Repurposing naproxen as a potential antiviral agent against SARS-CoV-2

The Outbreak of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections in late 2019 in China and many other countries around the world necessitate immediate action to develop new drugs against the virus. Repurposing of existing drugs for new targets is a fast, safe and unexpansive approach for this goal. Studies have shown that naproxen could specically interact with the RNA binding domain of nueclporteins of RNA viruses such as the inuenza virus. Therefore, this study aimed to evaluate the binding properties of naproxen to the nucleocapsid protein of SARS-CoV-2. 3D structure of N and C terminal domains of SARS-CoV- 2 nucleocapsid were constructed and each were docked with naproxen and analyzed during 100 ns of molecular dynamics. The results showed that naproxen interacts with the N terminal domain of the nucleocapsid via two salt bridges with Arg 88 and 92 and a network of h-bonds. Molecular dynamics analysis was also revealed that all the coordinations of naproxen with N terminal domain were kept during 100 ns of simulation time. The results of this study provide insights how naproxen can specically interact with the conserved RNA binding module of the nucleocapsid of SARS-CoV-2 that would inhibit the packaging of viral genome into capsid and virus assembly. Therefore we recommend evaluating the antiviral effects of naproxen against SARS-CoV-2 in in vitro studies and clinical trials. for the rapid development of new anti-SARS COV–2 Using target-based virtual ligand screening, 18 putative proteins encoded by SARS COV- were against the ZINC drug database and a local database.the of this showed three including 3-chymotrypsin-like protease (3CLpro), Spike, RNA- dependent RNA polymerase (RdRp), and papain-like protease (PLpro) with several lead compounds. However, such compounds were not identied for the nucleocapsid study also indicate naproxen might compete with viral RNA for biding with nucleocapsid therefore inhibiting viral capsid assembly and viral particle formation. Based on promising results obtained in this study we suggest further in vitro experiments and clinical trials to use naproxen in the treatment of COVID–19 patients.


Introduction
Coronaviruses, the causative agents of many pathological conditions in humans and animals are large single-stranded RNA viruses belonging to Coronaviridae. They are divided into 4 genera Alphacoronavirus, Betacoronavirus, Gammacoronavirus, and Deltacoronavirus 1 The genus Betacoronavirus contains some the most important respiratory viruses such as Severe acute respiratory syndrome-related coronavirus (SARS-CoV), Middle East respiratory syndrome-related coronavirus (MERS-CoV) and the newly identi ed one severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) 2 . SARS-CoV-2 was identi ed as the cause of a cluster of pneumonia cases in Wuhan, a city in the Hubei Province of China, at the end of 2019 3 Since the rst reports of cases from China, the number of infected diseases has been rising daily in this country and other countries 4 . The origin of SARS-CoV-2 was in enzootic bat viruses and then transmitted from presymptomatic or asymptomatic infected cases 5 . SARS-CoV-2 is closely related to SARS-CoV containing a 30kb genome organized into several ORFs necessary for viral infection and replication in the host cells. About 21 kb of the genome encodes for two large polyproteins pp1a and pp1ab that are processed into several nonstructural proteins of the virus by viral protease. SARS COV-2 also encodes for 4 structural proteins namely the spike (S), membrane (M), envelope (E), and nucleocapsid (N) proteins. The rst three proteins are essential parts of viral capsid and envelope while the N protein takes part in the packaging of the genome into viral particle 6,7 .
Crystallography studies of SARS-CoV and SARS CoV-2 nucleocapsids showed that the protein has two distinct domains at each terminus separated by intrinsically disorderedcentral Ser/Arg (SR)-rich linker. The N and C terminal domains are responsible for viral genome binding and nucleocapsid dimerization/oligomerization respectively. The N terminal domain contains a conserved RNA binding domain that not only takes parts in the formation of viral ribonucleoprotein (RNP) complex but also in regulating the viral replication and transcription by maintaining highly ordered RNA conformation 8,9 The conserved N and C terminal domains of the nucleocapsid of the virus have been proposed as potential drug targeting sites for inhibition of viral replication and assembly 8 . Two strategies can be considered for this purpose. First, structure-based drug design screening and synthesis for these domains that are time-consuming and expensive with a high rate of failure and potential unpredicted adverse effect on the human body. Second, repurposing commonly used drugs that are available in the market with known safety and adverse effect.
The incubation period for COVID-19 is thought to be within 14 days after exposure, with most cases present symptoms approximately ve days following exposure 10 . Infections with SARS COV-2 are not severe in many cases, although most patients have had a critical illness [11][12][13][14][15][16] . Pneumonia is a serious manifestation of infection, characterized primarily by fever, cough, and dyspnea. Additionally, gastrointestinal symptoms such as nausea and diarrhea have been reported in some cases as an uncommon manifestation [12][13][14] . Management of patients with suspected or documented COVID-19 has two sections include appropriate infection controland supportive care 17 . There are no speci c antiviral therapies for COVID-19 and there are many efforts for drug repurposing in all of the worlds 18,19 . For example, lopinavir-ritonavir combination therapy, as protease inhibitor which is used for the treatment of HIV infection, has antiviral activity against the SARS-CoV in vitro 20 and MERS-CoV in animal studies 21 and it could be used for treatment of COVID-19 in some patients 22,23 . Also, a randomized clinical trial in China is performing to evaluate the e cacy and safety of remdesivir as a nucleotide analogue in COVID- 19 17 due to its activity against SARS and MERS-CoVin both in vitro and animal studies 24,25 .
Naproxen is a non-selective NSAIDs that inhibit the activity of both COX isoenzyme. NSAIDs is among the most widely used therapeutics with antipyretic and analgesic effect and they are choice treatment in various in ammatory diseases such as arthritis, rheumatism as well as relieving the pain of everyday life 26 Previous study shows that Naproxen had effective in in uenza A and B as an antiviral activity in animal and clinical study [27][28][29] . It has been shown that naproxen interacts directly with the essential amino acids for viral RNA binding in the N terminal domain of the wild-type in uenza A nucleoprotein, therefore, prevents the formation viral ribonucleoprotein (RNP) complex and subsequent virus production 29,30 .
As most COVID19 patients complaints for having fever and myalgia 13,14,31 and presence of welldocumented evidence for the speci c interaction of naproxen with RNA binding domain of viral nucleoproteins 29,30 , the aim of this study was to in silico evaluation of the antiviral effect of naproxen on SARS COV-2 and further recommendation of it as both antiviral and anti-in ammatorydrug in clinical trials.

