Network pharmacology approach to decipher signaling pathways associated with target proteins of NSAIDs against COVID-19

Non-steroidal anti-inflammatory drugs (NSAIDs) showed promising clinical efficacy toward COVID-19 (Coronavirus disease 2019) patients as potent painkillers and anti-inflammatory agents. However, the prospective anti-COVID-19 mechanisms of NSAIDs are not evidently exposed. Therefore, we intended to decipher the most influential NSAIDs candidate(s) and its novel mechanism(s) against COVID-19 by network pharmacology. FDA (U.S. Food & Drug Administration) approved NSAIDs (19 active drugs and one prodrug) were used for this study. Target proteins related to selected NSAIDs and COVID-19 related target proteins were identified by the Similarity Ensemble Approach, Swiss Target Prediction, and PubChem databases, respectively. Venn diagram identified overlapping target proteins between NSAIDs and COVID-19 related target proteins. The interactive networking between NSAIDs and overlapping target proteins was analyzed by STRING. RStudio plotted the bubble chart of the KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway enrichment analysis of overlapping target proteins. Finally, the binding affinity of NSAIDs against target proteins was determined through molecular docking test (MDT). Geneset enrichment analysis exhibited 26 signaling pathways against COVID-19. Inhibition of proinflammatory stimuli of tissues and/or cells by inactivating the RAS signaling pathway was identified as the key anti-COVID-19 mechanism of NSAIDs. Besides, MAPK8, MAPK10, and BAD target proteins were explored as the associated target proteins of the RAS. Among twenty NSAIDs, 6MNA, Rofecoxib, and Indomethacin revealed promising binding affinity with the highest docking score against three identified target proteins, respectively. Overall, our proposed three NSAIDs (6MNA, Rofecoxib, and Indomethacin) might block the RAS by inactivating its associated target proteins, thus may alleviate excessive inflammation induced by SARS-CoV-2.

. Later, the infectious disease experts detected severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), can rapidly transmit from person to person through interaction or respiratory droplets 2 . As a consequence of its tremendous spread globally, WHO announced a changing level from epidemic to pandemic disease (COVID- 19) on March 11, 2020 3 . Although the symptoms were identical to pneumonia, however, many COVID-19 patients showed no physical sign, thus transmitting the virus to others, as silently spread 4 . Due to a reliable vaccine's unavailability, clinicians utilize antiviral drugs and NSAIDs as a significant viable option for COVID-19 patients 5 . A recent study has reported that the use of NSAIDs is safe for COVID-19 treatment without exposing specific adverse effects 6 . Though there is a lack of evidence whether combined NSAIDs treatment could worsen COVID-19 symptoms 7 , but researchers suggested that anti-inflammatory therapies might suppress the fatal cytokine storm of COVID-19 patients 8 . Additionally, WHO announced that no evidence of unwanted side effects was reported, particularly the risk of death with NSAIDs' administration in COVID-19 patients 9 .
NSAIDs are commonly used to treat diverse anti-inflammatory symptoms due to their excellent therapeutic efficacy 10 . For example, some evidence suggests that NSAIDs are associated with mitigating depression, bladder function recovery, reduction of psychiatric events, and a decrease in cancer risk, all of which connected directly to anti-inflammatory effects 11 . Most NSAIDs are known as inhibitors of COX-1 (Cyclooxygenase-1) or/and COX-2 (Cyclooxygenase-2) involved in the synthesis of prostaglandin and thromboxanes 12 . Recently, both COX-1 and COX-2 are expressed in inflamed tissues constitutively, and Indomethacin inhibited the two forms of COX effectively 13 . Also, Indomethacin is a potent NSAIDs against rheumatoid arthritis, used as an immune response enhancer against HIV (Human Immunodeficiency Virus), inhibiting harmful immune response induced by COX-2 13,14 . Generally, COX-1 is expressed in most normal cells, and COX-2 is induced by an abundance of physiological stimulus 15 . An animal test demonstrated that the inhibition of COX-1 up-regulates COX-2 expression level, results in the prevention of aggravated effects against inflammatory response 16 . It was reported that the administration of dual COX-1/2 inhibitors slightly diminished viral DNA replication but could not induce viral DNA cleavage 17 . Hence, COX-1/2 inhibitors are involved in host immune responses while cannot take effect on pathogens. However, one potential drug is Indomethacin, which possesses both anti-inflammatory

Results
Information of NSAIDs. A total of nineteen NSAIDs and one prodrug (FDA approved) were selected. Table 1 displayed the NSAIDs' chemical information and TPSA (Topological Polar Surface Area). Among the selected NSAIDs, nineteen NSAIDs were found as an active drug, and one "nabumetone" was a prodrug, and its metabolite form is 6-methoxy-2-naphthylacetic acid (6MNA). Figure 2 exhibited the chemical structure of these NSAIDs.

