In silico synergistic drug repurposing for combating novel coronavirus (COVID-19) outbreaks

As the number of novel coronavirus (COVID-19) cases continues to rise, there is a global need for rapid drug development. In this study, we propose a systems pharmacology approach to reposition FDA-approved drug candidates for coronavirus, identify targets and suggest a synergistic drug combination using network pharmacology. We collected 67 genes associated with coronavirus, performed an enrichment analysis to obtain coronavirus-associated disease- pathway and constructed protein-protein interaction (PPI) network based on 67 genes. Total 37 significant disease-pathways were retrieved, and associated FDA-approved drugs were listed for drug repurposing candidates. Our PPI network showed 51 targets from 67 genes and identified IL6 and TNF as potential targets for coronavirus. From the FDA drug list, we selected four drugs that are experimentally used or studied for coronavirus to construct two- drug combinations. From six drug-drug networks, we identified hydroxychloroquine + ribavirin combination had the highest number of overlapping targets (IL6, IL2, IL10, CASP3, IFNA1) from PPI network target list, suggesting a potent synergistic drug combination for coronavirus. With the aim to support the rapid drug development, we suggest a new approach using systems-level drug repurposing for COVID-19 treatment.


Introduction
Coronaviruses (CoVs) are enveloped, single-stranded RNA viruses that cause a pneumonia-like disease 1  (https://www.cdc.gov/coronavirus/2019-ncov/locations-confirmed-cases.html#map).. One research team in Guangzhou, China, has reported that, at the whole-genome level, COVID-19 is 99% identical to coronavirus isolated from pangolins, ant-eating mammals often used in traditional Chinese medicine 4 . The infection, which exhibits clinical symptoms of fever, dry cough, dyspnea and headache, produces a progressive respiratory illness due to alveolar damage that is fatal in an uncertain percentage of cases 4 . To date, no specific antiviral or therapeutic agents are available to treat coronaviruses, with the only management strategy being supportive care 2,3 . As the number of COVID-19 cases continues to rise, there is a global need for candidate intervention approaches. In this study, we suggest strategies that might be options for resolving the current coronavirus outbreak.
Rapid virus evolution and the threat of antiviral resistance has motivated research investment in various drug-discovery strategies 5 . One strategy for more timely introducing novel antiviral drugs is 'drug repurposing'. Drug repurposing (or repositioning) is the strategy of creating novel clinical opportunities for known approved or investigational drugs that are different from the original medical indication 6 . Drug repurposing offers three benefits for drug development. First, safety concerns are diminished because the drug has already been found safe in clinical models 7 . Second, development time and costs can be reduced in preclinical, phase I and II studies 8  Computational research is a promising time-saving alternative to experimental research that may identify novel targets and pathways 9 . Here, we identified diseases and pathways associated with coronaviruses in silico based on coronavirus-associated genes that might suggest additional drug repurposing options. First, we propose a systems pharmacology approach to reposition FDA-approved drugs for coronavirus diseases by systemically incorporating associated pathway. Next, we construct a protein-protein interaction (PPI) network based on coronavirus-associated genes to identify potential targets. Lastly, we suggest a polypharmacology-based paradigm, such as compound combinations for designing multi-targeted therapy to achieve more effective clinical responses 10 . 'Synergistic effect' is the result of combining two or more compounds that produce a greater effect than 'additive effects' 11 . Despite the significance of compound combinations in therapeutics, compound combinations in computational research is currently limited 12 . Herein, we suggest systems-level drug-drug interaction networks that capture both on and off-target effects that may reveal functional links between coronavirus, and a drug combination that may synergistically inhibit coronavirus.

Methods
Drug repurposing candidate study: Coronavirus target gene search and associated disease search Synergistic drug study: Drug-Drug network Four drugs were selected from retrieved FDA drug list. Two drug combination networks were obtained from STITCH (http://stitch.embl.de/) database. All networks were constructed within high confidence (0.7) minimum required interaction score.
FDA-approved treatments in accordance with coronavirus-associated diseases from KEGG database were obtained from FDA and Mayo clinic (https://www.mayoclinic.org/).. Diseases that had P-values over 0.05 were considered significant and listed in Table 1. Gene count represents the number of genes from 67 genes that are associated with the disease. Interestingly, several drugs, including oseltamivir, interferon and corticosteroids, have been used in patients with coronavirus (SARS or MERS), a finding in accord with our in silico results 13 . Moreover, a recent study on COVID-19 reported that chloroquine or hydroxychloroquine (for malaria), which is ranked in our disease results, showed antiviral efficiency in vitro 14 . In the case of recent studies on biomarkers, angiotensin-converting enzyme 2 (ACE2) has received attention as a new biomarker of SARS and COVID-19, a finding in agreement with treatments for viral myocarditis and Reninangiotensin system both of which use ACE inhibitors, such as enalapril and lisinopril as treatments.
Therefore, coronavirus-associated disease-pathways and FDA-approved drug list were proposed for drug repurposing candidates and further examined to uncover molecular mechanisms in host infection against COVID-19 4 .
Based on retrieved 67 genes, protein-protein interaction (PPI) network was retrieved from STRING database and visualized using Cytoscape (Figure 2, Figure 3). The node size reflects the number of interactions (degree) between proteins, and proteins with high number of degree were considered potent targets in our study.
From our in silico PPI network, 51 proteins from 67 genes lists were identified to be connected to form a network (S2). The proteins that showed the highest degree of interaction among all proteins was IL-6, followed by TNF, which can be considered potential targets for future coronavirus studies. Among these targets, DDP4 was recently reported to be a functional receptor for the MERS virus, a finding in agreement with our in silico results 3 . We also identified ACE2 that has also received attention as a new biomarker of SARS and COVID-19, as confirmed by our search 4 .
Lastly, we applied the systematic dynamics that can be investigated through drug-drug network to observe coronavirus targets associated within the predicted drug combinations.
Based on the disease and FDA-approved drug list (Table 1), we selected four diseases (influenza A, Malaria, Hepatitis C, and viral myocarditis), and matching drugs (oseltamivir, hydroxychloroquine, ribavirin, and lisinopril) that are currently used in clinic or within study for COVID-19 to construct drug-drug networks (Figure 4, Figure 5).
Total six networks were constructed, and only four networks were interconnected to each other ( Figure 5 a-d). Figure 5-a, Hydroxychloroquine + Ribavirin combination had the highest number of overlapping targets compared to PPI target lists (5 targets: IL6, IL2, IL10, CASP3, IFNA1), followed by Figure 5 Figure 1 Candidate study: data collection and disease-pathway enrichment Target study: Data collection and PPI network construction Protein-protein interaction network Drug combination approach for drug repurposing Supplemental information.pdf