A computational approach to evaluate caffeoylquinic acids and flavonoids in Pluchea indica Less. leaves as potential anti-HIV agents

Context : The attachment of human immunodeficiency virus type 1 glycoprotein 120 (HIV-1 gp120) to the CD4 receptor of human immune cells is the beginning of HIV-1 infection. Stimulation of reactive oxygen species (ROS) production through upregulation of nicotinamide adenine dinucleotide phosphate (NADPH) oxidase-2 (NOX-2) and -4 (NOX-4), and cytochrome P450 2E1 (CYP2E1) of the virus can be a potential target for anti-HIV agents. Aims : To evaluate the inhibitory effects of caffeoylquinic acids (CQAs) and flavonoids of Pluchea indica leaves against the binding of HIV-1 gp120 with CD4 receptor and their antioxidant activities via interactions with NOX-2, NOX-4, and CYP2E1 through in silico study. Methods : Ten CQAs and nine flavonoids of P. indica were docked to the 3TGS (gp120 HIV-1), 2CDU (NOX-2), 3A1F (NOX-4), and 3T3Z (CYP2E1) receptors using the AutoDockTools 1.5.7. Physicochemical and pharmacokinetics properties were predicted using the pkCSM online tool, while toxicity was predicted using the ProTox-II webserver. Results : Mostly, all of the CQAs and flavonoids were able to bind to all receptors. 3,4-Di-O-caffeoylquinic acid has the lowest binding energy (-8.79 kcal/mol) against 3TGS (gp120). 5-O-Caffeoylquinic acid and apigenin have great potential as antioxidants due to their good binding with NOX-2 and CYP2E1. However, CQAs might have ADME problems. Most test compounds did not cause hepatotoxicity, carcinogenicity, or mutagenicity. All test compounds have no cytotoxic potential. However, all CQAs have the potential to be immunotoxins. Conclusions : The findings indicated that 3,4-di-O-caffeylquinic acid could be a potential inhibitor of HIV-1 gp120-CD4 binding, while 5-O-caffeoylquinic acid and apigenin demonstrated strong antioxidant activities via NOX-2 and CYP2E1 inhibition. However, in-depth studies, including experimental in vitro and in vivo studies, are required to validate the anti-HIV activity of the compounds further.


INTRODUCTION
Human immunodeficiency virus type 1 (HIV-1) is a lentivirus that infects the human immune system (World Health Organisation, 2023), especially cells with CD4 receptors on their surface, such as T helper cells, macrophages, dendritic cells, and astrocytes (Seitz, 2016).In 2022, 39 million people were living with HIV in the world (World Health Organisation, 2023).Until now, treatment with antiretroviral (ARV) is still the primary choice for suppressing viral replication (Shin et al., 2021).Targets for HIV replication inhibition from ARV include attachment, fusion or entry, and viral enzymes ( Popović-Djordjević et al., 2022;Sierra-Aragón and Walter, 2012).
Attachment is an early stage in the HIV-1 life cycle, which can be a crucial target with great opportunities in the discovery and development of ARV (Caffrey, 2011), HIV-1 pre-exposure prophylaxis therapy (Malik et al., 2017;Mirani et al., 2019), even protection from infection (Bruxelle et al., 2021).Glycoprotein 120 (gp120) is one of the HIV-1 proteins (besides gp41) that plays a role in the initial attachment (Malik et al., 2017).Gp120 has also been reported to play a role in stimulating the production of reactive oxygen species (ROS), such as O2 •− and H2O2, in various cell lines (i.e., astrocytes and microglia) (Ivanov et al., 2016;Reshi et al., 2014).Stimulation of ROS production occurs through upregulation of cytochrome P450 2E1 (CYP2E1), proline oxidase (POX), and activation of the nicotinamide adenine dinucleotide phosphate (NADPH) oxidase-2 (NOX-2) and -4 (NOX-4) (Ivanov et al., 2016).High amounts of ROS can increase the risk of accelerating the progression of infection towards acquired immunodeficiency syndrome (AIDS) (Reshi et al., 2014).Fostemsavir is a first-in-class drug as an attachment inhibitor, which was approved by the United States (US) Food Drugs Association (FDA) in July 2020 for the treatment of patients with multidrug-resistant HIV-1 infection (Hiryak and Koren, 2021).However, as with other ARVs, the possibility of adverse side effects (viz., poor tolerability, toxicities, and drug-drug interactions, among others), availability of types and affordability of drugs in certain countries (Forsythe et al., 2019) are challenges.Thus, the discovery and development of new medicinal compounds are urgently needed.
Generally, drug discovery and development from medicinal plants begin with metabolite extraction, purification, preclinical testing, and clinical trials on humans.The different stages can take a long time, require a lot of energy, and be very costly.In addition, the failure rate at the clinical trial stage is very high, with possibly only one out of 5000 lead compounds reaching the market for therapeutic use (Ezzat et al., 2019).Recently, computational approaches have been used to screen candidate compounds to be developed as drugs particularly through in silico molecular docking studies (Najmi et al., 2022;Shaker et al., 2021).Through molecular docking studies, predictions of interactions and binding affinity between candidate compounds and target receptors can be studied simultaneously.In addition, the assessment of pharmacokinetic properties, such as ADME and toxicity, can be done in silico (Rudrapal and Chetia, 2020).
Therefore, the evaluation of CQAs and flavonoids from P. indica leaves as anti-HIV targeting gp120 HIV-1 and antioxidants targeting interactions with CYP2E1, NOX-2, and NOX-4 through computational molecular docking studies was carried out.Their performance was also compared with reference drugs, such as fostemsavir, BMS-806, dextromethorphan, apocynin, and propofol.In addition, in silico ADME and toxicity of the tested CQAs and flavonoids were also predicted in this study.