NSAIDs connected to the 781 target proteins or COVID-19 targeted proteins. A total of 781
NSAIDs related proteins were identified (Supplementary Table S1) through two public databases (SEA and STP). Figure 3A showed the overlapping target proteins (228 target proteins) selected from the two databases (Supplementary Table S2). The number of 466 COVID-19 targeted proteins was identified from the PubChem database (Supplementary Table S3). Figure 3B illustrated that the final 26 overlapping target proteins were selected between 228 overlapped target proteins and 466 COVID-19 targeted proteins (Supplementary Table S4). Figure 4 demonstrated that the final 26 overlapping target proteins were selected by STRING, which represents 26 nodes   The nodes indicated a total number of signaling pathways, target proteins, and NSAIDs. The edges represented the relationships of the three components. The STN relationship suggested that the network might be potential therapeutic efficacy against COVID-19. The STN network displayed that 13 target proteins associated with 26 signaling pathways built by a size map. Among the 13 target proteins, both MAPK8 and MAPK10 have the highest degree (22), followed by MAPK14 (15), BAD (7), and CXCL8 (7). Additionally, among the 19 NSAIDs, Diclofenac has the greatest degree (10), followed by 6MNA (9), Fenoprofen (8), and Sulindac (8).

Discussion
STN networking analysis demonstrated that the therapeutic effect of NSAIDs against COVID-19 was directly related to 26 signaling pathways-13 target proteins-19 NSAIDs. The results of the KEGG pathway enrichment analysis of 13 target proteins suggested that 26 signaling pathways were associated with the occurrence and development of the COVID-19 symptoms. The relationships of 26 signaling pathways with COVID-19 symptoms were succinctly discussed as follows. PPAR (Peroxisome Proliferator-Activated Receptor) signaling pathway: a report shows that PPARγ (Peroxisome Proliferator-Activated Receptor-gamma), PPARα (Peroxisome Proliferator-Activated Receptor-alpha), and PPARβ/δ (Peroxisome Proliferator-Activated Receptor-beta/ delta) agonists have anti-inflammatory and immunomodulatory functions 25 . MAPK (Mitogen-Activated Protein Kinase) signaling pathway: The mechanisms of p38 MAPK inactivation might be an effective therapy against the SARS infected cells 26 . Additionally, MAPK stimulates cytokine production such as IL-10 (Interleukin 10), TNF-α (Tumor Necrosis Factor-Alpha), IL-4 (Interleukin 4), and IFN-γ (Interferon-gamma) 27 . It is evident that MAPK inhibitor can alleviate inflammatory responses against COVID-19. ErbB (Erythroblastic Leukemia Viral Oncogene Homolog) signaling pathway: ErbB signaling reduces the proinflammatory activation in cardiac cells 28 . RAS (Renin-Angiotensin System) signaling pathway: The inhibition of ACE (Angiotensin Converting Enzyme) connected to RAS signaling pathway could reduce tissue damage in COVID-19 patients 29 . Thus, the study indicates that blocking of the RAS signaling pathway can reduce inflammatory response level. cGMP-PKG (Cyclic GMP-Protein Kinase G) signaling pathway: the activation of cGMP-PKG signaling inhibits inflammatory response in the prostate, and also decreases CCL5 (C-C Motif Chemokine Ligand 5) released in CD8 + T cells (Cluster of Differentiation 8 T cells) 30 . cAMP (Cyclic Adenosine Monophosphate) signaling pathway: the elevation of cAMP leads to diverse cellular effects, such as airway smooth muscle relaxation, repressed effects on cellular inflammation, and immune responses 31 . NF-κB (Nuclear Factor kappa-light-chain-enhancer of activated B cells) signaling pathway: activation of the NF-κB signaling pathway gives rise to the inflammation induced by www.nature.com/scientificreports/ www.nature.com/scientificreports/ the SARS-CoV infection. In contrast, NF-κB inhibitors are the potential antivirals, even against SARS-CoV and can also contribute to other pathogenic human coronaviruses 32 . FOXO (Forkhead box protein O1) signaling pathway: decrease of FOXO3 (Forkhead box protein O3) in T cells inhibits apoptosis, enhances multifunction of CD8 cells, and elevates viral control 33 . Sphingolipid signaling pathway: Sphingolipids play a vital role to protect the lung from damages, and its control may give a good therapeutic efficacy 34 . WNT (Wingless/Integrated) signaling pathway: WNT signaling