Molecular Docking and In-Silico Analysis of Natural Biomolecules against Dengue, Ebola, Zika, SARS-CoV-2 Variants of Concern and Monkeypox Virus

The emergence and rapid evolution of human pathogenic viruses, combined with the difficulties in developing effective vaccines, underline the need to develop innovative broad-spectrum antiviral therapeutic agents. The present study aims to determine the in silico antiviral potential of six bacterial antimicrobial peptides (AMPs), two phytochemicals (silvestrol, andrographolide), and two bacterial secondary metabolites (lyngbyabellin A, hapalindole H) against dengue virus, Zika virus, Ebola virus, the major variants of SARS-CoV-2 and monkeypox virus. The comparison of docking scores obtained with natural biomolecules was performed with specific neutralizing antibodies (positive controls for ClusPro) and antiviral drugs (negative controls for Autodock Vina). Glycocin F was the only natural biomolecule tested to show high binding energies to all viral surface proteins and the corresponding viral cell receptors. Lactococcin G and plantaricin ASM1 also achieved high docking scores with all viral surface proteins and most corresponding cell surface receptors. Silvestrol, andrographolide, hapalindole H, and lyngbyabellin A showed variable docking scores depending on the viral surface proteins and cell receptors tested. Three glycocin F mutants with amino acid modifications showed an increase in their docking energy to the spike proteins of SARS-CoV-2 B.1.617.2 Indian variant, and of the SARS-CoV-2 P.1 Japan/Brazil variant, and the dengue DENV envelope protein. All mutant AMPs indicated a frequent occurrence of valine and proline amino acid rotamers. AMPs and glycocin F in particular are the most promising biomolecules for the development of broad-spectrum antiviral treatments targeting the attachment and entry of viruses into their target cell.


Introduction
The emergence and evolution of disease-causing viruses have posed a phenomenal threat to human health and have become a tremendous challenge to modern medicine and the global economy. These viruses are largely zoonotic in origin, i.e., they originate in a particular animal reservoir species and are transmitted to humans [1]. Depending on its potential to infect and transmit among humans, an emerging virus can result in a few sporadic cases, leading to a localized outbreak or can develop into an epidemic or even into a global pandemic in the worst scenarios. Such events of emergence over the past decades are numerous and varied in occurrence [2]. The management of these viral infections has become extremely challenging due to the ability of viruses to mutate and evolve over time under the influence of environmental, ecological and socio-economic factors, in particular, increasing globalization and climate change [2,3]. Examples of these outbreak-causing agents include the dengue virus, influenza viruses that cause swine and Int. J. Mol. Sci. 2022, 23, 11131 2 of 23 avian flu, emerging viruses such as Ebola, Zika, MERS, and SARS coronaviruses, and now the monkeypox virus [4,5]. Some of these viruses undergo high mutation rates and/or genome re-assortment, which allow antiviral drug resistance, host immune evasion, and reduced response to vaccines via antigenic shift and antigenic drift [6,7]. Viral interactions based on ecology, genetics, and host cell entry that determine the emergence of these viruses are extremely complex, making it impossible to predict the mechanism of the next epidemic or pandemic [8,9]. Thus, when a viral outbreak emerges and raises fears of a pandemic, a coordinated global response is needed that includes individual protection, social distancing, quarantines, information campaigns, and the development of antiviral treatments and vaccines [10,11]. Vaccine development takes time, but the development of antiviral treatments can be anticipated through the development of broad-spectrum drugs targeting a wide range of viruses [12][13][14][15].
The recent COVID-19 pandemic has shown that strategies to initiate the development of suitable innovative treatments for emerging viruses are crucial [16]. In recent years, studies that focus on natural products such as bioactive secondary metabolites and antimicrobial peptides (AMPs) from microorganisms and plants have been of great interest among researchers with the aspiration of identifying novel antiviral drugs against emerging viruses that cause epidemics and pandemics [17][18][19]. Recently emerging viruses that lead to pandemics include human immunodeficiency virus, SARS, MERS, hantavirus, dengue, West Nile virus, and Ebola and Zika virus [20]. Viral pathogens that caused major global pandemics include avian influenza  [21]. Potential viral targets for these new antiviral drugs include capsid or envelope structural proteins [22,23]. Advances in computer technology permeate many aspects of drug discovery in the present day. Such technologies include virtual screening for hit identification and techniques for lead optimization which contribute towards low cost and safe screening of potential agents for the purpose of rational drug discovery. Virtual screening can be categorized into structure-based and ligand-based methods [24]. Molecular docking is the most commonly applied technique for virtual screening of molecular interactions since the early 1980s. Computer programs based on a variety of algorithms have been developed to perform molecular docking studies, as this virtual screening technique has become an increasingly important and critical tool in pharmaceutical research [25].
This study was conducted with a view to identifying new broad-spectrum natural antiviral agents targeting the attachment of the viral particle and its entry into the target cell. We herein report the binding potential of bioactive molecules naturally produced by bacteria and plants to surface viral proteins and corresponding cell receptors using molecular docking computer software in a virtual setting.