Preparation of proteins
The type of target receptors, their characteristics, and the positive controls used in this study are presented in Table S1.Before docking, each receptor was prepared by removing water molecules and ions from the receptor molecule.The native ligand was separated from the receptor molecule.Polar hydrogen was added to the receptor structure.Receptor structure preparation was carried out with PyMol (v.2.5.5).Fig. S1 shows the structures of the receptors that were prepared and used in this study.

Preparation of ligands
The ligands used in this study were compounds from the CQA and flavonoid groups found in P. indica leaves through literature searches.The list of lig-ands used in this study is presented in Table 1.The structure of each compound was downloaded from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/).Optimization of the shape of 3D compounds was carried out by adding hydrogen and minimizing energy using Avo-gadro2.Once completed, the file was saved in the pdb format.

Validation of the docking method
The docking method was validated by re-docking the native ligand on the active site of the receptor.The docking method validation parameters were met if the root mean square deviation (RMSD) value was ˂2 Å. Specifically for the 3A1F receptor, the docking method was validated using VAS2870 on the binding pocket previously predicted by Vinay et al. (2020).

Determination of predicted physicochemical parameters
Prediction of these parameters was carried out via the pkCSM webserver (https://biosig.lab.uq.edu.au/pkcsm/prediction), which was accessed for free.The respective ligand molecules were uploaded in canonical SMILE format.The parameters of the physicochemical properties of each ligand determined in this study were molecular weight (g/mol), log P, number of rotatable bonds, hydrogen bond donors, hydrogen bond acceptors, and surface area.

Determination of predicted toxicity parameters
Prediction of toxicity parameters for each ligand was carried out using the ProTox-II, which was accessed for free via https://toxnew.charite.de/protox_II/).The predicted toxicity parameters were the lethal dosage 50 (LD50) value, toxicity class, hepatotoxicity, carcinogenicity, immunotoxicity, mutagenicity, and cytotoxicity properties of each compound.

Data analysis
Validation of the docking method was accepted if the RMSD value of the redocking result was ˂2 Å.The binding affinity energy (G) and the inhibition constant (Ki) values from the molecular docking study of each CQA and flavonoid tested were compared with each reference drug.In addition, chemical interactions between residues on the target protein and the tested ligand with the most negative binding affinity energy values were observed and presented in the form of 2D images using Discovery Studio Visualizer 2021 (v21.1.0.20298).Predictions of the physicochemical properties of each compound investigated were compared with the Lipinski rule (Turner and Agatonovic-Kustrin, 2007).The ADME and toxicity predictions for each compound tested were compared with the criteria for each test parameter based on literature (if any).

Evaluation of molecular docking
The overlay of the native ligand crystal with the re-docked ligand is presented in Fig. 1.Based on the results obtained, the RMSD value of each re-docked native ligand gave a result of <2.0 Å.It means that the docking method can be used to dock the compounds being tested.Meanwhile, blind-docking was carried out on the 3A1F receptor using ordinates and grid box dimensions based on a study by Vinay et al. (2020).
The tested compounds are predicted to provide low binding affinity energy (G) and show the best inhibition constant (Ki) value.For example, 3,4-di-Ocaffeoylquinic acid has a Ki value of 0.36 M (for 3TGS, gp120 HIV-1) lower than the Ki value of the reference drugs (Fostemsavir, Ki = 1.34 M and BMS-806, Ki = 0.86 M) for 3TGS.Otherwise, if the binding affinity energy is high, then the Ki value of the test compound is poor.