Homology modeling
The full-length protein sequence of severe acute respiratory syndrome coronavirus 2's nucleocapsid (NP) was retrieved from NCBI Protein Database, accession number QIC53221. Since there is no complete template for the whole sequence (419 amino acids) and also the important functional parts of the protein located at its N (RNA binding domain) and C terminal domains(Dimerization domain) (Fig 1), two separate homology models were created.
Initially, the domains indicated in g 1 were extracted and searched against the PDB database using BLAST-P to nd an optimal template for each one.PDB ID 1ssk which is belonged to the Nterminal RNAbinding domain of the SARS CoVnucleocapsid protein 32 was selected as a template for the Nterminal domain (% Identity 93%). Similarly, Crystal structure of the SARS coronavirus nucleocapsid protein dimerization domain (PDB ID 2GIB) 33 at 1.75 Å resolution was used as the template for the Cterminal domain (% Identity 93%). Thereafter, the sequences and templates were aligned. The initial 30 models were created for each sequence and were sorted according to the GB/VI model. Since there is a very high similarity between the sequences and the templates there was no concern about the protein's loops reconstruction. The top model was re ned to the highest level by energy minimization until the root mean square gradient falls below 0.001. Finally, the quality of the models was evaluated by the Ramachandran plot and SAVE Server. All of the modeling steps were conducted on the Molecular Operating Environment (MOE) 2019.01.02.In addition to the protein targets, the 3D model of the conserved stem-loop of the SARS COV-2 RNA packaging signal (Fig 2) was also predicted by the DNA/ RNA modeling module of Molsoft ICM-Pro 3.8.7c. This signal is critical for the NP to correctly recognize the viral genomic RNA and assemble it into the capsid.

Protein-Ligand Docking
To analyze the exact binding pose and calculate the approximation of the binding free energy between Naproxen with both domains of the NP protein, molecular docking was accomplished by molsoft ICM-Pro 3.8.7c. Initially, the modeled structures were prepared by adjusting hydrogen atoms, xing the partial charges of the residues, etc. The ligand (Naproxen) was also retrieved from PubChem and converted to the 3D format. Subsequently, the binding pockets the both models were predicated and inspected visually. Thereafter, the docking grid was generated around the selected pocket. Finally, the fully exible ligand was docked in the grid guided pocket of the semi-exible N and Cterminal domains using a welltested scoring function which has equipped with exible ring sampling on the y for better ligand tting. The best binding pose was selected according to the docking score and visual inspections of the interacted residues. This pose was used in molecular dynamics simulation.Additionally, the most commonly prescribed NSAID for the alleviation of the SARS COV-2 induced lung in ammation, ibuprofen, was docked in the same pocket of both proteins as an irrelevant compound for comparison purposes.

