Dieckol and Its Derivatives as Potential Inhibitors of SARS-CoV-2 Spike Protein (UK Strain: VUI 202012/01): A Computational Study

The high risk of morbidity and mortality associated with SARS-CoV-2 has accelerated the development of many potential vaccines. However, these vaccines are designed against SARS-CoV-2 isolated in Wuhan, China, and thereby may not be effective against other SARS-CoV-2 variants such as the United Kingdom variant (VUI-202012/01). The UK SARS-CoV-2 variant possesses D614G mutation in the Spike protein, which impart it a high rate of infection. Therefore, newer strategies are warranted to design novel vaccines and drug candidates specifically designed against the mutated forms of SARS-CoV-2. One such strategy is to target ACE2 (angiotensin-converting enzyme2)–Spike protein RBD (receptor binding domain) interaction. Here, we generated a homology model of Spike protein RBD of SARS-CoV-2 UK strain and screened a marine seaweed database employing different computational approaches. On the basis of high-throughput virtual screening, standard precision, and extra precision molecular docking, we identified BE011 (Dieckol) as the most potent compounds against RBD. However, Dieckol did not display drug-like properties, and thus different derivatives of it were generated in silico and evaluated for binding potential and drug-like properties. One Dieckol derivative (DK07) displayed good binding affinity for RBD along with acceptable physicochemical, pharmacokinetic, drug-likeness, and ADMET properties. Analysis of the RBD–DK07 interaction suggested the formation of hydrogen bonds, electrostatic interactions, and hydrophobic interactions with key residues mediating the ACE2–RBD interaction. Molecular dynamics simulation confirmed the stability of the RBD–DK07 complex. Free energy calculations suggested the primary role of electrostatic and Van der Waals’ interaction in stabilizing the RBD–DK07 complex. Thus, DK07 may be developed as a potential inhibitor of the RBD–ACE2 interaction. However, these results warrant further validation by in vitro and in vivo studies.


Introduction
The WHO (World Health Organization) declared the outbreak of SARS-CoV-2 (severe acute respiratory syndrome coronavirus 2) as public health emergency of international concern on 30 January 2020, and a pandemic on 11 Mar 2020, due to the rapid spread of the virus and absence of any preexisting medications [1]. SARS-CoV-2 causes COVID-19, a disease of the respiratory system with symptoms varying from mild flu-like to more severe respiratory disorders characterized by high fever, body ache, tiredness, cough, antiviral drugs (Remdesivir, Ritonavir, Lopinavir, Ribavirin, and Fapiravir), Ivermectin, etc. As of Mar 2021, 12 vaccines have been authorized for public use as a preventive measure against COVID-19 by at least one national regulatory authority. These vaccines  19. In addition, other approaches such as plasma therapy, decoy-soluble ACE2 proteins, drug repurposing, natural products, and blocking peptides are underway. However, developing a new therapeutic agent is a time-consuming, highly challenging, and costly. Thus, there is a pressing need to identify a specific drug or vaccine against various SARS-CoV-2 proteins and develop them for the therapeutic of COVID- 19. In this study, we employed computational tools to screen a library of marine seaweed compounds against the Spike protein RBD of the SARS-CoV-2 UK variant (VUI 202012/01). It is anticipated that the outcome of this study may contribute to the existing knowledge of anti-COVID-19 compounds and thus may contribute to the development of effective drugs against the SARS-CoV-2 UK variant.

Homology Modeling and Validation of UK SARS-CoV-2 RBD
The three-dimensional model of the UK SARS-CoV-2 strain was generated by performing homology modeling in SWISS-MODEL. On the basis of sequence identity and sequence coverage, we considered chain A of 6ZBP as a template. The sequence alignment of UK SARS-CoV-2 RBD and chain A of 6ZBP is shown in Figure 1A, and the normalized QMEAN4 score of the generated model is presented in Figure 1B. The QMEAN4 score of the generated model was less than 1, which implies that its three-dimensional structure was comparable to the non-redundant set of PDB structures. Further, the overall quality of the generated RBD model was evaluated with the help of PSVS software to enumerate parameters such as Ramachandran, Verify3D, Procheck, and Molprobity clash scores ( Table 1). Analysis of Ramachandran plot from Procheck and Richardson's lab suggests that all the residues of the RBD model were with most favored and additional/generously allowed regions ( Figure 1C and Table 1). It is noteworthy that none of the residues were in the disallowed regions. Moreover, RBD model and template superimposition suggested that RBD occupied a similar conformational space ( Figure 1D). The relative positions of Tyr453Phe and Asn501Tyr mutations are also depicted.    In structure-based drug design methodology, an extensive database comprising structurally diverse ligands is searched using HTVS and molecular docking to identify potential inhibitors [11,12]. In this study, we performed HTVS to screen compounds from a marine seaweed library against the RBD of SARS-CoV-2 Spike protein. In HTVS, 809 out of 1110 compounds present in the library (i.e., 72.9%) were able to bind RBD with binding energies varying from −6.951 to −0.333 kcal mol −1 . The compounds with docking energy ≤−4.000 kcal mol −1 in HTVS (i.e., 51 compounds) were subjected to SP docking (Table S1). On the basis of SP docking, compounds with ≤−5.000 kcal mol −1 docking energy (i.e., five compounds) were further shortlisted and screened by XP docking. These compounds were BE011, GA004, GA005, GA006, and GA007; their structure, source, and chemical nature are presented in Table 2. The XP docking energy of these five compounds was between −5.439 and −8.326 kcal mol −1 ( Table 2). On the basis of the lowest XP docking energy, we identified the compound BE011 (Dieckol) as the most potent binder of RBD and hence selected for further studies. Dieckol is abundantly isolated from the brown alga Ecklonia cava found on Jeju Island, Korea [13]. Dieckol has been reported to possess various biological properties such as anti-cancer, antithrombic and profibrinolytic activities, hepatoprotective, antioxidant, anti-diabetic, anti-hyperlipidemic, photochemopreventive, and anti-HIV [14][15][16][17][18][19][20][21][22][23]. Interestingly, Dieckol isolated from brown algae Ecklonia cava has been reported to possess inhibitory activity against 3CLpro or Mpro of SARS-CoV [24].

