Deciphering the molecular pathogenesis behind neurological manifestations of SARS-CoV-2 and drug repurposing, a systems biology approach

Introduction: As the COVID-19 pandemic spreads worldwide, reports about the neurological complications of SARS-CoV-2 are excessively increasing. However, there is still insucient high-throughput data on neuronal cells infected with SARS-CoV-2 to help predict its neural pathogenesis. HCoV-OC43 is another member of the beta coronavirus family that has conrmed neuro-invasive effects and has available neural omics data. This study predicts the critical genes, biological processes, and pathways mediating in SARS-CoV-2 neurological manifestations using a systems biology approach. Method: We retrieved raw data related to SARS-CoV-2 and HCoV-OC43 infections from gene expression omnibus datasets (GSE147507 and GSE13879 respectively). We constructed gene regulatory networks for both infections, detected signicant regulatory motifs by FANMOD software, and created their subnetworks. We also constructed PPI networks and identied the MCODE clusters. In the intersection of merged subnetworks of two viruses, the most critical genes were veried in GRN & PPI networks. We drug-repurposed for the selected target genes and performed the functional enrichment analysis using DAVID and String databases. Results: Some of the top KEGG pathway results included NF-kappa B, Toll-like receptor, NOD-like receptor, MAPK, and Neurotrophin signaling pathways. The most essential identied genes included IL6, TNF, HOXA5, POU2F2, ITGB3, STAT1, YY1, E2F6, ESR1, FOXO3, FOXO1, MEF2A, ATF3, ATF4, DDIT3, TCF4, BCL2L2, and BMP4. These genes were also involved in mechanisms of other viral infections of the nervous system. This study repurposes nine medicines with effects on COVID-19 neurological complications. Some of the repurposed drugs were previously registered in clinical trials for COVID-19 treatment. Conclusion: We recommended some identied crucial genes and medications to investigate further their potential role in treating COVID-19 neurological complications. transcription and gene We the host response protein-protein interaction their possible shared molecular mechanisms to predict the molecular mechanisms of the new responsible for COVID-19 neural manifestations. We compared the host response's gene regulatory networks to the two infections to predict crucial genes accountable for their neural The crucial shared genes were then selected as target genes to repurpose new drug candidates, and the new repurposed drug candidates were then validated using the experimental in vitro study model. The study has investigated the effect of the SARS-CoV-2 on the lung epithelial cell lines only after 24 hours post-infection. After 24 hours of infection by SARS-CoV-2, the total RNA of the infected and mock cells were extracted and sequenced using an Illumina NextSeq 500 platform. We selected the differentially expressed genes (DEG) of the two cell lines based on log 2 FC>0.5 or <-0.5fold and false discovery rate p-value <0.05. study investigated the genes mediating in the central nerve system (CNS) involvement of SARS-CoV-2 using the shared genes between both OC43 and SARS-CoV-2 GRN and PPI networks. The most critical genes included six up-regulated shared genes TNF, HOXA5, POU2F2, ITGB3, and STAT1) and 12 down-regulated shared genes (YY1, E2F6, ESR1, FOXO3, FOXO1, MEF2A, ATF3, ATF4, DDIT3, TCF4, BMP4). other studies critical genes and beyond."

Integrated Protein-Protein Interaction rEference) (http://cbdm-01.zdv.uni-mainz.de/) separately to retrieve the protein interaction maps. We selected the protein-protein interactions with a con dence score threshold of 0.7 for each gene set. We then imported the interaction tables to Cytoscape software (version 3.7.1) (https://cytoscape.org/) to visualize them and perform topological network analysis (Shannon, Markiel et al. 2003). To map the third PPI network for each gene set, we also used the Bisogenet app to retrieve the HPRD database's interactions. We merged the three PPI networks in Cytoscape and analyzed it, and nominated the top 10% of the nodes having the highest degree and betweenness centrality as the hub and bottleneck nodes, respectively. Hub genes are de ned as highly connected nodes in the PPI network, and networks are usually sensitive to delete the hub genes. Both hub and bottleneck nodes are generally necessary for fundamental cellular processes. (Liu, Yi et al. 2019).
We used the Molecular Complex Detection (MCODE) app to screen the clusters and nding their seed genes. The clusters are highly interconnected regions in a PPI network, and seed genes have the highest degree value in a cluster (Bader and Hogue 2003). The clusters with an MCODE-score bigger than ve were then merged to perform functional enrichment analysis.