Protein-Protein Docking
To elucidate the possible interfering effect of naproxen binding on the RNA binding activity of N protein and also dimerization of C terminal domain, two independent protein-protein docking studies were performed using Global FFT Dock of Molsoft ICM-Pro 3.8.7c. In the 1 st study, the conserved stem-loop of RNA packaging signal that was modeled earlier was docked at the empty RNA binding interface of the Nterminal domain. The top-scored pose was selected based on the interaction energy. The 2 nd study was performed similarly. However, the N protein with naproxen bonded at its RNA binding pocket was used as the receptor. Any interference of reproducing the 1 st study pose was investigated. A similar procedure with some modi cation was done for the Cterminal domain. The docking pocket for the Cterminal domain is the dimerization interface.

MM-GBSA Scoring
To estimate the correct free energy of binding between naproxen and its targets, the molecular mechanics energies approach combined with the generalized Born and surface area (MM-GBSA) was applied as a comprehensive modular nature approach for binding free energy calculation. To calculate this value, the MM-GBSA module of Schrodinger suite 2019-2 was used under OPLS3e force eld in VSGB 2.0 solvation model. A 5 Å full exibility of the receptor around the ligand with minimizing the sampling method was taken into account when the calculations were conducted.

Molecular Dynamics (MD) Simulation
The realistic interaction of naproxen with both proteins and also its effect on the protein structure was determined by 100 ns MD simulation in an explicit water model. For this purpose, the nal modeled structures which were already docked with naproxen were dissolved in a layer of TIP3 water molecule with 10 Å distance from the edge of the protein. The systems were neutralized by adding 50mM counterions (NaCl). For simulating the biological condition, the MD was performed in NAMD 2.13 molecular dynamics engine (GPU accelerated version) under constant pressure and temperature under the NPT ensemble. Amber14-EHT force eld was used for calculating the interaction potentials throughout the simulation. After preparing the systems, the MD was initiated according to the following protocol. Initially, the system was relaxed by 1 ns minimization step. Thereafter it was heated up to 310 K for 5 ns to bring the system out of local minima. The protocol was followed by 10 ns equilibration at the target temperature. Finally, the production step was executed for 100 ns with 2 fs timestep to solve the equation of motion. All light bonds length was constrained with the SHAKE algorithm. The system was sampled every 2 ps and the snapshot was analyzed for RMSD, RMSF, and interaction energy changes.