HTVS, SP, and XP Molecular Docking of Marine Seaweed Compounds against RBD of Spike Protein
In structure-based drug design methodology, an extensive database comprising structurally diverse ligands is searched using HTVS and molecular docking to identify potential inhibitors [11,12]. In this study, we performed HTVS to screen compounds from a marine seaweed library against the RBD of SARS-CoV-2 Spike protein. In HTVS, 809 out of 1110 compounds present in the library (i.e., 72.9%) were able to bind RBD with binding energies varying from −6.951 to −0.333 kcal mol −1 . The compounds with docking energy ≤ −4.000 kcal mol −1 in HTVS (i.e., 51 compounds) were subjected to SP docking (Table S1). On the basis of SP docking, compounds with ≤−5.000 kcal mol −1 docking energy (i.e., five compounds) were further shortlisted and screened by XP docking. These compounds were BE011, GA004, GA005, GA006, and GA007; their structure, source, and chemical nature are presented in Table 2. The XP docking energy of these five compounds was between −5.439 and −8.326 kcal mol −1 ( Table 2). On the basis of the lowest XP docking energy, we identified the compound BE011 (Dieckol) as the most potent binder of RBD and hence selected for further studies. Dieckol is abundantly isolated from the brown alga Ecklonia cava found on Jeju Island, Korea [13]. Dieckol has been reported to possess various biological properties such as anti-cancer, antithrombic and profibrinolytic activities, hepatoprotective, antioxidant, anti-diabetic, anti-hyperlipidemic, photochemopreventive, and anti-HIV [14][15][16][17][18][19][20][21][22][23]. Interestingly, Dieckol isolated from brown algae Ecklonia cava has been reported to possess inhibitory activity against 3CLpro or Mpro of SARS-CoV [24].

HTVS, SP, and XP Molecular Docking of Marine Seaweed Compounds against RBD of Spike Protein
In structure-based drug design methodology, an extensive database comprising structurally diverse ligands is searched using HTVS and molecular docking to identify potential inhibitors [11,12]. In this study, we performed HTVS to screen compounds from a marine seaweed library against the RBD of SARS-CoV-2 Spike protein. In HTVS, 809 out of 1110 compounds present in the library (i.e., 72.9%) were able to bind RBD with binding energies varying from −6.951 to −0.333 kcal mol −1 . The compounds with docking energy ≤ −4.000 kcal mol −1 in HTVS (i.e., 51 compounds) were subjected to SP docking (Table S1). On the basis of SP docking, compounds with ≤−5.000 kcal mol −1 docking energy (i.e., five compounds) were further shortlisted and screened by XP docking. These compounds were BE011, GA004, GA005, GA006, and GA007; their structure, source, and chemical nature are presented in Table 2. The XP docking energy of these five compounds was between −5.439 and −8.326 kcal mol −1 ( Table 2). On the basis of the lowest XP docking energy, we identified the compound BE011 (Dieckol) as the most potent binder of RBD and hence selected for further studies. Dieckol is abundantly isolated from the brown alga Ecklonia cava found on Jeju Island, Korea [13]. Dieckol has been reported to possess various biological properties such as anti-cancer, antithrombic and profibrinolytic activities, hepatoprotective, antioxidant, anti-diabetic, anti-hyperlipidemic, photochemopreventive, and anti-HIV [14][15][16][17][18][19][20][21][22][23]. Interestingly, Dieckol isolated from brown algae Ecklonia cava has been reported to possess inhibitory activity against 3CLpro or Mpro of SARS-CoV [24]. Table 2. Extra-precision (XP) molecular docking of selected ligands (having ≤ −5.000 kcal mol−1 in SP docking) against RBD of spike protein.