ADME prediction
Table 4 describes the predicted absorption, distribution, metabolism, and excretion (ADME) properties of each ligand tested using the pkCSM online tool.Absorption studies can be carried out using various approaches.Generally, cell-based assays using cell lines such as Caco-2 and Madin-Darby canine kidney (MDCK) are performed for this purpose (Van de Waterbeemd et al., 2007).This study found that the Caco-2 permeability of apigenin has a value of >0.9, while the other test compounds have a value of <0.9.The intestinal absorption (human) of apigenin is >30% (93.25%).It shows that the apigenin compound has the potential to be well absorbed in the small intestine.Meanwhile, most CQAs and flavonoids in P. indica were predicted to have intestinal absorption values (human) >30%.The skin permeability value of the tested compounds was in the range of −2.735 (<2.5), which means that these compounds could penetrate the skin easily.All test compounds were predicted as P-glycoprotein substrates.All tested flavonoids were not inhibitors of P-glycoprotein I and II.4,5-Di-O-caffeoylquinic acid methyl ester and 3,4,5tri-O-caffeoylquinic acid methyl ester were predicted to be inhibitors of P-glycoprotein I. 3,4,5-Tri-Ocaffeoylquinic acid methyl ester and 1,3,4,5-tetra-Ocaffeoylquinic acid were P-glycoprotein II inhibitors.
The distribution parameters determined include VDss (human) (D1), fraction unbound (human) (D2), BBB permeability (D3), and CNS permeability (D4).The VDss value accepted was >−0.15.Thus, based on the prediction results, all tested compounds in this study met the criteria.Unfortunately, all tested compounds have BBB permeability values that did not meet the criteria (log BBB <0.3), which means that they have difficulty in penetrating the BBB.In this study, only kaempferol, luteolin, and apigenin were predicted to meet acceptable CNS permeability criteria because they have log P −3.It means that these three compounds have the potential to penetrate the central nervous system.All tested compounds did not affect CYP2D6 and were not metabolized by it.The di-, tri-, and tetra-CQAs in this study were predicted to be able to act as CYP3A4 substrates.Quercetin, kaempferol, myricetin, luteolin, and apigenin might act as CYP1A2 inhibitors.3,4,5-Tri-O-Caffeoylquinic acid methyl ester could possibly act as CYP3A4 substrate and CYP3A4 inhibitor, while luteolin as CYP1A2 and CYP2C9 inhibitors.Apigenin was predicted as CYP1A2 and CYP2C19 inhibitors.
The excretion parameters determined were total clearance (E1) and renal OCT2 substrate (E2).The total clearance values of all flavonoids and mono-CQAs in this study were positive.It means they could be excreted quickly.On the other hand, the compounds di-, tri-, and tetra-CQAs have negative values.Furthermore, all compounds in this study were not substrates of the OCT2, which is involved in the uptake and secretion of cationic drugs.