Computational model of N and C Terminal Domains
The atomic-resolution modelsof both N and C terminal domains of SARS COV-2 were shown in Fig 3 in ribbon representation with their transparent electrostatic surface (Blue and Red are positive and negative residues respectively). As already described, the quality of the models was evaluated by both the Ramachandran plot (inset of Fig 3) and also SAVE server. The Ramachandran plots show there is no more than 0.5% of the residues were in the not allowable area which is more than the acceptable threshold (5%). The results of the SAVE server were also con rmatory. For N and C domains, 98.54% and 80.41% of the residues haveaveraged 3D-1D score ≥ 0.2 respectively. The N terminal domain is mainly β content. However, the C portion is α/β. The RMSD values between both protein and their templates are less than 0.2 Å. According to these results, the modeled structures meet the criteria for a good model.Besides, both proteins contain an exposed pocket (dashed box) surrounded by some charged residues. The key residues of the N terminal domain's pocket include R88 and R92 which have a critical role in the RNA binding function. However, the dimerization interface contains many residues although some of them such as F34, R37, and F46 are so important 33  Since the docking algorithm was induced exibility in the RNA molecule, its structure undergoes some conformation changes compared to its initial state. As is clear, the best-docked pose indicates the RNA was exactly bound to the positively charged pocket (dock score -7.26 Kcal/mol) which was described in Fig 3. The residues involved in the binding process were also indicated in Fig 5b. Notably, two key residues, R88 and R92, were also present in the interaction interface. These residues coordinate the RNA molecule in the pocket through electrostatic interactions with the negatively charged backbone phosphate atoms. There are also two other important residues (Y109 and Y111) which boost the binding of the RNA molecule via pi-pi stacking interactions.
Similarly, the dimerization interface of the C terminal domain structure was also elucidated by superimposition guided manual protein-protein docking of the modeled structure to itself. The result was shown in Fig 6. As already described before and can be seen in this gure, this area includes many residues. Some of them may have a little more important role in binding such as F34, R37, and F46. However, the dimerization process requires almost equal participation from all residues.
The predicted top docking conformation of naproxen with the N and C terminal domains structures are shown in Fig 7 and 8. For the N terminal domain, the top-scored pose of the drug was accurately docked to the predicted active pocket of the protein that has been shown in Fig 3 (docking score -21.3 Kcal/mol). The ligand was stabilized inside the pocket through many interactions including but not limited to two salt bridges with both key Arginine residues (R88 and R92), a pi-anion with Y109, two piamid interactions with A50 and Y111 and two hydrogen bonds with backbones of G47 and T49. Among these interactions, the rst two salt bridges are crucial for the proper binding of the RNA packaging signal 32 . Therefore, if a compound can mask these residues strongly, it may prevent the assembly of the virus by hindering the RNA packaging process. Moreover, the protein docking investigation was revealed when naproxen is bound to the described pocket, the RNA packaging signal cannot attach to this site which is a promising condition. When the top docked pose of the naproxen is compared with ibuprofen as anirrelevant drug (docking score -17.64 Kcal/mol), it can be seen that the binding strength of naproxen is greater. Ibuprofen binding site is similar to naproxen. It has many interactions with the pocket's residues which most of them are various types of pi stacking interactions (Fig 7c). It also had an H-bond with S51 which is absent for the naproxen interaction. However, it is missed a salt bridge interaction with one of the key arginine residues (R88). This may be the main reason for the lower binding a nity of this drug to the N terminal domain compared to naproxen. Besides, there is also a salt bridge to R37. By the way, the protein docking study is shown even when naproxen occupied its site in the Cterminal domain, the dimerization process did not hinder ( g 9). This could be due to the already described fact that the dimerization interface is very wide and go through among many residues and the naproxen binding occupies only a very small portion of it. Therefore, to the best of our knowledge and according to this result, it seems the naproxen is not an ideal intervention for inhibiting the NP dimerization. It is noteworthy that ibuprofen had a considerably better docking score than naproxen for this domain. However, the MM-GBSA evaluation did not con rm its superiority to naproxen (table 1).

Binding Free Energy Prediction
Since the MM-GBSA calculation is more accurate than any other scoring functions of molecular docking, the binding a nity of naproxen with the docked structures was reevaluated by this method again for double reassurance of the previous ndings. The result is completely in accordance with the docking calculation and rank the naproxen binding capability to the N terminal domain at the top (Table 1) even though they used different force elds. In conclusion, the reliability of the docking results was indicated.