Results
The results of the in silico experiments reveal potential profound interactions between some biomolecule drug candidates and viral proteins or cellular virus receptors associated with dengue, Ebola, Zika, monkeypox viruses, and variants of SARS-CoV-2. The higher docking energy score was used to determine the docking strength of each ligand molecule with its respective receptor. The docking energy scores of reference material were used to compare the effectiveness of each biomolecule drug candidate and its antiviral potential. The data for molecular docking analyses for reference materials are summarized in Table 1. Using Autodock Vina software, the cut-off value for predicting high docking energy between a ligand and a receptor has been set at −6 kcal/mol in previous studies [26,27]. The ClusPro energy score reflects the attempt to achieve the native source with the lowest free binding energy [28]. The antiviral antibodies used as positive controls and cellular virus receptors indicated very high binding energy when docked with their corresponding surface 3 of 23 viral proteins, with docking scores ranging from −749.6 to −1239.9 kcal/mol. Despite their mechanism of action targeting the replication machinery of the viral genome, the antiviral drugs Brincidofovir, Molnupiravir, Remdesevir, and Sofosbuvir showed docking scores with surface viral proteins slightly above the cut-off value described in the literature. The antiviral drug Tecovirimat inhibiting the envelope protein p37 which is essential for the extracellular transmission of the MPV also achieved a relatively low docking score for the viral protein A42R. The comparison of docking scores obtained with natural biomolecules (AMPS, phytochemicals, or bacterial secondary metabolites) was performed with the negative and positive controls used in the study, i.e., specific neutralizing antibodies (positive controls for ClusPro) and antiviral drugs (negative controls for Autodock Vina). A docking score obtained with ClusPro that was greater than or equal to the results obtained with the positive control was considered high. A docking score obtained with Autodock Vina that was higher than the results obtained with the negative control was considered high.
Glycocin F showed high binding energies to all viral surface proteins and corresponding cell virus receptors tested (Tables 2-8). Plantaracin ASM1 and lactococcin G also showed  high docking scores with all viral surface proteins tested and all cell receptors for DENV,  ZIKV, EBOV, and SARS-VoV-2 variants, but not with monkeypox receptors (Tables 3 and 4). The docking energy scores of bacterial AMPs ranged from −771 to −975.2 kcal/mol with DC-SIGN, from −793.5 to −1336.2 kcal/mL with AXL, from −756.8 to −1163.1 kcal/mol with TIM-1, from −609.3 to 1018.9 kcal/mol with ACE2, from −958.2 to −1505.5 kcal/mol with Toll-like receptor 5, from −719 to −1114.9 kcal/mol with CR3/Mac-1, and from −579.9 to −1134.2 kcal/mol with CD36 ( Table 2). The docking energy scores of bacterial AMPs with spike (S) proteins of SARS-CoV-2 ranged as follows; for bacteriocin plantaricin ASM1 the range was −1237.5 to −1399.3 kcal/mol (Table 3), for bacteriocin lactococcin G the range was −1009.9 to −1262.3 kcal/mol (  Table 3). The highest docking energy for the Ebola virus glycoprotein was recorded with bacteriocin glycocin F, which indicated a docking score of −1208.2 kcal/mol. The interaction of specific amino acid residues of the bacterial AMPs and the surface viral proteins are also presented in Tables 3-8. PyMOL indicated the presence of a variety of interacting amino acid residues for each docked molecule.          The intersection between N-Terminal of S1 Domain and C-Terminal of S1 Domain       (Table 9). Andrographolide showed high docking energies with the viral proteins of EBOV, SARS-CoV-2 B.1.1.7 UK, and SARS-CoV-2 P.1 Japan/Brazil variants (Table 10). Lyngbyabellin A showed high docking scores only with SARS-CoV-2B.1.1.7 UK, SARS-CoV-2 P.1 Japan/Brazil, SARS-CoV-2 B.1.617.2 Indian, and SARS-CoV-2 B.1.1.529 Omicron variants (Table 11). Silvestrol recorded the highest docking energy of −8.4 kcal/mol for the spike protein of the SARS-CoV-2 B.1.351 South African variant (Table 9). Andrographolide showed the highest docking energy for the spike proteins of the SARS-CoV-2 B.1.1.7 UK variant with a value of −7.7 kcal/mol (Table 10). Lyngbyabellin A recorded the highest docking score of −9.0 kcal/mol for the spike protein of the SARS-CoV-2 B.1.1.7 UK variant (Table 11). Hapalindole H had the same highest docking energy for the spike proteins of the SARS-CoV-2 B.1.351 South African and SARS-CoV-2 P.1 Japan/Brazil variants with a value of −8.2 kcal/mol (Table 12). Silvestrol, lyngbyabellin A, andrographolide, and hapalindole H showed high docking scores to the DENV cellular receptor, but low docking energies to the ZIKV, MPV, and SARS-CoV-2 cellular receptors ( Table 2). Hapalindole H but not silvestrol, lyngbyabellin A, and andrographolide showed high docking energy to the EBOV cellular receptor ( Table 2).    The intersection between N-Terminal of S1 Domain and S2 Domain −6.9     The 3D representations of the molecular docking results with the highest docking scores for each viral protein are shown in Figure 1. Best binding energy illustrations for the six targets for silvestrol, andrographolide, lyngbyabellin A, and hapalindole H with the surface viral proteins and interacting amino acid residues are illustrated in Figures 2-5, respectively. The AMPs with the highest binding energies for the viral proteins were mutated and the mutants were then evaluated for their affinity for the viral proteins. Among the mutant AMPs tested, only three glycocin F mutants showed an increase in their docking energy with the spike protein of SARS-CoV-2 B.1.617.2 Indian variant, the spike protein of SARS-CoV-2 P.1 Japan/Brazil variant, and the DENV envelope protein compared to the corresponding wild-type AMPs, indicating improved molecular docking stability (Table 13 and Figure 6). This was not the case for the bacteriocin plantaricin ASM1 against the Zika virus protein E and against the spike (S) protein of SARS-CoV-2 B.1.351 South African variant, nor was it the case for the bacteriocin glycocin F against the spike (S) protein of SARS-CoV-2 P.1 Japan/Brazil variant (Table 13 and Figure 6). The frequent occurrence of amino acid rotamers, valine (VAL), and proline (PRO) has been detected in all mutant bacterial AMPs.      Amino acid residues of the wild-type protein appear in purple, whereas rotamers of amino acid residues of mutant protein appear in gray.  The AMPs with the highest affinities for the viral proteins were mutated and the mutants were then evaluated for their affinity for the viral proteins. The increased binding affinities of the mutants are indicated in bold.

