(+)-Usnic acid and its salts, inhibitors of SARS‐CoV‐2, identified by using in silico methods and in vitro assay

The pandemic caused by severe acute respiratory Coronavirus-2 (SARS-CoV-2) has been ongoing for over two years, and treatment for COVID-19, other than monoclonal antibodies, is urgently required. Accordingly, we have investigated the inhibitors of SARS-CoV-2 protein targets by high-throughput virtual screening using a marine natural products database. Considering the calculated molecular properties and availability of the compounds, (+)-usnic acid was selected as a suitable hit. In the in vitro antiviral assay of (+)-usnic acid by the immunofluorescence method, IC50 was 7.99 μM, which is similar to that of remdesivir used as a positive control. The generalized Born and surface area continuum solvation (MM/GBSA) method was performed to find the potent target of (+)-usnic acid, and the Mpro protein showed the most prominent value, −52.05 kcal/mol, among other SARS-CoV-2 protein targets. Thereafter, RMSD and protein–ligand interactions were profiled using molecular dynamics (MD) simulations. Sodium usnate (NaU) improved in vitro assay results with an IC50 of 5.33 μM and a selectivity index (SI) of 9.38. Additionally, when (+)-usnic acid was assayed against SARS-CoV-2 variants, it showed enhanced efficacy toward beta variants with an IC50 of 2.92 μM and SI of 11.1. We report the in vitro anti-SARS-CoV-2 efficacy of (+)-usnic acid in this study and propose that it has the potential to be developed as a COVID-19 treatment if its in vivo efficacy has been confirmed.

NSPs play an essential role in viral replication and transcription. Of the 16 NSPs, the above-mentioned four NSPs are known as major target proteins. NSP3, also called papain-like protease (PLpro), cleaves NSP1 and NSP2 from the polyproteins pp1a and pp1ab. GRL0617, a SARS-CoV inhibitor, has been reported to exhibit in vitro antiviral efficacy against SARS-CoV-2 with IC 50 of 2.1 μM by binding to PLpro 19,20 . NSP5, also called 3CLpro or Mpro, is the main protease involved in the maturation of other NSPs and the cleavage of polyproteins pp1a and pp1ab. MI-09 and MI-30, synthetic molecules with a five-membered bicycloproline moiety and γ-lactam ring, have shown potent inhibitory activities against SARS-CoV-2 Mpro with an IC 50 of 15.2 and 17.2 nM respectively. Further, these compounds have shown in vivo antiviral efficacy with good PK properties and no significant toxicity 21 . NSP12, also called RNA-dependent RNA polymerase (RdRP), catalyzes the replication and transcription of the SARS-CoV-2 RNA genome. Remdesivir is an FDA-approved drug only for emergency use in COVID-19 treatment. The incorporation of remdesivir into the growing RNA product induces RdRp stalling and creates a barrier to RNA translocation 22 . NSP13 is an RNA helicase that interacts with NSP12 to enhance viral replication efficiency 23 . Posaconazole and grazoprevir have recently been identified as potential repurposing drugs that target helicase by in silico studies 24 .
The influx of the SARS-CoV-2 can be prevented by inhibiting the host proteins involved in the viral infection. Furin is the protease responsible for the proteolytic activation of the spike at the S1/S2 site 25 . Although many studies point out that furin is not the only protease involved in the cleavage of the S1/S2 site, treatment with a furin inhibitor, decanoyl-RVKR-chloromethylketone (CMK), blocks the entry of the virus 26 . TMPRSS2, a crucial protease in SARS-CoV-2 entry into the host cell, is inhibited by camostat mesilate and nafamostat, both of which are approved treatments for other medical uses. The in vitro antiviral IC 50 of camostat mesilate is measured as 4.2 nM 27 . Further, camostat mesilate has shown in vivo antiviral activity against SARS-CoV with a survival level of above 60% when dosed in mice at 30 mg/kg 28 . Furthermore, camostat mesilate has undergone clinical trials for the treatment of COVID-19; however, it did not show any particular efficacy 29 . Cathepsin L, a lysosomal protease and a marker for various cancers, plays a crucial role in SARS-CoV-2 infection by cleaving the S protein at a position different from the furin cleavage site 30,31 . Aloxistatin (E-64d), a broad cysteine protease inhibitor, reduces the cell entry of SARS-CoV-2 pseudovirions by 92.5% at 30 μM 32 . AAK1, the numb-associated kinase (NAK) family, is responsible for clathrin-mediated endocytosis. Barcitinib is a selective Janus kinase 1/2 inhibitor that shows a high affinity for AAK1. The phase 3 clinical trial for the treatment of COVID-19 has been conducted using barcitinib and the 28-day mortality rate has been reduced by 38.2% compared to that of the placebo 33 .
Natural products have long been used to treat various diseases including cancer, metabolic diseases and infectious diseases. Natural products that hang down through folk remedies are profiled along with advances in chemistry and are regarded as useful resources for drug development 34,35 . Over 300,000 natural products obtained from various resources including plants, animals, microorganisms and marine organisms have been reported and new compounds with distinct chemical scaffolds are continually being reported 34,36 . For marine natural products that have not been thoroughly studied compared to terrestrial compounds, more than 1,000 compounds with novel skeletons have been reported every year 37 . Natural products are structurally diverse and many of them exhibit characteristic bioactivities. Among the natural products, there are many substances with antiviral effects 35 . Spirooliganone, uncinoside, baicalein and curcumin are natural products with antiviral effects 38 . Specifically, Tamiflu, a treatment for influenza A and B, is based on a natural product derived from the Chinese spice, star anise (Illicium verum) 39 . Thus, the use of natural products as a starting point for new drug discovery will be a viable strategy for the successful development of therapeutics that can effectively control COVID- 19. In this study, the hit compound, (+)-usnic acid, was identified by the in silico molecular docking of more than 20,000 marine natural products against eight SARS-CoV-2 targets. The antiviral efficacy of the hit compound was evaluated using immunofluorescence. To rationalize the antiviral efficacy of the hit compound, MM-GBSA and molecular dynamics (MD) approach were employed. Finally, the potency of (+)-usnic acid was examined under various conditions including the salt formation of (+)-usnic acid (sodium and potassium salts) and in vitro assays against some SARS-CoV-2 variants. Through this study, we have demonstrated that (+)-usnic acid could be a good drug lead for the treatment of the deadly COVID-19.