Molecular Dynamics Simulation
To con rm the results that have been obtained from the molecular dockings and thermodynamic investigations, two rounds of MD simulations were performed on the top docked pose of both N and C terminal domains. Before analyzing the primary results, the stability of parameters of the systems was checked by plotting temperature and energy of the system during the equilibration step which revealed the systems were reached to a steady-state from ( uctuation of both parameters was less than 5% standard error during this step). The primary results of the simulation are shown in Fig 10. The RMSD plot for "Naproxen -N terminal Domain" showing the backbone of the structure deviated less than 7 Å during the whole 100 ns of the production step ( g 10a). Analysis of this plot indicates the trajectory reached to almost a steady state after 40 ns (variation of the RMSD is less than ±1Å). Twenty snapshots were taken from the system every 5 ns and visually inspected for the residues that are engaged in the interactions. In all of the samples, both key arginine residues (R88 and 92) have fully interacted with naproxen through electrostatic forces. In addition, analyzing the trajectory for h-bond formation shown, there is also at least one strong h-bond (acceptor-donor distance <3 Å) with naproxen throughout the simulation (e-value <0.05) (Fig 11a). This is a very promising nding, since the hydrogen bonding is crucial for stabilizing the ligand inside the receptor pocket.
The interaction energy was also completely in accordance with the other results especially the MM-GBSA, docking, and number of H-bond. It starts at -190.9 kcal/mol for the rst frame and continues fairly steady to the end of the simulation with an average value of -184.4 Kcal/mol for all 50.000 frames.
Fluctuation for each residue of this domain was demonstrated in the RMSF plot (Fig 10b). The average value for all of the residues is almost 4 Å with a lower and upper limit between 2.2 Å and 7.4 Å. These values indicate a medium uctuation in the whole protein. The results also have shown that the key arginine residues reside in relatively low uctuating region which is an indication of their important roles. Moreover, almost all regions that had a higher RMSF value have belonged to the loop structures.
All the above parameters were also investigated for the C terminal domain. As demonstrated in g 10c and d, this domain had relatively higher values for the RMSD and RMSF plots. For example, although the system correctly heated and equilibrated, however, the RMSD plot did not reach a perfect steady-state (the values gradually increased) until 80 ns. Likewise, the RMSF values for this domain approximately 2 Å higher than the N terminal domain (of course these are two different independent domains and it is not correct to compare them with each other). This is completely an expectable nding because of the nature of this domain which is designed for the dimerization, a process that requires higher exibility not even in the contacting interface but also in the whole domain. The average RMSD value for this domain is about 5.81Å that is about 0.7 Å higher than the N terminal domain.
The interaction strength of naproxen with this domain is considerably lower than to the N terminal domain which was already speci ed by the MM-GBSA evaluation and also docking study. The interaction energy is started from -107.8 Kcal/mol and gradually decreased to lower than -250 Kcal/mol by 30 ns. However, it suddenly increased to more than -20 Kcal/mol by 35 ns. After this point, the interaction energy gradually become stronger until the end of simulation which reached -73 Kcal/mol with an average of -96.5 Kcal/mol for the whole period of simulation. This is about half the strength of naproxen interaction energywith the N terminal domain. Superimposing the number of H-bond on the interaction energy plot (Fig 11b) was revealed the sharp decrease in the interaction energy is probably due to the loose H-bonds at that time. Similar to the N terminal domain, 20 snapshots were taken from the system and visually inspected for he existed interactions. As expected, there is no stable pattern of interactions between snapshots.