Discussion
The multitude of emerging viruses such as dengue, Zika, Ebola, monkeypox, and the recent occurrence of the COVID-19 pandemic show that the current antiviral therapeutic arsenal is not sufficient [2,4]. Vaccine development is costly and time-consuming, and large-scale administration can be difficult [29]. Furthermore, the development of antiviral treatments is hampered by the high evolutionary power of certain viruses such as coronaviruses or the Ebola virus. All this highlights the need to develop new broad-spectrum antiviral molecules, i.e., active against a wide range of viruses [30][31][32]. Ideally, these broadspectrum antiviral treatments should be able to be active against new, totally unknown viruses or emerging viruses that are mutants or variants of known viruses.
Dengue, Ebola, Zika, and SARS-CoV-2 viruses use their structural surface proteins to attach and induce entry into host cells and initiate pathogenesis. Therefore, a therapeutic agent capable of inhibiting the attachment of these viruses to their corresponding host cell receptor would block the initiation of viral replication and early infection [33,34]. Viral surface proteins that bind to the surface receptor of the target cell are good candidates for vaccine development because they stimulate the production of blocking antibodies [35]. Various natural biomolecules are capable of blocking in vitro viral infection of a given virus type, usually by blocking the action of the replication machinery of the viral genome [36,37]. The present study was based on the assumption that blocking viral surface proteins and/or their corresponding cell receptors could also be achieved with natural biomolecules. We selected various natural biomolecules of bacterial or plant origin for which data in the literature have shown antiviral properties against different viruses. We tested in silico their ability to interact with proteins involved in viral recognition of cell surface receptors of the main emerging viruses. Thus, we targeted viral envelope proteins of dengue, Ebola, Zika, and SARS-CoV-2 viruses and corresponding cell surface receptors, in search of broadspectrum antiviral molecules. Given its recent emergence in various countries, we also looked at the monkeypox virus. We relied on the few data published and/or accessible in databases to select the viral protein A42R and the cellular receptors Toll-like receptor 5, CR3/Mac-1, CD36, and FcγRIIA. Figure 7 illustrates the proposed mode of antiviral action of these biomolecules, which consists of blocking either the viral surface proteins or the corresponding cell receptors, or both.
to attach and induce entry into host cells and initiate pathogenesis. Therefore, a therapeutic agent capable of inhibiting the attachment of these viruses to their corresponding host cell receptor would block the initiation of viral replication and early infection [33,34]. Viral surface proteins that bind to the surface receptor of the target cell are good candidates for vaccine development because they stimulate the production of blocking antibodies [35]. Various natural biomolecules are capable of blocking in vitro viral infection of a given virus type, usually by blocking the action of the replication machinery of the viral genome [36,37]. The present study was based on the assumption that blocking viral surface proteins and/or their corresponding cell receptors could also be achieved with natural biomolecules. We selected various natural biomolecules of bacterial or plant origin for which data in the literature have shown antiviral properties against different viruses. We tested in silico their ability to interact with proteins involved in viral recognition of cell surface receptors of the main emerging viruses. Thus, we targeted viral envelope proteins of dengue, Ebola, Zika, and SARS-CoV-2 viruses and corresponding cell surface receptors, in search of broad-spectrum antiviral molecules. Given its recent emergence in various countries, we also looked at the monkeypox virus. We relied on the few data published and/or accessible in databases to select the viral protein A42R and the cellular receptors Toll-like receptor 5, CR3/Mac-1, CD36, and FcγRIIA. Figure 7 illustrates the proposed mode of antiviral action of these biomolecules, which consists of blocking either the viral surface proteins or the corresponding cell receptors, or both. Silvestrol, a secondary metabolite derived from Aglaia spp. showed broad-spectrum antiviral potential in the in silico study conducted as it was predicted to interact with viral surface proteins of two viral families (ZIKV, and some SARS-CoV-2 variants). However, binding predictions only showed a high docking score with DENV cellular receptor for silvestrol. Thus, silvestrol could block the interaction of DENV with target cells by binding to the cell receptor. For the ZIKV, and SARS-CoV-2 viruses, silvestrol would block solely viral surface proteins. Other mechanisms for the antiviral action of silvestrol have been previously reported in the literature [38][39][40]. A study conducted by Müller et al. showed that silvestrol inhibits the replication of HCoV-229E and MERS-CoV by suspending capdependent viral mRNA translation [41]. Various studies have shown that silvestrol can inhibit the eIF4A-dependent translation of viral mRNA of EBOV, ZIKV and hepatitis E virus [42,43]. A study by Henss et al. also showed that silvestrol can delay protein synthesis of the chikungunya virus and reduce viral RNA replication [44]. A recent investigation showed that silvestrol at a concentration of 10 nM reduces viral titers of SARS-CoV-2 up to 100-fold in infected human bronchial epithelial cells [45].