HTVS, SP, and XP Molecular Docking of Marine Seaweed Compounds against RBD of Spike Protein
In structure-based drug design methodology, an extensive database comprising structurally diverse ligands is searched using HTVS and molecular docking to identify potential inhibitors [11,12]. In this study, we performed HTVS to screen compounds from a marine seaweed library against the RBD of SARS-CoV-2 Spike protein. In HTVS, 809 out of 1110 compounds present in the library (i.e., 72.9%) were able to bind RBD with binding energies varying from −6.951 to −0.333 kcal mol −1 . The compounds with docking energy ≤ −4.000 kcal mol −1 in HTVS (i.e., 51 compounds) were subjected to SP docking (Table S1). On the basis of SP docking, compounds with ≤−5.000 kcal mol −1 docking energy (i.e., five compounds) were further shortlisted and screened by XP docking. These compounds were BE011, GA004, GA005, GA006, and GA007; their structure, source, and chemical nature are presented in Table 2. The XP docking energy of these five compounds was between −5.439 and −8.326 kcal mol −1 ( Table 2). On the basis of the lowest XP docking energy, we identified the compound BE011 (Dieckol) as the most potent binder of RBD and hence selected for further studies. Dieckol is abundantly isolated from the brown alga Ecklonia cava found on Jeju Island, Korea [13]. Dieckol has been reported to possess various biological properties such as anti-cancer, antithrombic and profibrinolytic activities, hepatoprotective, antioxidant, anti-diabetic, anti-hyperlipidemic, photochemopreventive, and anti-HIV [14][15][16][17][18][19][20][21][22][23]. Interestingly, Dieckol isolated from brown algae Ecklonia cava has been reported to possess inhibitory activity against 3CLpro or Mpro of SARS-CoV [24]. Table 2. Extra-precision (XP) molecular docking of selected ligands (having ≤ −5.000 kcal mol−1 in SP docking) against RBD of spike protein.

Investigation of Physicochemical Pharmacokinetic, Drug-Like, and Medicinal Properties
Determining physicochemical, pharmacokinetics, drug-like, and medicinal properties using in silico tools is widely accepted as a fast and accurate method [25]. In this study, the physicochemical and ADMET properties of the compound with lowest XP docking energy, i.e., BE011 or Dieckol, was determined. The results suggested that Dieckol was not suitable to be developed as a candidate drug molecule owing to the violations of Lipinski's, Ghose's Veber's Egan's, and Muegge's rules (Tables 3-5). Lipinski's rule of five suggests that an orally active drug should not have more than one violation of the following rules: molecular weight ≤ 500 Da, number of hydrogen bond donor ≤ 5, hydrogen bond acceptor ≤ 10, XlogP3 ≤ 5. We observed that Dieckol violated three Lipinski's rules as it had a molecular weight of 742.55 Da, 11 hydrogen bond donors, and 18 hydrogen bond acceptors (Table 3). Moreover, Dieckol is not soluble and has poor gastrointestinal absorption. To improve the physicochemical and ADMET properties, we generated 10 derivatives of Dieckol (DK01-10) in silico using ChemSketch and optimized their energies using OPLS3e forcefield. The physicochemical and ADMET properties of all the Dieckol derivatives were determined using SwissADME, and the results are presented in Tables 3-5.
We observed that amongst all the Dieckol derivatives, DK07 (IUPAC name is 8-{3hydroxy-4-[(7-hydroxynaphthalen-2-yl)oxy]phenoxy}-1,4-benzodioxin-5-ol) showed acceptable drug-like properties. The physicochemical properties of DK07, such as molecular weight, rotatable bonds, hydrogen bond donors, hydrogen bond acceptors, and XlogP3, were 416.38 Da, 4, 3, 9, and 4.88, respectively (Table 3). Most importantly, DK07 was moderately soluble and had a total polar surface area of 97.61 Å 2 . It displayed high gastrointestinal absorption and no blood-brain barrier crossing ability, and did not act as a Pglycoprotein substrate. DK07 acted as an inhibitor of CYP2C19, CYP2C9, CYP2D6, and CYP3A4. The skin permeability of DK07 was estimated to be −5.38 cm/s (Table 4). Most importantly, DK07 had only 1 Ghose's violation and obeyed all the other drug-likeness filters such as Lipinski's, Veber, Egan's, and Muegge's rules. It did not display any PAINS and Brenk alerts, although there were two lead-likeness alerts (Mol wt >350, and XLOGP3 >3.5). The bioavailability score and synthetic accessibility of DK07 were 0.55 and 3.70, respectively (Table 5). Since DK07 obeyed all the properties of a drug-like molecule and its ADMET properties were within acceptable limits, we analyzed the interaction between RBD and DK07 in detail and performed molecular dynamics and free energy calculation.