Methods
Hit discovery through high throughput virtual screening (HTVS). An in-house database containing natural products derived from marine sources (animals, plants, bacteria, and fungi) was selected as the HTVS ligand for the SARS-CoV-2 protein targets. The in-house database contained 24,669 marine natural products. Ligands were prepared using the LigPrep software offered by Schrödinger 40 . The OPLS3e force field was applied to the ligands generating tautomers and possible states at a target pH of 7.0 ± 2.0.
The grids were generated based on the eight SARS-CoV-2 targets listed in Table 1. For the protein grid generation, the grid box was located within the centroid of the selected binding site residues. The size of the grid box was set to dock the ligands within 20 Å. The ligands were docked using the Glide module from Schrödinger with HTVS precision 41 . The molecular properties such as absorption, distribution, metabolism and excretion (ADME) of the ligands were calculated using QikProp software 42 .
Compound isolation and structure confirmation. The (+)-usnic acid was isolated from the extracts of marine fungus, Mycosphaerella sp., which was collected from a marine sediment at Donghae-si, Korea. The fungus was cultivated in 6 L of potato dextrose broth (PDB) dissolved in seawater in 27 °C, at 140 rpm shaking incubator for 7 days. The broth was extracted with ethyl acetate and yielded 4.01 g extract. The crude extract was fractionated into 3 fractions with a silica gel column chromatography using CH 2 Cl 2 and MeOH as solvent. Fractions were further purified by C18 HPLC (Phenomenex luna C18 column, 250 mm × 10 mm, 5 μm) using 65% CH 3  www.nature.com/scientificreports/ spectrometer. Electrospray ionization source (ESI) mass data was obtained using Agilent Technologies 6120 quadrupole mass spectrometer coupled with Agilent Technologies 1260 series HPLC. The optical rotation of (+)-usnic acid was measured in chloroform using a 10 mm path length cell on a Digital Polarimeter P-2000, Jasco Inc. For the cell preparation, the 384-tissue culture plates were seeded with 1.2 × 10 4 vero cells per well. After 24 h, compounds were prepared at 10 points by twofold serial dilutions in dimethyl sulfoxide (DMSO) to 50 μM and treated to the cells. One hour after the compound treatment, cells were transferred to BSL-3 facility and the viruses were infected at MOI of 0.008. The plates were incubated at 37 ℃ for 24 h. After fixing the cells with 4% paraformaldehyde (PFA), permeabilization was performed. Anti-SARS-CoV-2 N primary antibody was treated and afterward, cells were stained with Alexa Fluor 488-conjugated goat anti-rabbit IgG secondary antibody and Hoechst 33342. The fluorescent images of the infected cells were obtained using a large capacity image analysis instrument, Operetta (Perkin Elmer).