Identi cation of miRs suppressing TFs
MicroRNAs post-transcriptionally suppress transcription factors. We considered TFs as target genes and retrieved experimentally validated miR-TF interactions from miRTarBase release 8.0.

GRN construction and motif detection
We integrated the four types of regulations (TF-gene, miR-gene, TF-miR, and miR-TF) to construct the transcription factor-microRNA-gene regulatory network visualized by Cytoscape software (version 3.7.1).
We then merged and exported them to FANMOD software to identify the network's signi cant 3-node regulatory motifs. FANMOD is a tool for network motif detection that uses a novel algorithm called RAND-ESU (Wernicke and Rasche 2006). We evaluated each of the possible 3-node motif types for their signi cance using random network generation and built the random networks 1000 times to compare them with the original input network. The 3-node motif types having Z-score >2.0 and p-value < 0.05 were considered signi cant. (The other FANMOD parameters were the same as our previous work with DOI: 10.1080/17435390.2018.1513090). We then created the motif-related speci c subnetworks for each of the signi cantly scored motifs separately. The motifs with the same FANMOD ID were merged and considered as a unique motif subnetwork.

Functional enrichment analysis
We performed functional enrichment analysis for Gene Ontology (GO,www.geneontology.org) and Kyoto Encyclopedia of Genes and Genomes (KEGG, https://www.genome.jp/kegg/) on the nodes of the merged PPI clusters and genes of the merged GRN motifs separately using the STRING (string-db.org) and DAVID databases (https://david.ncifcrf.gov) (Huang da, Sherman et al. 2009) (The Gene Ontology Consortium 2019). Biological processes and biochemical pathways with p-value < 0.05 were considered statistically signi cant. We selected biological processes and KEGG pathways, which were shared between the enrichment results of both databases.

Identi cation of critical genes and drug repurposing
HCoV-OC43 showed neuroinvasive effects in mice and humans; besides, neurological manifestations in patients with SARS-CoV-2 are recently reported (Asadi-Pooya and Simani 2020). Therefore, we used GSE13879, in which the human neuronal cell line was infected with HCoV-OC43 and compared its gene regulatory network with the SARS-CoV-2 network to predict genes responsible for the central nervous system (CNS) involvement in SARS-CoV-2 patients. Therefore, the shared genes between the intersections of the motif-related subnetworks of HCoV-OC43 & SARS-CoV-2 were selected as target genes to repurpose new drug candidates. The genes that played the role of the hubs/bottlenecks of the GRN/PPINs and the genes participating in PPI MCODE clusters were considered the most critical target genes for drug repurposing. We enriched these target genes using the STRING and DAVID databases and reported biological processes and KEGG pathways with p-value <0.05 enriched in both databases. We retrieved the drug-gene interactions among the medications from three drug databases, including Drug Gene Interaction Database (DGIdb) version 3.0.2 (http://dgidb.org/), PharmGKB (https://www.pharmgkb.org/) and DrugBank database version 5.1.6 (https://www.drugbank.ca/) (Wishart, Feunang et al. 2018). We then constructed and visualized the interaction network between the selected genes (available in GRN and PPI networks) and their related drugs using the Cytoscape software. (The medications targeting all the shared target genes were also identi ed and represented in online Supplementary table S14 & S15) We validated our repurposed drug candidates for possible treatment of neural manifestations of COVID-19 by comparing them with the medications registered in clinical trial platforms for COVID-19 treatment, extracted from the DrugBank database (Wishart, Feunang et al. 2018). The repurposed drugs were also discussed and veri ed by other experimental reports available in the literature review. The study design and the scheme of the work ow are summarized in g.1.
In this study, we rst investigated how DEGs and their related TF/miRs are involved in biological processes and pathways related to the host response to HCoV-OC43 & SARS-CoV-2 infections using gene regulatory network and protein-protein interaction network. We extracted the DEGs based on p-value<0.05 and log 2 fold change (FC) <-0.5 or >0.5. The DEGs identi ed for the only microarray dataset available for HCoV-OC43 and RNA sequencing results of two lung epithelial cell lines (NHBE & A549) infected with  SARS-CoV-2 are available in table 1 and online supplementary table S1. We rst elicited the DEGs related to HCoV-OC43 24h post-infection, and the shared DEGs between 48&72h. The DEGs related to SARS-CoV-2 were identi ed for the two cell lines separately and merged to produce the SARS-CoV-2 up and downregulated gene sets separately for further analysis.
Protein-protein interaction network (PPIN) construction: We retrieved the protein-protein interaction maps among the DEGs using the HPRD (by BIOSOGENET app), STRING, and HIPPIE databases. We then merged them to create a PPIN for up and down-regulated genes of each infection, separately. The PPI networks were visualized and topologically analyzed using the Cytoscape software. The number of nodes and edges of each network are represented in table 2, and the topological network analysis results are available in online supplementary table S2. We selected the top 10% of the PPI network nodes having the highest degree and betweenness centrality as the hub and bottleneck nodes, respectively (available in online supplementary table S3).
We investigated the shared DEGs between the HCoV-OC43 (up and down-regulated) networks related to 24h post-infection with the SARS-CoV-2 PPI networks. The few numbers of DEGs shared between the two infections are represented in online supplementary g.S1 and g.S2 The number of the shared genes was not su cient (six down-regulated and 20 up-regulated shared genes) to construct an intersectional network for further analysis. The shared DEGs between the PPI networks related to HCoV-OC43 48&72h post-infection and SARS-CoV-2 were then investigated and used for further analysis.