Investigation of Physicochemical Pharmacokinetic, Drug-Like, and Medicinal Properties
Determining physicochemical, pharmacokinetics, drug-like, and medicinal properties using in silico tools is widely accepted as a fast and accurate method [25]. In this study, the physicochemical and ADMET properties of the compound with lowest XP docking energy, i.e., BE011 or Dieckol, was determined. The results suggested that Dieckol was not suitable to be developed as a candidate drug molecule owing to the violations of Lipinski's, Ghose's Veber's Egan's, and Muegge's rules (Tables 3-5). Lipinski's rule of five suggests that an orally active drug should not have more than one violation of the following rules: molecular weight ≤ 500 Da, number of hydrogen bond donor ≤ 5, hydrogen bond acceptor ≤ 10, XlogP3 ≤ 5. We observed that Dieckol violated three Lipinski's rules as it had a molecular weight of 742.55 Da, 11 hydrogen bond donors, and 18 hydrogen bond acceptors (Table 3). Moreover, Dieckol is not soluble and has poor gastrointestinal absorption. To improve the physicochemical and ADMET properties, we generated 10 derivatives of Dieckol (DK01-10) in silico using ChemSketch and optimized their energies using OPLS3e forcefield. The physicochemical and ADMET properties of all the Dieckol derivatives were determined using SwissADME, and the results are presented in Tables 3-5.    We observed that amongst all the Dieckol derivatives, DK07 (IUPAC name is 8-{3hydroxy-4-[(7-hydroxynaphthalen-2-yl)oxy]phenoxy}-1,4-benzodioxin-5-ol) showed acceptable drug-like properties. The physicochemical properties of DK07, such as molecular weight, rotatable bonds, hydrogen bond donors, hydrogen bond acceptors, and XlogP3, were 416.38 Da, 4, 3, 9, and 4.88, respectively (Table 3). Most importantly, DK07 was moderately soluble and had a total polar surface area of 97.61 Å 2 . It displayed high gastrointestinal absorption and no blood-brain barrier crossing ability, and did not act as a P-glycoprotein substrate. DK07 acted as an inhibitor of CYP2C19, CYP2C9, CYP2D6, and CYP3A4. The skin permeability of DK07 was estimated to be −5.38 cm/s (Table 4). Most importantly, DK07 had only 1 Ghose's violation and obeyed all the other drug-likeness filters such as Lipinski's, Veber, Egan's, and Muegge's rules. It did not display any PAINS and Brenk alerts, although there were two lead-likeness alerts (Mol wt >350, and XLOGP3 >3.5). The bioavailability score and synthetic accessibility of DK07 were 0.55 and 3.70, respectively (Table 5). Since DK07 obeyed all the properties of a drug-like molecule and its ADMET properties were within acceptable limits, we analyzed the interaction between RBD and DK07 in detail and performed molecular dynamics and free energy calculation.

Interaction of Spike Protein RBD with ACE2 and Dieckol Derivative DK07
The RBD of SARS-CoV-2 Spike protein has emerged as the most suitable target for drug designing and development. It plays an essential role in recognizing ACE2 receptors on the host cell, thus mediating the internalization of viral RNA into the host. Here, we retrieved the X-ray crystal structure (PDB Id: 6M0J) of RBD in complex with ACE2 and analyzed the interaction pattern using BIOVIA Discovery Studio Visualizer [26]. We found that RBD and ACE2 interact through salt bridges, hydrogen bonds, and hydrophobic interactions ( Table 6). Two salt bridges were formed by ACE2 Lys417:NZ with RBD Asp30:OD1 (3.09 Å) and Asp:OD2 (3.00 Å), while an electrostatic interaction was formed between Lys31:NZ of ACE2 and Glu484:OE1 of RBD (4.39 Å). Nine conventional hydrogen bonds were mediated by Gln24:OE1, Glu35:OE1, Asp38:OD2, Tyr41:OH, Gln42:NE2 (two bonds), Tyr83:OH, Lys353:NZ, and Lys353:O of ACE2 with Gly446:O, Tyr449:OH, Tyr449:OH, Asn487:OD1, Asn487:ND2, Gln493:NE2:B, Gly496:O, Thr500:OG1, and Gly502:N of RBD with a distance in the range of 2.69-3.24 Å (Table 6). We also observed one pi-donor hydrogen bond between ACE2 Tyr83:OH and RBD Phe486 (4.09 Å). The ACE2-RBD complex was further stabilized by six hydrophobic interactions between Lys31, His34, Met82, Tyr83, Lys353:C,O, and Gly354:N of ACE2 and Leu455:CD1, Phe486, Tyr489, and Tyr505 of RBD, with a distance in the range of 3.85-5.14 Å ( Table 6). The previous report also suggested that a small stretch of 23 amino acid residues (Glu23 to Leu45) in the peptidase (PD) domain of ACE2 to be majorly involved in the interaction with RBD of Spike protein [27]. (chain E of 6M0J) was performed using "Glide-2018" (Schrodinger, LLC, New York, NY, USA). An analysis of binding pose and interaction pattern between DK07 and Spike protein RBD revealed that DK07 was bound to the ACE2 binding groove of RBD through multiple interactions such as electrostatic interaction, hydrogen bonding, and hydrophobic interactions (Figure 2). The amino acid residues Arg403:NH2 and Lys417:HZ3 formed pication electrostatic interactions with DK07. We observed four strong conventional hydrogen bonds formed by Glu406:OE2 (2.48 Å), Arg408:HE (2.11 Å), Arg408:HH21 (2.41 Å), and Gly496:O (2.43 Å), and three pi-donor hydrogen bonds formed by Arg403:NH2 (4.10 Å), Lys417:HZ3 (3.15 Å), and Gly496:HN (3.12 Å). Moreover, two pi-pi T-shaped hydrophobic interactions were formed by Tyr505 (5.92 and 5.17 Å), while Lys417:CD formed a pi-sigma hydrophobic interaction (3.83 Å). Other residues such as Asp405, Arg408, Gln409, Phe453, Tyr495, Phe497, Tyr501, and Tyr505 were engaged in Van der Waals' interaction with DK07. Although Van der Waal's force is usually weak, it plays significant role in enhancing the attraction and binding of ligands to protein. Interestingly, the amino acid residues Lys417, Gly496, and Tyr505 of Spike protein RBD were found to be commonly involved in the interaction with ACE2 and DK07, thus suggesting that DK07 was bound at the same location where ACE2 interacts with Spike protein RBD. The analysis of docking energy (-9.254 kcal mol −1 ) and the corresponding binding affinity (6.13 × 10 6 M −1 ) of DK07 towards Spike protein RBD suggested a strong interaction between RBD and DK07 and hence strong inhibition potential of DK07 (

MD (Molecular Dynamics) Simulation Analysis
MD simulation is a powerful tool to analyze the structure and dynamics of the protein-ligand complex. We performed 50 ns MD simulation of the RBD-DK07 complex and measured various parameters, as described below.