Computational calculation of MM-GBSA and MD simulations.
The extra precision (XP) docking of (+)-usnic acid was performed against eight selected SARS-CoV-2 receptors. Based on the docking studies, the Prime MM-GBSA offered by Schrödinger was used to calculate the binding-free energy of the docked ligand for each protein target 51 . The binding free energy was calculated using the following formula: ∆G(Solv) is the difference in the GBSA solvation energy of the (+)-usnic acid-protein complex compared to the sum of the energies of the uncomplexed protein and ligand. ∆E(MM) is the difference in the minimized energy of the (+)-usnic acid-protein complex compared to the sum of the energies of the uncomplexed protein and ligand. ∆G(SA) is the difference in the surface area energy compared to the sum of the energies of the uncomplexed protein and ligand. Prime MM-GBSA was operated in the OPLS3e force field using the variable dielectric surface generalized Born (VSGB) solvation model.
The thermodynamic behavior, stability and interaction of (+)-usnic acid with Mpro were assessed by MD simulations. The MD calculations were performed using Desmond software 52,53 . The MD system was built using the TIP3P 3-site water model. The system was neutralized by adding 3 sodium ions and the salt concentration was adjusted to 0.15 M using the sodium and chloride ions. For the MD simulations, an isothermal-isobaric (NPT) ensemble class was applied with the temperature and pressure set to 300 K and 1.01325 bar respectively. The Nose-Hoover chain method was employed for the thermostat method and the Martyna-Tobias-Klein method was applied to the barostat method. The total simulation time was 100 ns, at the recording interval of 100 ps. The analysis of the MD simulation was executed using the simulation interaction diagram enclosed in the Desmond module.
Anti-SARS-CoV-2 assay of the (+)-usnic acid salts. The anti-SARS-CoV-2 assay of the salt form of (+)-usnic acid with the counter-ions, Na + or K + , was evaluated. The salt form of (+)-usnic acid was synthesized by titration with sodium hydroxide and potassium hydroxide and, thereafter, it was lyophilized. The structure was confirmed by NMR. The efficacy of the (+)-usnic acid salts was evaluated using the immunofluorescence as described above.

Analysis of the HTVS of marine natural products against SARS-CoV-2 proteins. The HTVS
for marine natural products was performed using the Glide software. An in-house database containing 24,669 marine natural product structures was analyzed. The docking scores of eight SARS-CoV-2 target proteins (PLpro, Mpro, RdRp, helicase, furin, TMPRSS2, cathepin L, and AAK1) were profiled. The number of compounds that were identified as possible hits was 2,251, with a docking score below -7. Further, the compounds were filtered with ADME properties calculated by Qikprop, following the criteria suggested by Schrödinger. Twenty-seven compounds were passed as potential hits. Considering the availability and ease of supply, four natural products were analyzed at a single dose (10 μM), and (+)-usnic acid showed efficacy against SARS-CoV-2 in an in vitro antiviral assay (Fig. S3). (+)-Usnic acid was selected as the hit compound for the further analysis. The calculated properties of the (+)-usnic acid are shown in Table 2. The (+)-usnic acid used in the experiment was structurally confirmed by NMR (Fig. S7). The purity was monitored by LC-MS and confirmed that it was over 99% (Fig. S9). The optical rotation was 500.4 ([α] 25 D ) and it was considered optically pure by comparing with standard compound.

Evaluation of the in vitro antiviral activities of (+)-usnic acid against SARS-CoV-2. The SARS-
CoV-2 antiviral efficacy of the selected compound, (+)-usnic acid, was evaluated by immunofluorescence. The dose response curves of (+)-usnic acid with three control compounds (chloroquine, lopinavir, and remdesivir) are shown in Fig. 1. The IC 50 value of (+)-usnic acid was 7.99 μM and CC 50 was over 50 μM. The selectivity index (SI) is calculated to be 6.26. The IC 50 values of the controls, chloroquine, lopinavir and remdesivir, were 11.48, 11.80, and 7.42 μM, respectively. The CC 50 values of the controls were all above 50 μM; the CC 50 of lopinavir was higher than 150 μM. The SI of chloroquine, lopinavir and remdesivir was 13.07, 4.24, and 6.74 respectively. The efficacy of (+)-usnic acid against SARS-CoV-2 is similar to that of remdesivir, which is approved by the FDA for the treatment of emergency patients with COVID-19. The (+)-usnic acid showed a lower IC 50 value compared to those of the two other controls, chloroquine and lopinavir.

