainab Iman Sani (Conceptualization [equal], Data curation [equal], Formal analysis [equal], Project administration [equal], Resources [equal], Software [equal], Writing—original draft [equal]), and Adegbenro Temitope (Data curation [equal], Formal analysis [equal], Investigation [equal], Methodology [equal], Project administration [equal], Resources [equal], Visualization [equal], Writing—original draft [equal])

Supplementary data

Supplementary data are available at Biology Methods and Protocols online.

Conflict of interest statement. The authors declare that there is no conflict of interest. The authors alone are responsible for the accuracy and integrity of the paper content.

Funding

None declared.

Data availability

All relevant data are within the manuscript.