The phytochemical compound andrographolide, derived from A. paniculate, also showed broad-spectrum antiviral potential in the present in silico study. As with silvestrol, high docking scores were obtained when andrographolide was tested against the cellular receptors of DENV. Binding predictions also showed strong docking scores with the surface protein of EBOV and two variants of SARS-CoV-2. Thus, andrographolide could block the interaction of DENV with target cells by interacting with the corresponding cellular receptors. For EBOV and SARS-CoV-2, andrographolide would block solely viral surface proteins. Data from the literature have shown that andrographolide has various antiviral properties [46]. Andrographolide inhibits the replication of DENV and reduces infection in human HepG2 and HeLa cells [47]. Ethanol extracts of A. paniculate containing andrographolide inhibit the activity of the simian retrovirus in human A549 cells [48]. Andrographolide reduces CHIKV infection in human HepG2 cells by interfering with viral protein synthesis [49]. Andrographolide inhibits the expression of viral enveloped glycoproteins C and D of herpes simplex virus type 1 [50]. Andrographolide inhibits the activity of DENV and ZIKV [51]. Andrographolide decreases the viral load of SARS-CoV-2 in human Calu-3 cells [52] and is thought to inhibit the main protease of SARS-CoV-2 [53,54].
Secondary metabolites isolated from cyanobacteria are known for their potential antiviral activity against viral pathogens such as HIV, measles virus, adenovirus, influenza, herpes simplex virus, and Coxsackie [55][56][57]. A recent in silico study conducted by Aminu et al. showed that an indole alkaloid compound known as hapalindole derived from marine cyanobacteria has strong docking energy with SARS-CoV-2 spike protein [58]. In our study, hapalindole was found to have a high docking energy not only to surface proteins of different variants of SARS-CoV-2 but also to those of ZIKV and EBOV. Binding predictions also showed strong docking scores with the cellular receptors of DENV and EBOV for hapalindole. Thus hapalindole may have antiviral properties against EBOV by blocking its envelope glycoprotein and the corresponding receptor, antiviral properties against SRARS-CoV-2 and ZIKV by blocking only viral envelope proteins and DENV by blocking the cell receptor.
Glycocin F was the only AMP, and more broadly, the only natural biomolecule tested to show high binding energies to all viral surface proteins and corresponding cell virus receptors. Lactococcin G and plantaricin ASM1 also showed promising broad-spectrum in silico antiviral potential by targeting all the surface viral proteins and most of the corresponding cell surface receptors. The binding affinities obtained between these three AMPs and the corresponding viral surface proteins or cell receptors were very high (from −825.2 to −1756.7 kcal/mol). Among the AMPs tested, the lantibiotic bacteriocin produced by L. lactis known as nisin, is the only FDA-approved microbial-derived AMP up to date [59,60]. The antiviral activity of nisin against the bovine viral diarrhea virus (BVDV) has already been demonstrated [61]. An in silico study conducted by Balmeh et al. showed that bacteriocin glycocin F derived from L. lactis and bacteriocin plantaricin ASM1 derived from L. plantarum have high docking energy with SARS-CoV-23CL protease, RNA-dependent RNA polymerase RdRp and spike (S) envelope protein [19]. Surfactin is a powerful natural antimicrobial derived from B. subtilis, which suppresses the proliferation of porcine epidemic diarrhea virus (PEDV) and transmissible swine gastroenteritis virus (TGEV) in epithelial cells by inhibiting viral membrane fusion with host cells at concentrations between 15-50 µg/mL [62]. This mechanism of action is consistent with the in silico results presented here showing that gardimycin, glycocin F, lactococcin G, and plantaricin ASM1 may interact with the surface proteins of various viruses and different variants of the SARS-COV-2 virus, and with their corresponding cell surface receptors. This could explain recently published results showing that AMPs produced by probiotic strains of Lactobacillus acidophilus reduce symptoms of hospitalized patients with COVID-19 and improve antibody production against SARS-CoV-2 [60,61,63,64].