MCODE cluster detection and functional enrichment analysis
The identi ed clusters of each PPI network and their seed genes using the MCODE plugin are represented in g.2 The number of PPI clusters for HCoV-OC43 up and down-regulated genes were eighteen and twenty-six subnetworks, respectively. Based on the MCODE-score of more than ve, we selected eight clusters of up-regulated and seven clusters of down-regulated genes.
We performed the Gene Ontology and Biochemical KEGG pathway enrichment analysis on the selected PPI clusters' union, using the STRING and DAVID database (based on p-value<0.05). We have described the results that were shared between the two databases in online supplementary tables S4 and S5. The Seed genes of the eight up-regulated clusters were RPLP2, NDUFA3, LSM8, ACKR3, FGA, CBX3, HOXC9, and SYT1. Seed genes of the seven down-regulated clusters included UFL1, APLN, TRA2B, DHX58, COG2, IARS, and HTR4.
The SARS-CoV-2 PPI up and down-regulated networks had nineteen and nine clusters, respectively. We selected the nine up-regulated clusters based on MCODE-score > 5 and the nine down-regulated clusters based on MCODE-score>3 (no down-regulated cluster score was higher than ve). We performed gene ontology and KEGG pathway enrichment analysis on the clusters' merged union, using the STRING database and the DAVID database (based on p-value<0.05). We described the shared biological processes and KEGG pathways in online supplementary tables S6 and S7. The Seed genes of the top nine up-regulated clusters included HLA-F, B2M, UBE2W, IFI44, FGA, TGM1, DDX28, IL1A, and HAUS3. The down-regulated clusters' Seed genes were RAB26, PDE5A, COL20A1, FBXW9, CEP44, TSC22D3, METTL7A, PPARGC1A, and CENPA.

Gene Regulatory Network construction and motif detection
We extracted four types of regulatory relationships (TF-gene, miR-gene, TF-miR, and miR-TF) for each gene set. The results are available in table 3 and online supplementary table S8. We then imported the four regulatory relationships to Cytoscape software (version 3.7.1) and constructed the GRN for each gene set separately. We selected the top 10% of the GRN nodes having the highest degree and betweenness centrality as the hub and bottleneck nodes, respectively (available in online supplementary table S9).
The four types of regulations were then fed into the FANMOD software as an incorporated list for each gene set. The signi cant 3-node motifs of the GRNS were identi ed and represented in online supplementary g.S3-g.S6 (p-value<0.05, K-score>2). We selected the motifs with at least two different color edges (represented in g.3 and g.4). The selected motif-related subnetworks were created and visualized using the Cytoscape software (version 3.7.1). We merged the motif-related subnetworks with the same FANMOD ID and reported the subnetworks' intersection in g.5. Our motif detection results showed that all the four gene sets (up and down-regulated genes of HCoV-OC43, and SARS-CoV-2) had the same type of signi cant motifs with FANMOD IDs 78, 14, and 164.
Biological process and Biochemical pathway enrichment analysis: We performed enrichment analysis on the DEGs and also on the TFs included in the related DEGs of the merged union of each gene set's motif-related subnetworks, using the STRING and DAVID databases (p-value<0.05). The biological processes and KEGG pathways enriched in both databases are available in online supplementary table S10-S13.
The sub-networks' total union related to the HCoV-OC43 up and down-regulated subnetworks had 325 and 560 genes/TFs. The union of SARS-CoV-2 up and down-regulated genes had 23 and 5 genes/TFs, respectively.