involves with the prime inflammatory pathways like intestinal inflammation. Also, the understanding mechanism of WNT ligands and cytokines manifest new treatment strategies for chronic colitis and other inflammatory diseases 35 . VEGF (Vascular Endothelial Growth Factor) signaling pathway: a report suggested that the activation of ACE2 (Angiotensin-Converting Enzyme 2) inhibits VEGFA (Vascular Endothelial Growth Factor A), which elevates vascular permeability and severity of endothelial damage 36  www.nature.com/scientificreports/ (Interleukin-17) signaling pathway: IL-17 receptor inhibitors are widely used to ameliorate the inflammatory acuteness to date. Furthermore, it is a potential target to suppress severe inflammation induced by COVID-19 41 . Fc epsilon RI signaling pathway: Fc epsilon RI interconnecting causes mast cell degranulation and synthesis of proinflammatory mediators 42 . TNF (Tumor Necrosis Factor) signaling pathway: TNF deficit is associated with dysfunctional secretion of inflammatory cytokine, leading to lung pathology and death during respiratory poxvirus infection, and thus TNF is a very significant element for regulating inflammation 43 . Neurotrophin signaling pathway: COVID-19 causes severe brain damage and destruction of the central nervous system derived from neurotrophin 44,45 . Insulin signaling pathway: Obesity-oriented insulin resistance is associated with the induction of proiifnnflammatory macrophage, leads to inflammation of adipose tissue 46 . GnRH (Gonadotropin-Releasing Hormone) signaling pathway: BBB (Blood Brain Barrier) disrupted by a viral infection, lymphocytes (B and T cells), monocytes, and granulocytes can penetrate in the brain parenchyma, which induces inflammation, resulting in dysregulation of GnRH neurons. Additionally, the inflammation of GnRH neurons inhibits GnRH transport through proinflammatory cytokines by impairing the cytoskeleton 47 . Prolactin signaling pathway: HIV patients have greater prolactin quantity compared to others. Besides, prolactin is regarded as a cytokine to stimulate the immune system 48,49 . Adipocytokine signaling pathway: Adipocytokines stimulate inflammation and disrupt immune response, which induces proinflammation in RA (Rheumatoid Arthritis) patients, leading to the development of bone damage 50 . Oxytocin signaling pathway: oxytocin interrupts proinflammatory cytokines' production by inactivating the eIF-2α-ATF4 (Eukaryotic Initiation Factor-2 alpha-Activating Transcription Factor 4) pathway 51 . Relaxin signaling pathway: relaxin inhibitors are good therapeutic targets to suppress inflammation caused by airway dysfunction 52 . AGE-RAGE (Advanced Glycation End product-Receptor of Advanced Glycation End product) signaling pathway in diabetic complications: The binding of AGE to its receptor RAGE can trigger cytokine production, thus, can cause tissue damages, while the blockage of AGE-RAGE can effectively ameliorate the inflammation 53 . Epithelial cell signaling in Helicobacter pylori infection: Helicobacter pylori interrupts T and B cell signaling to work the immune system. It is apparent that COVID-19 patients with Helicobacter pylori might be vulnerable to inflammatory responses 54 . Generally, SARS-CoV-2 invades the lungs and throat, induces excessive inflammation, which causes cytokines' secretion, resulting in severe complications like acute respiratory failure, pneumonia, and acute liver injury 55 . The leading cause is that the downregulation of ACE2 results in an angiotensin-II (Ang II) increase, which might spur the progression of COVID-19 through activated RAS 56 .
It was discovered that ACE2 is the functional receptor for the SARS-CoV-2 to trigger an infection in the lung alveolar epithelial cells. The internalization of the virus leads to downregulating the ACE2 on the host cell surface that could cause the elevation and demotion of Ang II and angiotensin 1-7 (Ang 1-7), respectively. Such an imbalance between these angiotensins may induce deleterious effects in the lung and heart. Thus, the SARS-CoV-2 affects humans through this mechanism [57][58][59][60] . Therefore, RAS blockade may restore the RAS balance by reducing the deleterious effects associated with Ang II 61 . Recent evidence showed that RAS inhibitors might be a promising target for relieving acute-severe pneumonia caused by the COVID-19 62 .