MD (Molecular Dynamics) Simulation Analysis
MD simulation is a powerful tool to analyze the structure and dynamics of the proteinligand complex. We performed 50 ns MD simulation of the RBD-DK07 complex and measured various parameters, as described below.

RMSD (Root Mean Square Deviation) Calculations
In MD simulation, RMSD is measured as a deviation in the structure of protein or protein-ligand complex with respect to a reference structure usually the initial frame. Figure 3A shows RMSD in Cα atoms of RBD alone (teal color) and RBD-DK07 complex (brown line). As compared to the initial frame, no significant fluctuations were observed in RMSD values of protein and protein-inhibitor complex throughout the simulation time. The mean RMSD values of RBD alone or in complex with DK07 were obtained as 1.5575 and 1.6305 Å, respectively. Since the variation in RMSD values of protein and protein-inhibitor complex were much lower than the acceptable limit of 2.0 Å, forming a stable RBD-DK07 complex was anticipated ( Figure 3A).

RMSD (Root Mean Square Deviation) Calculations
In MD simulation, RMSD is measured as a deviation in the structure of protein or protein-ligand complex with respect to a reference structure usually the initial frame. Figure 3A shows RMSD in Cα atoms of RBD alone (teal color) and RBD-DK07 complex (brown line). As compared to the initial frame, no significant fluctuations were observed in RMSD values of protein and protein-inhibitor complex throughout the simulation time. The mean RMSD values of RBD alone or in complex with DK07 were obtained as 1.5575 and 1.6305 Å, respectively. Since the variation in RMSD values of protein and proteininhibitor complex were much lower than the acceptable limit of 2.0 Å, forming a stable RBD-DK07 complex was anticipated ( Figure 3A).

RMSF (Root Mean Square Fluctuation) Calculations
In MD simulation, RMSF value of a protein is generally measured to access the fluctuations in the side chains due to the binding of a ligand. Figure 3B depicts the RMSF of Spike protein RBD (teal color) in the presence of DK07 during MD simulation and compared it with the experimentally determined B-factor (red color) obtained during X-ray crystallography. Minor fluctuations in RMSF values of RBD side chains might have been due to the entry and binding of DK07 into the groove of the protein. Throughout the MD simulation, the RMSF values coincided with the B-factor values, thus suggesting that the binding of DK07 did not alter the overall conformation of RBD. , as compared with Bfactor, which was determined during X-ray crystallography (teal color).

Secondary Structure Analysis
The interaction between a protein and inhibitor may alter the secondary structure elements (SSE) of the protein. In this study, we monitored the changes in SSE of RBD due to the binding of DK07 during simulation ( Figure 4A). The total SSE of RBD in complex with DK07 was 29% (α-helix = 7% and β-sheets = 22%), which was in agreement with the

RMSF (Root Mean Square Fluctuation) Calculations
In MD simulation, RMSF value of a protein is generally measured to access the fluctuations in the side chains due to the binding of a ligand. Figure 3B depicts the RMSF of Spike protein RBD (teal color) in the presence of DK07 during MD simulation and compared it with the experimentally determined B-factor (red color) obtained during X-ray crystallography. Minor fluctuations in RMSF values of RBD side chains might have been due to the entry and binding of DK07 into the groove of the protein. Throughout the MD simulation, the RMSF values coincided with the B-factor values, thus suggesting that the binding of DK07 did not alter the overall conformation of RBD.

Secondary Structure Analysis
The interaction between a protein and inhibitor may alter the secondary structure elements (SSE) of the protein. In this study, we monitored the changes in SSE of RBD due to the binding of DK07 during simulation ( Figure 4A). The total SSE of RBD in complex with DK07 was 29% (α-helix = 7% and β-sheets = 22%), which was in agreement with the reported values of SSEs 32% (α-helix = 11% and 21% β-sheets). The results indicated that the binding of DK07 to RBD did not considerably modify its secondary structure. reported values of SSEs 32% (α-helix = 11% and 21% β-sheets). The results indicated that the binding of DK07 to RBD did not considerably modify its secondary structure.

Protein-Ligand Interaction Analysis
The total number of contacts between RBD and DK07 during simulation was determined to vary in the 4-20 range, with an average of 12 contacts ( Figure 4B). Moreover, the involvement of amino acid residues in making contact with DK07 during simulation showed that Arg403, Glu406, Lys417, Tyr453, and Tyr505 were involved in making contact for most of the simulation ( Figure 4C).