Ligands
A total of 19 ligands, including 9 reference molecules (4 antibodies and 5 antiviral drugs), 6 AMPs of bacterial origin, 2 phytochemicals, and 2 cyanobacterial secondary metabolites, were selected for the molecular docking experiments (Table S1). The anti-bodies and cellular virus receptors were used as positive controls for interaction with corresponding viral surface proteins. Antiviral drugs targeting enzymes involved in viral genome replication were used as negative controls for interaction with viral surface proteins. The three-dimensional (3D) structures of the selected macromolecules were downloaded from the Research Collaboratory for Structural Bioinformatics Protein (RCSB) PDB database (https://www.rcsb.org/) accessed on 10 August 2021 and from the National Library of Medicine (NLM) PubChem PDB database (https://pubchem.ncbi.nlm.nih.gov) accessed on 10 August 2021 in SDF format.

Modeling and Preparation of Selected Macromolecules
The PDB structures of viral proteins and phytochemical compounds were modified using AutoDockTools (version 1.5.7, La Jolla, CA, USA), where water molecules were discarded and hydrogen bonds and Kollman charges were added. The modified SDF and PDB files were converted to Protein Data Bank, Partial Charge (Q), and Atom Type (T) (PDBQT) format before being analyzed to optimize docking efficiency.