Interestingly, our study identified that the three target proteins (MAPK8, MAPK10, and BAD) are mainly associated with the RAS signaling pathway. MAPK8 and MAPK10 are members of the MAPK family which are the key mediators of inflammation, vasoconstriction, and thrombosis. Besides, overwhelming heart and lung injury in COVID-19 infection might be due to the overactivation of MAPK 63 . Therefore, these proteins' inactivation can also be a viable strategy for relieving COVID-19 induced organ injury. In addition, disposal of inflammatory cells by promoting cell death can be an innovative approach to control excessive inflammation. In this regard, the anti-apoptotic Bcl-2 gene's inhibition can also be a potential target to lessen inflammation 64,65 . Our findings also explored that MAPK8 MAPK10 and BAD proteins are related to three, 12, and two NSAIDs, www.nature.com/scientificreports/ respectively. During the molecular docking analysis, 6MNA, Rofecoxib, and Indomethacin revealed promising binding affinity along with the highest docking score against MAPK8, MAPK10 and BAD proteins, respectively. The result suggested that the three NSAIDs' key mechanism against COVID-19 might be to inhibit inflammation of lung cells by inactivating the RAS signaling pathway, and blockers of MAPK8, MAPK10 and BAD might suppress cytokine storm. Among various NSAIDs, Indomethacin is a current drug of interest to clinicians. Primary care physicians (New York) reported that Indomethacin had been prescribed to a large number of COVID-19 patients and observed quick recovery from cough, pain, and other symptoms. Such improvements and well-being benefits were not evident in the case of ibuprofen and hydroxychloroquine implementation ( Little 66 ). Notably,   18 . In contrast, 6MNA (an active metabolite of Nabumetone) and Rofecoxib are also potential antiinflammatory drugs, but studies disclosed that they are less potent than Indomethacin 73,74 . A clinical study recently demonstrated that Indomethacin has potent anti-inflammatory (decrease in IL-6) and antiviral efficacy; taking SRF (Sustained Release Formulation) with 75 mg twice a day achieved full effects in 3 days for patients infected by COVID-19 75 . Hence, such compelling outcomes indicate that Indomethacin can be considered alone or in combination for antiviral therapy, which may assist in combating human coronavirus (SARS-CoV-2). However, there are also some limitations to our analysis. This study has provided a predictive viewpoint of NSAIDs' mechanism  www.nature.com/scientificreports/ ence" mode 78 . Additionally, COVID-19 targeted proteins were identified by retrieving COVID-19 in PubChem (https:// pubch em. ncbi. nlm. nih. gov/). Also, the topological polar surface area (TPSA) value identified by Swis-sADME is included to verify the NSAIDS' cell permeability; particularly, its permeability is typically limited when the TPSA value is more than 140 Å 279 . The final overlapping proteins between NSAIDs and COVID-19-targeted proteins were identified and visualized by Venny 2.1 (https:// bioin fogp. cnb. csic. es/ tools/ venny/).
Signaling pathway enrichment analysis of overlapping proteins via KEGG database. KEGG database provides correlation of target proteins and signaling pathways through functional annotation 81 . A bubble chart of signaling pathways associated with COVID-19 infection plotted by RStudio. The bubble chart demonstrates a hub signaling pathway (Lowest rich factor) between NSAIDs and COVID-19 related proteins.
The construction of STN network. The Signaling pathway(s)-Target protein(s)-NSAIDs were used to construct a signaling pathway-target protein-NSAID (STN) network. In this STN network, different colors and shapes (nodes) stand for the signaling pathways (green circle), target proteins (yellow square), and NSAIDs (orange triangles). Gray lines (edges) indicated the interaction of signaling pathways-target proteins-NSAIDs. The STN networks were utilized to construct a size map, based on degree of values. In this network, green circles (nodes) represented signaling pathways; yellow squares (nodes) represented target proteins, and orange triangles (nodes) represented NSAIDs; its size represented degree value. The size of yellow squares stands for the number of connectivity with signaling pathways; the size of orange triangles stands for the number of connectivity with target proteins. The merged networks were constructed by using RStudio.
Preparation for MDT of NSAIDs. The ligand molecules were converted .sdf from PubChem into .pdb format using Pymol, and the ligand molecules were converted into .pdbqt format through Autodock.

Data availability
All data generated or analyzed during this study are included in this published article (and its "Supplementary Information S1" files).