Radius of Gyration (rGyr) and Different Surface Area Analysis
The radius of gyration (rGyr) is considered a significant indicator of the protein's folding state in different conditions. Here, the rGyr of the RBD-DK07 complex was measured to gain an insight into the compactness of protein during simulation ( Figure 5A). The rGyr of RBD fluctuated between 5.16 and 6.75 Å, with an average value of 5.84 Å. The solvent-exposed surface area of a protein under different conditions is generally accessed to look for any conformational changes [28]. Here, molecular surface area (MolSA), solvent accessible surface area (SASA), and polar surface area (PSA) of RBD in complex with DK07 were measured during simulation to explore the exposure of the protein to the solvent molecules and thus to access its conformational stability ( Figure 5B-D). MolSA, SASA, and PSA of DK07 were in the range of 508.36-585.17, 343.89-636.88, and 463.26-598.53 Å 2 , respectively. The average values of MSA, SASA, and PSA were estimated to be 553.62, 475.11, and 534.58 Å 2 , respectively. Although the values of surface areas fluctuated for the initial part of the simulation, they became stabilized and remained within acceptable error once favorable contacts were made between DK07 and RBD. The results of rGyr and surface areas confirmed the formation of a stable RBD-DK07 complex.

Protein-Ligand Interaction Analysis
The total number of contacts between RBD and DK07 during simulation was determined to vary in the 4-20 range, with an average of 12 contacts ( Figure 4B). Moreover, the involvement of amino acid residues in making contact with DK07 during simulation showed that Arg403, Glu406, Lys417, Tyr453, and Tyr505 were involved in making contact for most of the simulation ( Figure 4C).

Radius of Gyration (rGyr) and Different Surface Area Analysis
The radius of gyration (rGyr) is considered a significant indicator of the protein's folding state in different conditions. Here, the rGyr of the RBD-DK07 complex was measured to gain an insight into the compactness of protein during simulation ( Figure 5A). The rGyr of RBD fluctuated between 5.16 and 6.75 Å, with an average value of 5.84 Å. The solvent-exposed surface area of a protein under different conditions is generally accessed to look for any conformational changes [28]. Here, molecular surface area (MolSA), solvent accessible surface area (SASA), and polar surface area (PSA) of RBD in complex with DK07 were measured during simulation to explore the exposure of the protein to the solvent molecules and thus to access its conformational stability ( Figure 5B-D). MolSA, SASA, and PSA of DK07 were in the range of 508.36-585.17, 343.89-636.88, and 463.26-598.53 Å 2 , respectively. The average values of MSA, SASA, and PSA were estimated to be 553.62, 475.11, and 534.58 Å 2 , respectively. Although the values of surface areas fluctuated for the initial part of the simulation, they became stabilized and remained within acceptable error once favorable contacts were made between DK07 and RBD. The results of rGyr and surface areas confirmed the formation of a stable RBD-DK07 complex.

Free Energy (Prime/MM-GBSA) Calculation of RBD-DK07 Interaction
The binding of a ligand into the pocket of a protein is governed by the thermodynamic contribution of ligand to the overall binding free energy of the protein-ligand complex. Thus, the molecular interactions formed between ligand and protein play a significant role in determining the overall stability and affinity of a ligand inside the binding pocket of protein [27]. Therefore, we performed Prime/MM-GBSA to ascertain the binding thermodynamics of DK07 towards RBD. The binding free energy (ΔGbind) of DK07 towards RBD was estimated to be −52.87 kcal mol −1 , which reflected a strong interaction between DK07 and RBD (Table 7). Although the ΔGbind of the RBD-DK07 complex was significantly higher than the docking ΔG of RBD-DK07 interaction, it might have been due to the limitations of forcefield (OPLS3e) used in the analysis. Prime calculates MM-GBSA using the relative binding affinity of protein-ligand without considering any simulation process. Furthermore, previous reports suggest a disagreement between docking ΔG and MM-GBSA ΔGbind determined using Prime module [29][30][31]. Many ligands have been shown to bind the RBD of Spike protein with an estimated value of ΔGbind in the range of −21.45 to −54.11 kcal/mol, which is close to our estimated value [32,33]. Interestingly, the Prime/MM-GBSA values of lomitapide, dihydroergotamine mesylate, and olaparib against the Spike protein of SARS-CoV-2 have been estimated to be −98.46, −84.67, and −80.19, respectively [34]. The decomposition of binding free energy (ΔGbind) into its constituents revealed that gas phase energy (ΔEgas = −127.75 kcal mol −1 ) favors the formation of the DK07-RBD complex, while solvation energy (ΔGsol = 74.88 kcal mol −1 ) opposed it. Moreover, the polar (ΔGpolar) and non-polar (ΔGnon-polar) solvation energies, which comprised overall solvation energy (ΔGsol) of DK07-RBD interaction were found to be 78.67 and −3.79 kcal mol −1, respectively. Further, the gas phase energy (ΔEgas) constituents such as Van der Waal's energy (ΔEvdW = −58.16 kcal mol −1 ) and electrostatic or Coulomb energy (ΔEelec = −69.99 kcal mol −1 ) stabilized the DK07-RBD complex. These results were in agreement with the molecular docking results, suggesting that electrostatic interactions between DK07 and Arg403:NH1 and Glu406:OE2 along with Van der Waals' interactions favored the binding of DK07 to RBD of Spike protein and stabilized the RBD-DK07 complex.