DISCUSSION
Gp120 HIV-1 is a viral protein that plays an essential role in the initial life of the virus in human target host cells with the CD4 receptor.In summary, at-tachment of gp120 to CD4 causes a conformational change of gp120, which subsequently allows the presentation of the chemokine co-receptor protein CXCR4 or CCR5 binding site together with gp41 (Caffrey, 2011;Sierra-Aragón and Walter, 2012).This process opens the way for HIV genetic material to enter the host cell and initiate the HIV cycle itself.CYP2E1 and NADPH oxidase (NOX-2 and NOX-4) are activated due to the presence of viral particles, one of which is gp120, causing excess ROS production in cells (Ivanov et al., 2016).Excessive amounts of ROS can cause activation of nuclear factor (NF)-B.This factor controls gene transcription, which can lead to increased HIV replication (Aquaro et al., 2008).Through molecular docking studies, predictions of the interaction of test compounds with specific receptor targets or proteins can help researchers search for and develop drug-candidate compounds (Najmi et al., 2022).
3TGS is a crystal structure of HIV-1 clade C strain C1086 gp120 core in complex with NBD-556 (Table S1).NBD-556 (native ligand 03G) is a small molecule (337.8 Da).This molecule binds to the Phe43 cavity of gp120 (Kwon et al., 2012) and shows its potential in targeting the inhibition of HIV-1 gp120 binding to host cells with the CD4 receptor (Tintori et al., 2013).In this study, it appears that the oxalamide NH in the native ligand 03G interacts with the Glu429, Met426, Gly473, and Asn425 amino acid residues with hydrogen bonds in the Phe43 cavity of 3TGS (Fig. 2) with a distance of 1.98, 2.14, 2.26, and 2.00 Å, respectively.According to Tintori et al. (2013), the oxalamide moiety of NBD-556 formed two hydrogen bonds with the carbonyl oxygen atoms of Asn425, Asp368, and Gly473 residues in the proposed Phe43 binding mode.In this study, the compounds 3,4-di-O-caffeoylquinic acid and quercetin 3-O-glucoside also interacted with Asn425 and Asp368 residues via hydrogen bonds.Previously, it was reported that the compound 3,4-di-O-caffeoylquinic acid interacts with HIV-1 integrase through molecular docking studies (Hu et al., 2010;Serina et al., 2016).This compound was assessed as having poor interaction with HIV-1 protease in silico (Serina et al., 2016).
Meanwhile, 3A1F and 2CDU are crystal structures of NOX.NOX is an enzyme that produces ROS (superoxide, O2 •− or hydrogen peroxide, H2O2).The enzyme was identified in the membranes of phagocytic immune cells, namely macrophages and neutrophils (Vermot et al., 2021).NOX plays a role in pathogen killing by pumping ROS into the phagosome, where ingested pathogens can be destroyed.NOX also plays a role in cellular signaling, which is related to the processes of apoptosis, proliferation, homeostasis, and gene regulation (Couret and Chang, 2016).NOX is also commonly known as phagocytic NADPH oxidase (phox).There are several types of NOX, including NOX-1, NOX-2, NOX-3, NOX-4, NOX-5, dual oxidase-1 (DUOX-1) and DUOX-2.At the beginning of the entry of viral proteins into host cells, gp120 triggers the expression of NOX-2 and NOX-4.NOX-2 (gp91 phox ) is an enzyme isolated from phagocytes with a molecular weight of 91 kDa.NOX-2 was the first NOX isoform identified.Meanwhile, NOX-4 is expressed most in the kidney, osteoclasts, fibroblasts, and endothelial cells.Its maturation depends on p22 phox .NOX-4 produces detectable H2O2 in vitro in the absence of superoxide dismutase (Vermot et al., 2021).By blocking NOX-2 and NOX-4, oxidative stress levels can be reduced (Reshi et al., 2014).The 3A1F receptor does not have a native ligand, so docking is carried out using the blind docking method, where the binding set is determined according to studies by Vinay et al. (2020) by using VAS2870 as the target ligand.VAS2870 is an inhibitor of NOX isoforms (except NOX-3) (Vinay et al., 2020).Meanwhile, the native ligand used in the 2CDU receptor is ADP.Apocynin and dextromethorphan are used as reference inhibitors of NOX (Jiang et al., 2013;Da Silva Costa et al., 2018).In this study, compounds apocynin, 4-O-caffeoylquinic acid, and quercetin 3-Oglucuronide interacted with Gln123 and Ser91 residues in the 3A1F binding site via hydrogen bonds.Meanwhile, dextromethorphan only showed hydrogen with Ser91.A study reported that phenolic derivative compounds anchored to the 3A1F receptor showed interactions with Ser91, Gln142, and Glu135 receptors via hydrogen bonds (Aqeel et al., 2020).Furthermore, in this study, the ligands ADP, 5-Ocaffeoylquinic acid, apigenin, and apocynin both interacted with Cys242 and Asp179 residues in the 2CDU receptor binding site.Meanwhile, with Asp179 residue, dextromethorphan showed carbon-hydrogen bond interactions.According to Da Silva Costa et al. ( 2018), Asp179 is an amino acid residue that interacted via hydrogen bonds with the test ligand, namely two selected caffeine analogs, and via carbonhydrogen bonds with dextromethorphan.
Cytochrome P450 (CYP) 2E1 (CYP2E1) is a family of heme-containing monooxygenase enzymes (Leung et al., 2013).In this study, the 3T3Z receptor, which is human CYP2E1 in complex with pilocarpine, was used.One study reported that the presence of gp120 showed the cause of increased expression of CYP2E1.They were involved in the production of ROS, which causes oxidative stress.Antioxidant activity associated with HIV-1 pathogenesis that acted on the pathway could potentially be a new drug target (Reshi et al., 2014).There are several mechanisms of CYP2E1 inhibition, including haem ligation (4-methyl pyrazole, 3-amino-1,2,4-triazole, and diallyl sulfide), haem interaction (disulphiram and phenethyl isothiocyanate), and competitive (propofol) (Lewis et al., 2000).In this study, 5-O-caffeoylquinic acid and apigenin were compounds that are good at interacting with 3T3Z (CYP2E1).
Failure of the drug in its clinical application and unmanageable toxicity are closely related to the poor pharmacokinetics profile of the drug.Thus, initial evaluation by predicting ADMET properties using online tools is straightforward.It can reduce research costs compared to in vitro studies and shorten research time (Dulsat et al., 2023).In this study, predictions of the physicochemical and ADME properties of the test compounds were carried out using the pkCSM tool.The tool is considered to have an extensive range of ADME parameter information.Another advantage is that this tool can be accessed for free (Dulsat et al., 2023).Drug-likeness is related to the Lipinski rule of five (Turner and Agatonovic-Kustrin, 2007).This study found that the 3,4-di-O-caffeoylquinic acid is an anti-HIV candidate with a mechanism of action via attachment inhibition gp120-CD4.The compound also has antioxidant properties via NOX inhibition.However, the compound violated three Lipinski's rules.Based on the rule, if or more criteria are violated, then the high-risk compound has oral bioavailability problems, such as poor absorption and permeation capabilities (Van de Waterbeemd et al., 2007).However, it should be noted that the rule does not definitely categorize whether a compound will be absorbed quantitatively well or poorly.In addition, compounds that do not violate the criteria are not always orally bioavailable.Thus, scientific proof regarding this matter is still needed.Even though the compound can be absorbed well, it can still have low bioavailability due to the presence of a high pre-systemic clearance system.In general, low bioavailability of marketed drugs may occur in oral dosage forms of hydrophobic chemical components and be absorbed slowly (Turner and Agatonovic-Kustrin, 2007).
As previously explained, the unmanageable toxicity of candidate compounds is also critical in their development as drugs.In this study, the ProTox-II webserver was used as a tool to predict the toxicity of tested compounds.The advantage of this tool is that it has extensive toxicity information, is easy to interpret, and is easily accessible for free (Banerjee et al., 2018).In this study, the toxicity of the compound 3,4-di-Ocaffeoylquinic acid is predicted to be in class 5 as a compound that is categorized as safe.However, this compound is predicted to have potential immunotoxicity properties.Behne et al. (2023) reported that the CQAs (especially 5-O-caffeoylquinic acid and 3,5-di-O-caffeoylquinic acid) found in coffee by-products have immunotoxin effects which could impact the immune system.However, the intake of both in coffee by-products is still considered relatively safe.