MM-GBSA and MD analysis of (+)-usnic acid against Mpro. The MM-GBSA method was used to
identify the most preferable SARS-CoV-2 target proteins for (+)-usnic acid. The binding free energies are shown in Fig. 2. The target protein that showed the highest difference in the binding-free energy is Mpro, -52.05 kcal/ mol. The binding free energies of remaining proteins were in the range of -39 and -29 kcal/mol, which had at least 10 kcal/mol difference with that of Mpro.
The detailed contributions of the binding-free energies are shown in Table 3. The binding free energy (dG bind ) was calculated from the sum of dG coloumb , dG covalent , dG Hbond , dG Lipo , dG Packing , dG SolvGB and dG vdW . The main contributor to Mpro, the protein with the lowest binding-free energy when complexed with (+)-usnic acid, was dG vdW with a value of -45.30 kcal/mol. Further, Mpro had relatively low dG Lipo compared to those of other proteins; -7.81 kcal/mol.
The MD simulation of Mpro complexed with (+)-usnic acid was monitored and is shown in Fig. 3. Protein and ligand RMSD are shown in Fig. 3a,b. The RMSD of the alpha-carbon (C α ) of the Mpro protein reached up to 3.5 Å in 30 ns and then remained equilibrated for the rest of measurement times. The RMSD fluctuation of C α remained within 1-3 Å, which is acceptable for small and globular proteins. The ligand RMSD rapidly rose to 5 Å in 20 ns. Then, the fluctuation of ligand RMSD remained in a steady-state between 4.8 and 5.6 Å for the other monitoring period.
Through the MD simulation, the amino acid of Mpro protein, that (+)-usnic acid mainly interacts with, was profiled and shown in Fig. 3c-d. The interaction through the H-bond and water bridges was predominant and there were slight hydrophobic interactions as well; however, ionic interaction was not detected. His164, Asp187 Table 2. The calculated molecular property of the selected compound, (+)-usnic acid. The molecular property includes docking scores of protein targets and predicted ADME properties (dipole, PSA, volume, SASA, QplogS, %HumanOralAbsorption). www.nature.com/scientificreports/  www.nature.com/scientificreports/ and Gln189 were the amino acids with the interaction fractions over 0.8, in which most of them were accounted by H-bonds and water bridges. According to the percentage interaction illustration, (+)-usnic acid interacted with Asp187 at most with 78%, followed by 65% of His164. At the 3D model in Fig. 3e, (+)-usnic acid bound to the Mpro binding pocket and interacted with His41, Cys145, Asp187 and Gln189 through hydrogen bonds.

Inhibition of the SARS-CoV-2 variants by (+)-usnic acid.
The antiviral efficacy of (+)-usnic acid against SARS-CoV-2 differs depending on the variants. The IC 50 of (+)-usnic acid against alpha and delta variants were 6.05 μM and 7.17 μM respectively, which were similar to the efficacy against the original strain. However, usnic acid showed enhanced efficacy against the beta variant where the IC 50 was 2.92 μM. The SI for the beta variant was 11.1, indicating more than twice the SI of the other two variants assayed (Table 5).  Table 4. Anti-SARS-CoV-2 efficacy of (+)-usnic acid and its salts (NaU and KU).