Free Energy (Prime/MM-GBSA) Calculation of RBD-DK07 Interaction
The binding of a ligand into the pocket of a protein is governed by the thermodynamic contribution of ligand to the overall binding free energy of the protein-ligand complex. Thus, the molecular interactions formed between ligand and protein play a significant role in determining the overall stability and affinity of a ligand inside the binding pocket of protein [27]. Therefore, we performed Prime/MM-GBSA to ascertain the binding thermodynamics of DK07 towards RBD. The binding free energy (∆G bind ) of DK07 towards RBD was estimated to be −52.87 kcal mol −1 , which reflected a strong interaction between DK07 and RBD (Table 7). Although the ∆G bind of the RBD-DK07 complex was significantly higher than the docking ∆G of RBD-DK07 interaction, it might have been due to the limitations of forcefield (OPLS3e) used in the analysis. Prime calculates MM-GBSA using the relative binding affinity of protein-ligand without considering any simulation process. Furthermore, previous reports suggest a disagreement between docking ∆G and MM-GBSA ∆G bind determined using Prime module [29][30][31]. Many ligands have been shown to bind the RBD of Spike protein with an estimated value of ∆G bind in the range of −21.45 to −54.11 kcal/mol, which is close to our estimated value [32,33]. Interestingly, the Prime/MM-GBSA values of lomitapide, dihydroergotamine mesylate, and olaparib against the Spike protein of SARS-CoV-2 have been estimated to be −98.46, −84.67, and −80.19, respectively [34]. The decomposition of binding free energy (∆G bind ) into its constituents revealed that gas phase energy (∆E gas = −127.75 kcal mol −1 ) favors the formation of the DK07-RBD complex, while solvation energy (∆G sol = 74.88 kcal mol −1 ) opposed it. Moreover, the polar (∆G polar ) and non-polar (∆G non-polar ) solvation energies, which comprised overall solvation energy (∆G sol ) of DK07-RBD interaction were found to be 78.67 and −3.79 kcal mol −1, respectively. Further, the gas phase energy (∆E gas ) constituents such as Van der Waal's energy (∆E vdW = −58.16 kcal mol −1 ) and electrostatic or Coulomb energy (∆E elec = −69.99 kcal mol −1 ) stabilized the DK07-RBD complex. These results were in agreement with the molecular docking results, suggesting that electrostatic interactions between DK07 and Arg403:NH1 and Glu406:OE2 along with Van der Waals' interactions favored the binding of DK07 to RBD of Spike protein and stabilized the RBD-DK07 complex. All the energies are in kcal mol −1 .

Preparation of Ligands
A library of marine seaweed metabolites containing 1110 unique compounds was retrieved from the Seaweed Metabolite Database [35]. The library harbors a diverse chemical space with known biological activities such as antiviral, antimicrobial, antiinflammatory, and anti-cancer [36]. Before molecular docking, the conformation of ligands was optimized by removing salt (if any) and producing different ionization states at pH 7.0 ± 1.0 using "Epik module in LigPrep tool" (Schrodinger-2018-4, LLC, NY, USA) followed by energy minimization using the optimized potential for liquid simulation (OPLS3e) forcefield as described previously [26,37].

Homology Modeling of UK SARS-CoV-2 Spike Protein RBD and Its Validation
A model of UK SARS-CoV-2 Spike protein RBD was generated by inserting Tyr453Phe and Asn501Tyr mutations in the Spike protein RBD using PyMol. The primary amino acid sequence of the mutated strain was used to search the templates in SWISS-MODEL. On the basis of GMQE score and percent identity, we used chain A of 6ZBP as a template to generate the model using SWISS-MODEL. The three-dimensional structure of the generated model was validated using the protein structure validation suite (PSVS). Parameters such as PROCHECK, VERIFY3D, and Ramachandran plot were generated and examined. Further, ProSA-web was used to assess the overall quality of the model.

Preparation of Target Protein (Spike Protein RBD of UK SARS-CoV-2)
The 3D coordinates of Spike protein RBD of SARS-CoV-2 (PDB Id: 6M0J; resolution 2.45 Å) were retrieved from the PDB-RCSB database [38]. In the X-ray crystal structure, Spike protein RBD was in complex with the ACE2 receptor protein. Before molecular docking, Spike protein RBD was pre-processed, deleting any heterogeneous atoms or ligands including ACE2 and removing non-catalytic water molecules. The bond orders were defined, and missing H-atoms were added using "Protein Preparation Wizard" (Schrodinger-2018-4, LLC, New York, NY, USA), as reported previously [37,39]. A network of H-bonds was defined, and the whole system was energy minimized through the OPLS3e forcefield. The ligand conformational search for molecular docking was accomplished inside a grid box of 80 × 80 × 80 Å dimension (located at −37.0 × 29.8 × 4.4 Å) generated by the "Receptor-Grid Generation Tool" (Schrodinger-2018-4, LLC, New York, NY, USA).

Molecular Docking
The library of marine seaweed compounds was screened against the Spike protein RBD at 3 different stages, namely, high-throughput virtual screening (HTVS), standard precision (SP) molecular docking, and extra precision (XP) molecular docking in "Glide" (Schrodinger-2018-4, LLC, New York, NY, USA), as reported previously [40,41]. The topscoring compounds with a docking score ≤ −4.000 kcal mol −1 in HTVS (51 ligands) were subjected to SP docking. The most promising compounds with a docking score ≤ −5.000 kcal mol −1 in SP docking (5 ligands) were subjected to XP docking. The binding affinity (K d ) of the ligands was calculated from the docking energy (∆G) for Spike protein RBD using the below equation [42,43].