Molecular Docking Analysis
The molecular docking of smaller macromolecules such as phytochemical compounds with the selected viral structural proteins was performed by using AutoDock Vina (version 1.1.2, La Jolla, CA, USA). The docking was performed at a default grid box dimension of 40 Å × 40 Å × 40 Å and the energy range was set at 4 and exhaustiveness was set at 8 [65]. In the case of larger macromolecules, i.e., bacterial AMPs and reference antibodies, molecular docking with the selected viral structural proteins was performed by the application of the ClusPro 2.0 supercomputer-based online server [66][67][68][69].

Estimation of Binding Free Energy/Docking Energy and Determination of the Root Mean Square Distance
AutoDock Vina software and ClusPro online server were used to calculate and estimate the docking energies/binding free energies of ligand-receptor complexes in kcal/mol. The balanced output of model rank 0 result of ClusPro was selected as the most accurate output. Whereas, the Autodock Vina log file of the docked molecule indicating the binding affinity allocated to a root mean square distance (RMSD) value of zero was selected as the best result.

Simulation of Molecular Ligand-Receptor Interactions
Simulation and visualization of molecular interactions indicating the active binding sites of ligand-receptor complexes and their amino acid sequences were performed using PyMOL (version 2.5.2) molecular visualization system [70].

Mutation and Structure Modeling of Bacterial AMPs
The two AMPs bacteriocin glycocin F and plantaricin ASM1, which showed the highest docking scores with certain viral surface proteins, were subjected to mutation by initiating the mutagenesis feature of the PyMOL software. Mutagenesis was performed in order to determine the possibility of further increasing the docking strength of bacterial AMPs. The basis of PyMOL mutagenesis involves the replacement of amino acid residues of AMPs with their corresponding rotamers that denote the highest percentage of mutation probability. Mutant AMPs remodeled as ligands were re-docked using ClusPro with their corresponding viral protein receptors to determine the increase in binding energies induced by the effects of the mutations. Interaction figures of the remodeled peptides of the mutant AMPs were designed and illustrated by PyMOL software [19,70].

Conclusions
In conclusion, glycocin F is the natural biomolecule with the highest potential to develop a broad-spectrum antiviral agent. Remarkably, the binding energies of glycocin F for surface viral proteins are at least as high as those of cell virus receptors. This highlights the interest in further studying the antiviral activity of these natural molecules to develop broad-spectrum antiviral agents. The interaction of glycocin F with different viral surface proteins and their cell receptors should now be demonstrated experimentally. As indicated by the in silico mutagenesis results, the sequence of this antimicrobial peptide can certainly be optimized, this will also have to be developed experimentally.