Discussion
From a marine natural products in-house database containing 24,669 natural products, (+)-usnic acid, a substance effective against SARS-CoV-2, was discovered as a hit compound through virtual screening and in vitro assays. The thermodynamic properties and amino acid interactions of (+)-usnic acid were analyzed using MD simulation. Furthermore, the efficacy of (+)-usnic acid and its salts against the SARS-CoV-2 variants was evaluated in this study. The HTVS of natural product libraries allowed the profiling of docking energies for eight SARS-CoV-2 target proteins within a few days. The docking energy was the priority in the selection and compounds exceeding the permitted range of calculated PK properties were removed. Twenty-seven natural products were identified as potential hits, and considering the ease of securing the compounds, four natural products were selected for in vitro evaluation at a single dose. The (+)-usnic acid showed potent efficacy against SARS-CoV-2 and we conducted further experiments with this compound.
Usnic acid (UA) was first reported in 1844, since then many bioactivities have been studied for this dibenzofuran compound and its derivatives 54 . Usnic acids are found mostly in lichens such as Alectoria, Cladonia, Evernia, Lecanora, Ramalina and Usnea 55 . The usnic acid shows wide-ranging bioactivities including antimicrobial, antifungal, antiprotozoal, insecticidal, anti-inflammatory, phytotoxic, anti-inflammatory, and antimitotic activities 56 . Moreover, the usnic acid shows weak antiviral activity against Herpes simplex type1, Polio type1, papiloma and influenza virus A (H1N1) 56,57 . Recently, usnic acid was produced from the marine fungus Mycosphaerella sp., which was isolated from marine sediment in Korea with several compounds in similar scaffolds 58 .
The in vitro immunofluorescence test against the SARS-CoV-2 showed that (+)-usnic acid showed efficacy with IC 50 , CC 50 and SI corresponding to 7.99 μM, more than 50 μM and 6.26 respectively. Compared to the three compounds (remdesivir, lopinavir and chloroquine) used as positive controls, (+)-usnic acid showed a similar profile to remdesivir, which was approved as a treatment for COVID-19 only for an emergency use. (+)-Usnic acid showed lower IC 50 values than those of lopinavir and chloroquine.
Although the antiviral efficacy of (+)-usnic acid was validated, we did not know the exact protein target. Therefore, we attempted to identify it using MM-GBSA. MM-GBSA was applied to the energy-minimized pose of docked (+)-usnic acid with eight SARS-CoV-2 protein targets. The binding free energy of Mpro protein was the lowest at -52.05 kcal/mol and the energy difference among those of other proteins was at least 10 kcal/mol. Thus, we determined that Mpro was the most plausible SARS-CoV-2 protein target for (+)-usnic acid. For further analysis, the MD simulations of Mpro with (+)-usnic acid were performed for 100 ns. The protein RMSD increased up to 3.5 Å for 30 ns and ligand RMSD rose to 5 Å for same period as well. The ligand moved slightly to inner pocket of Mpro compared to the binding pose of docking analysis. Instead of a hydrogen bond with Gly143, (+)-usnic acid became more stabilized by H-Bonds with Asp187 and Gln189. Afterwards, (+)-usnic acid was remained stabilized for other 100 ns monitoring period. As a result, the most likely target protein of SARS-CoV-2 for (+)-usnic acid is Mpro.
Subsequently, the antiviral efficacy of (+)-usnic acid under various conditions was evaluated. First, the sodium and potassium salts of (+)-usnic acid were prepared. The IC 50 and SI of NaU were slightly improved, while those of KU remained almost the same as that of (+)-usnic acid. This improvement in the salts could be explained by the enhanced water solubility. The CC 50 of both salts was greater than 50 μM. Next, the antiviral efficacy of (+)-usnic acid against the SARS-CoV-2 variants, which are causing serious pandemic worldwide, was evaluated. The (+)-usnic acid against the original SARS-CoV-2 strain showed profiles similar to those of the alpha and delta variants, while it exhibited a superior effect on the beta variant. The reason for this is still unclear and further research is needed to understand why (+)-usnic acid exhibits better efficacy against the beta variant, but it is hypothesized that the K90R mutation on Mpro, 95% prevalent in beta affected the protein stability of the Mpro 59,60 . As lys90 is located on the opposite side of the binding site, it does not significantly affect the binding of (+)-usnic acid. But as a result, protein folding of Mpro becomes destabilized and the overall efficacy of (+)-usnic acid against beta variant seem enhanced.
Usnic acid has been widely used as an ingredient for daily necessities including toothpastes and mouthwashes 55 . It has also been used as a dietary supplement for weight loss 61 . Unfortunately, usnic acid exhibits hepatotoxicity because of the uncoupling of oxidative phosphorylation, resulting in the destruction of mitochondrial respiration at very high concentrations 62 . Nevertheless, usnic acid is expected to have sufficient efficacy and www.nature.com/scientificreports/ safety for the treatment of COVID-19. The maximum period required for COVID-19 recovery is usually 7 days and paxlovid and molnupiravir, approved for emergency use, are employed for five days only. The use of usnic acid within five days may not show any significant toxicity since there are records of human consumption for a relatively long period of over two weeks 63 . Therefore, it is suggested that usnic acid could be used in emergency situations for the treatment of COVID-19, if its in vivo efficacy is proven. Additionally, oral administration is expected to be possible based on a study showing the improved bioavailability of KU in a mouse tumor xenograft model 64 .
In conclusion, this study identified (+)-usnic acid as a potential drug lead against COVID-19 by the HTVS of a marine natural products database. Further, the in vitro efficacy of (+)-usnic acid has been validated in several SARS-CoV-2 variants. The protein target of (+)-usnic acid was explored using MM-GBSA and MD approaches. SARS-CoV-2 Mpro was identified as a potent protein target of (+)-usnic acid. Although there is a toxicity issue on usnic acid, it is easy to be obtained and has a record of its human use. Therefore, we propose (+)-usnic acid as a potential therapeutic tool for COVID-19 treatment, considering the desperate need to end the deadly pandemic.

Data availability
The datasets used and/or analyzed during the current study available from the corresponding author on reasonable request.