The shared functional enrichment results between the PPINs & GRNs
We performed gene ontology and KEGG pathway enrichment analysis on the nodes of merged MCODE clusters of each PPIN and merged motif-related subnetworks of each GRN separately. The shared biological processes and KEGG pathways between enrichment results of PPIN & GRN of each gene set are available in tables 4 and 5.
Some of the top biological processes for HCoV-OC43 DEGs included translation, mRNA metabolic process, protein localization, protein modi cation, and response to virus and stress. Besides, the top ten biological process terms of SARS-CoV-2 up-regulated genes were mostly about immune response, cytokine-mediated, and interferon signaling pathways. Some of the top KEGG pathways of HCoV-OC43 DEGs included Parkinson's disease, Alzheimer's disease, Huntington's disease, NOD-like receptor signaling pathway, Measles, TNF, and NF-kappa B signaling pathway. Some of the top KEGG pathways of SARS-CoV-2 up-regulated genes were In uenza A, Herpes simplex infection, NOD-like receptor signaling pathway, Rheumatoid arthritis, Measles, TNF signaling pathway, Hepatitis C, and Cytokine-cytokine receptor interaction. The number of shared down-regulated DEGs between PPI and GRNs was insu cient to perform the enrichment analysis for SARS-CoV-2 down-regulated genes (online supplementary table S7 and S13).

Construction of target gene-drug interaction network and drug repurposing
To predict the genes possibly mediating in neural manifestations of SARS-CoV-2, we identi ed the intersection between the unions of the motif-related subnetworks of HCoV-OC43 and SARS-CoV-2 GRNs (up and down-regulated DEGs, separately). These intersections contained the shared TFs, miRs, and target genes. The number of shared genes between the up-regulated GRN networks was 31, and the number was 35 for the downregulated GRN networks, depicted in g.6 and g.7.
We also enriched the target genes using the STRING and DAVID databases and described the shared biological processes and KEGG pathway results (p-value <0.05) between the databases in online supplementary table S14 and S15. The top related KEGG pathways were NF-kappa B, Toll-like receptor, NOD-like receptor, In uenza A, Herpes simplex infection, Measles, HTLV-I infection, Hepatitis B, MAPK, and Neurotrophin signaling pathways.
We searched for the medications interacting with the 31 up-regulated and 35 down-regulated target genes in Drug Gene Interaction Database (DGIdb), PharmGKB, and DrugBank databases. In DrugBank, we retrieved interactions of the target genes with previously FDA-approved and pharmacologically active drugs. For the up-regulated shared target genes, we found 20 interactions in DrugBank, 184 interactions in DGIdb, and 68 interactions in PharmGKB and the overall number of the repurposed drug/compounds were 251 (online supplementary table S16). The numbers were 37, 150, and 42 interactions for the downregulated shared target genes, respectively, and the overall number was 182 (online supplementary table S17).
We extracted the drug-gene interactions for the six and twelve critical up and down-regulated target genes, respectively. Among the six up-regulated genes, four of them had drug interactions, and we visualized their 144 drug-gene interactions (129 unique drugs) using Cytoscape ( g.8). Among the twelve down-regulated genes, six had drug interactions. The number of drug interactions visualized for the six down-regulated genes was also 136. (133 unique drugs) ( g.8).
We then investigated our candidate drugs among the 563 medications currently available in clinical trials for their possible effect against SARS-CoV-2, extracted from the DrugBank database (https://go.drugbank.com/covid-19 ). The number of medications repurposed for the up-regulated critical target genes currently registered for SARS-CoV-2 treatment clinical trials was 35, and the number was 20 for the down-regulated (table 8).