CONCLUSION
Based on this in silico study, CQAs and flavonoids from P. indica have the potential to act as anti-HIV and antioxidant agents.3,4-Di-O-Caffeoylquinic acid was suggested to act as an anti-HIV by inhibiting gp120-CD4 attachment and as an antioxidant via NOX inhibition.The flavonoid that has the potential to be a good anti-HIV in this study was quercetin 3-Oglucoside.Antioxidant activity through NOX-2 inhibition was demonstrated by the 5-O-caffeoylquinic acid and apigenin, while NOX-4 inhibition was demonstrated by the 4-O-caffeoylquinic acid and quercetin 3-O-glucuronide.Meanwhile, 5-O-caffeoylquinic acid and apigenin were potential CYP2E1 inhibitors.By targeting the initial process of infection, drug candidates can be developed for HIV prevention and prophylaxis therapy.Based on ADMET predictions, there is a possibility that 3,4-di-O-caffeoylquinic acid compound experiences problems with oral bioavailability and immunotoxicity.However, in-depth studies, including experimental in vitro and in vivo studies, are required to validate the anti-HIV activity of the compounds further.Citation Format: Hikmawanti NPE, Saputri FC, Yanuar A, Jantan I, Yeni Y, Mun'im A (2024) A computational approach to evaluate caffeoylquinic acids and flavonoids in Pluchea indica Less.leaves as potential anti-HIV agents.J Pharm Pharmacogn Res 12(4): 701-721.https://doi.org/10.56499/jppres23.1896_12.4.701 Publisher's Note: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers.Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Open Access: This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/ licenses/by/4.0/), which permits use, duplication, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Figure 1 .
Figure 1.Overlay of the crystal structure of the native ligand (color by element) and the redocking result ligands (magenta) of the 3TGS (A), 2CDU (B), and 3T3Z (C) receptors.

Table 1 .
Compounds used in this study.

Table 1 .
Compounds used in this study(continued…)

Table 2 .
Molecular docking results of selected P. indica compounds.

Table 3 .
Physicochemical of ligands based on Lipinski's rule.

Table 4 .
Prediction of ADME properties of ligands.

Table 5 .
Prediction of toxicity properties of ligands.