Discussion
In this study, we showed comparative interaction of naproxen to the N terminal RNA-binding domain and C terminal dimerization/oligomerization domain of the nucleocapsid protein of SARS COV-2. We showed several key residues of the N terminal domain coordinate with naproxen that would compete for RNA binding with nucleocapsid. We also stimulated the binding dynamics of naproxen to N and C terminal domains during 100 ns of using molecular dynamics.
Nucleocapsid is one of the structural proteins of SARS COV-2 that is important for many functions including viral ribonucleoprotein (RNP) complex, regulation of replication or transcription. Crystallographic studies of the SARS-CoV N terminal domain revealed that the domain is folded mostly with antiparallel β-sheets core with a positively charged protruding hairpin, similarly, the same folding pattern has been seen for modeled and crystal structure of SARS-CoV-2 nucleocapsid 9 . Although the overall structure N terminal domains are the same among many betacoronavirus, the surface charge distribution pattern is markedly different 8 indicating unique RNA binding module for SARS COV-2. The C terminal domain of nucleocapsid also folds into a conserved structural shape resembling the letter C in the monomeric form among coronaviruses. As it has been modeled in this study C terminal domain folds into extended conformation with the dimerization surface at the center of the domain. The amino acids at the dimerization surface have been described for SARS-CoV which are mainly involved in hydrophilic and hydrophobic interactions 33 . Truncated nucleocapsid protein of SARS-CoV missing amino 285-422 inhibited the dimerization of nucleocapsid and virion assembly 34 The molecular dynamics simulation that was done here, con rmed the data that were obtained by molecular docking and MM-GBSA evaluation and showed naproxen strongly interacted with the RNA binding pocket of the N terminal domain structure. By comparing with empty structure, it is clear that its exibility did not affect signi cantly when naproxen bound (data not shown). It also should be noted that an intrinsic property of NP is its exibility 35 . Our analysis following the other studies showed the two key arginine residues (R88 and R92) are bonded to naproxen in the whole simulation time. These residues are very important for the native RNA binding process and also present in the nucleocapsid of other viruses such as in uenza 30,32 . In silicomutagenesis of these residues, the a nity of naproxen to the binding site drastically decreased. In addition to the electrostatic interactions, a network of H-bond between the ligand, protein and surrounding water molecules contributes to the stabilizing of the naproxenN terminal domain complex. Figure x11 shows this effect clearly. In this gure you can, the interaction energy completely obeys from the H-bond count between naproxen and the structures (this also was seen for the C terminal domain). For the N terminal domain, these strong interactions mask the RNA binding groove and compete with the viral RNA binding which similar to other studies on the anti-viral effect of this drug 36 .We also examine the binding potential of naproxen to the C terminal domain because this requires protein dimerization which is essential for its normal function. Fig 10c, d show the way that this domain behaves. It has more exibility than the N terminal domain which is completely in accordance with other ndings that pointed out the exibility of the C-terminal of NP is reminiscent 37,38 .Possibly the higher exibility interferes with the strong binding of naproxen with the key residues at the dimerization interface. By the way, as described in the result, occupying this site by naproxen cannot hinder the dimerization process which was deduced from the protein docking study.
High-throughput virtual screening and repurposing of the existing drugs for targeting SARS COV-2 proteins are the two main strategies for the rapid development of new anti-SARS COV-2 drugs. Using target-based virtual ligand screening, 18 putative proteins encoded by SARS COV-2 were screened against the ZINC drug database and a local database.the results of this study showed three potential targets including 3-chymotrypsin-like protease (3CLpro), Spike, RNA-dependent RNA polymerase (RdRp), and papain-like protease (PLpro) with several lead compounds. However, such compounds were not identi ed for the nucleocapsid 39 After the recent global outbreak of SARS COV-2 full-genome sequencing and phylogenic analysis indicated that SARS COV-2 is a betacoronavirus and similar to severe acute respiratory syndrome (SARS) virus. The receptor-binding gene structure of SARS COV-2 is alike with SARS coronavirus, and there is the same receptor for cell entry in both 40 . Additionally, the determination of the detailed 3Dstructures of key virus proteins, it is very e cient to apply computer-aided drug design techniques to quickly identify promising drug repurposing candidates for virus elimination 41 . So, the aim of this insilico study was to determine the antiviral effect of naproxen on SARS COV-2 and the results of the present study showed that naproxen had interaction with two types of arginine, ARG A 49 and ARG A 45, which play a major role in RNA binding in whole transcriptional processes. Naproxen binds to the nucleocapsid site and does not allow the virus to bind to the cells. Previously, Zheng et al show that the NP of in uenza B virus (BNP) has a higher binding a nity to naproxen than in uenza A virus NP (ANP) and speci cally, naproxen targets the NP at residues F209 (BNP) and Y148 (ANP) 27 . In another study,Lejal et al by in silico screening showed that naproxen combined antiviral and anti-in ammatory effects by targeting both NP and COX2 in in uenza A virus 36 .
The result of the present study about COVID- 19 and other studies about in uenza show that nucleocapsid has a main role in the life cycle of coronaviruses and it is a target for naproxen. Nucleoprotein acts as a single-stranded RNA binding protein that takes on a signi cant role in ribonucleoprotein particles (RNPs) transporting between the nucleus and cytoplasm and it is a structural protein of RNPs that is required for virus replication 42 . Docking studies show that several known drugs such as Car lzomib, Eravacycline, Valrubicin, Lopinavir, and Elbasvir. Car lzomib, Streptomycin, and formoterol have an inhibitory effect against SARS COV-2 43,44 . These drugs in-comparison with Naproxen are expensive and additionally, Naproxen is a drug in controlling some symptoms of these patients, such as fever and myalgia, so, it seems that is preferred over other drugs. Therefore, due to the antiviral role of naproxen, it is recommended to use naproxen in patients infected with SARS COV-2 without drug contraindication to help control fever and virus clearance time.

Conclusion
The urgent demand for the development of new antiviral drugs for the treatment of COVID-19 patients has led us to repurpose a commonly used antipyretic and analgesic drug for speci c treatment of SARS COV-2 infect patients. In this study using molecular docking and molecular dynamics analysis, we showed that naproxen interacts speci cally with some key amino acids in the RNA binding pocket of the N terminal domain of SARS COV-2 nucleocapsid. The results of this study also indicate naproxen might compete with viral RNA for biding with nucleocapsid therefore inhibiting viral capsid assembly and viral particle formation. Based on promising results obtained in this study we suggest further in vitro experiments and clinical trials to use naproxen in the treatment of COVID-19 patients.      The C terminal domain was dimerized while naproxen bound to its pocket. This is suggested that the compound may fail from inhibiting the dimerization process which is vital for the correct NP function.

Figure 10
Backbone RMSD, whole residue RMSF and interaction energy plots of naproxen bound N (a, b) and C terminal domains (c, d). The RMSF plot of both proteins showed considerable uctuation in the structures, although it is higher for the C terminal domain (d). The yellow highlights in (c) indicate the location of the residues in the binding pockets.