Molecular Dynamics (MD) Simulation
The dynamics and stability of protein-ligand complex were evaluated by performing MD simulation with the help of "Desmond" (Schrodinger-2018-4, LLC, New York, NY, USA). An orthorhombic simulation box was used to perform MD simulation after placing the protein-ligand complex at the center of the box, at least 10 Å away from the boundaries. TIP3P explicit solvent molecules were used to solvate the simulation box, and the system was neutralized by adding proper counterions. The physiological conditions were mimicked by adding 150 mM NaCl and adjusting the system's temperature and pressure to 298 K and 1 atm bar. The energy of system was minimized by 2000 iterations with convergence criteria of 1 kcal/mol/Å using OPLS3e forcefield. A production run of 50 ns was computed using NTP (isobaric-isothermal) ensemble. The temperature was maintained using a Nose-Hoover Chain thermostat, and the pressure was kept constant with the help of Matrtyna-Tobias-Klein barostat [47,48]. During the production run, a time step of 2 fs was fixed, and the energy and structure were logged at an interval of 10 ps in the trajectory. Analysis of MDS trajectory was performed using "Simulation Interaction Diagram tool" (Schrodinger, LLC, NY, USA) to determine the root mean square deviation (RMSD), root mean square fluctuation (RMSF), radius of gyration (rGyr), molecular surface area (MolSA), solvent accessible surface area (SASA), and polar surface area (PSA).

Determination of Binding Free Energy Using Prime/MM-GBSA
The binding free energy of ligand having lowest XP docking energy was determined using the molecular mechanics (MM) and generalized Born surface area (GBSA) using "Prime/MM-GBSA module" (Schrodinger-2018-4, LLC, New York, NY, USA) as reported previously [49]. The binding free energy (∆G bind ) of the protein-ligand complex was measured using the following relations: ∆G bind = ∆H − T∆S ∆H = ∆E gas + ∆G sol (4) G sol = G polar + G non-polar (7) G non-polar = γSASA + β The binding free energy (∆G bind ) is the difference in free energy of the complex and the individual free energies of protein and ligand. In other words, binding free energy is defined as the difference between enthalpic (∆H) and entropic contributions (T∆S), where enthalpy (∆H) is the sum of gas phase energy (∆E gas ) and solvation free energy (∆G sol ). Further, the gas phase energy (∆E gas ) is the sum of internal (E int ), Van der Waals' (E vdW ), and electrostatic or Coulombic (E elec ) energies. The internal energy (E int ) term can further be divided into energies of the bonds (E bond ), angles (E angle ), and torsions (E torsion ). Conversely, solvation free energy (G sol ) is the total of polar (G polar ) and non-polar (G non-polar ) solvation energies. The polar solvation energy (G polar ) was determined by employing the generalized Born (GB) implicit solvation model. The non-polar solvation energy was calculated by taking the values of surface tension proportionality constant (γ) as 0.00542 kcal mol −1 Å 2 , and non-polar solvation free energy of a point solute (β) as 0 kcal mol −1 . Moreover, SASA was calculated using a linear combination of pairwise overlap (LCPO) model.

Conclusions
Computational tools have been found to reduce time and effort to identify a potential therapeutic solution against different diseases [50][51][52]. The present study focused on identifying potential compounds that can block ACE2-RBD interaction. We screened a library of marine seaweed compounds against the UK strain of SARS-CoV-2 harboring multiple mutations in Spike protein. We employed different computational tools such as homology modeling, HTVS, molecular docking, molecular dynamic simulation, and free energy calculations. In view of the absence of three-dimensional crystal structure of RBD of UK SARS-CoV-2 strain, we generated a homology model of RBD using SWISS-MODEL. On the basis of extra precision molecular docking scores, we identified BE011 (Dieckol) as the most promising inhibitor of the RBD model. Dieckol is abundantly found in brown alga Ecklonia cava from Jeju Island, Korea. The structure analysis of Dieckol suggests that it does not possess drug-like properties and hence cannot be used as a lead inhibitor against RBD. Further, we utilized the structure of Dieckol as scaffold to generate different derivatives using ChemSketch and subjected them to XP molecular docking and ADMET analysis. The Dieckol derivative, i.e., DK07 was identified as the most potent inhibitor of RBD and was found to possess acceptable physicochemical, pharmacokinetic, drug-likeness, and ADMET properties. The IUPAC name of DK07 is 8-{3-hydroxy-4-[(7hydroxynaphthalen-2-yl)oxy]phenoxy}-1,4-benzodioxin-5-ol. DK07 binds to the RBD at the ACE2-RBD interface and interacts with key amino acid residues. Free energy calculation and molecular dynamics simulation of RBD-DK07 suggest the involvement of hydrogen bonding, electrostatic interactions, and Van der Waals' interactions as the driving source in the formation of a stable RBD-DK07 complex. The findings of this study require additional proof from in vitro and in vivo experiments to validate the potential of DK07 to bind UK strain of SARS-CoV-2 Spike protein RBD and prevent its interaction with ACE2.