Discussion:
This study analyzed and compared the DEGs related to lung epithelial cell lines (NHBE and A549) treated with SARS-CoV-2 and N-Tera2 differentiated human neuronal cells treated with HCoV-OC43 to investigate the possible molecular mechanisms behind the neurological manifestations of SARS-CoV-2 infection. We constructed TF-miR-gene regulatory and protein-protein interaction networks to identify the critical nodes (hubs, bottlenecks, motif members, and MCODE cluster members), biological processes, and pathways mediating in the two infections' pathogenesis. The shared critical genes between the HCoV-OC43 effect on neural cells and SARS-CoV-2 could shed light on the molecular mechanisms of brain conditions in COVID-19 patients (Fig. 9). Herein, we discuss and validate some of the predicted genes and pathways probably mediating in neural manifestations of COVID-19 using other experimental literature.
The SARS-CoV-2 RNA acts as a viral pathogen-associated molecular pattern (PAMPs) that can be identi ed by pattern-recognition receptors (PRRs) like toll-like receptors (TLRs) and NOD (nucleotidebinding oligomerization domain)-like receptors (NLRs) (Kopitar-Jerala 2015). The surface receptors were among the top enrichment results of KEGG pathways of the shared DEGs between the two infections and are the rst molecules activated in the innate immune system against the CNS pathogens. An in ammatory cascade initiates after triggering TLRs 3, 7, 8, and 9 by activating the NF-kB signaling pathway and IFN α/β/γ gene expression (Sabroe, Parker et al. 2008, Totura, Whitmore et al. 2015. The NF-kB pathway was also another pathway enriched for the shared DEGs. It regulates gene expression by kB sites present in promoter and enhancer regions of various essential genes such as chemokines, cytokines, adhesion molecules, and pro-in ammatory transcription factors. Therefore, it regulates neuronal survival and neuronal in ammatory reactions. The NF-κB also induces pro-IL-1β, pro-IL-18, TNF, and IL-6 (Tergaonkar, Correa et al. 2005, Tergaonkar 2006, Wong and Tergaonkar 2009). The TNF and IL-6 were identi ed as hub/bottlenecks in the GRN/ PPI networks of our in-silico analysis.
SARS-CoV-2 can also be recognized by NLRP3 (a kind of NLRs). NLRP3 forms an in ammasome complex with caspase-1 and cleaves IL-1β and IL-18 to mature forms . These cytokines (IL-1β, IL-18, TNF, and IL-6) induce further NF-κB nuclear translocation, activation of the JAK/STAT pathway, and phosphorylation of p38 MAPK (Battagello, Dragunas et al. 2020). The pathways were also included among our top ten enrichment results of the shared genes. Activation of p38s regulates immune response and in ammatory processes in SARS-CoV-2 infection (Feng, Fang et al. 2019, Grimes andGrimes 2020). In CNS, elevated activity of MAPK signaling can modulate neuronal survival and homeostasis (Feng, Fang et al. 2019, Grimes andGrimes 2020). Besides, in the JAK/STAT pathway, cytokines like INF and IL-6 bind to their receptors, induce JAK proteins cross-phosphorylation and then recruit STAT proteins. The phosphorylated STAT proteins then dimerize and translocate into the nucleus and regulate gene expression as transcription factors and (O'Shea, Schwartz et al. 2015). STAT1 and STAT3 were also identi ed as two crucial genes in our GRN/PPI networks. STAT1 has an essential role in the IFN signaling type I and type II and the JAK/STAT pathway (Pasieka, Cilloniz et al. 2011, Kulkarni, Scully et al. 2017. The role of STAT-1 has been previously elucidated in the innate immune response to other neurotropic viruses such as the severe acute respiratory syndrome coronavirus (SARS), Herpes simplex virus type 1 (HSV-1), and West Nile virus (WNV) (Pasieka, Cilloniz et al. 2011, Mahlakõiv, Ritz et al. 2012, Winkelmann, Luo et al. 2016).
The neurotrophin signaling pathway was another enriched result between the shared genes of our insilico study. The pathway makes crosslinks with various intracellular signaling cascades, including NF-kB and MAPK pathways. Neurotrophins, such as nerve growth and brain-derived neurotrophic factors, induce neurons' survival, development, and function (Reichardt 2006). In the nervous system, IL-6 and TNF-α are typically expressed at relatively low levels. However, their expression is up-regulated under different pathological conditions like in ammation and viral infections (Yang, Lindholm et al. 2002, Oh, McCloskey et al. 2010). SARS-CoV-2 can activate glial cells in the CNS and induce a pro-in ammatory state. IL-6 and TNF-α play essential roles in the Systemic In ammatory Response Syndrome (SIRS), leading to brain damage (Li, Fu et al. 2004, Liguori, Pierantozzi et al. 2020, Serrano-Castro, Estivill-Torrús et al. 2020, Wan, Yi et al. 2020). The IL-6 level also increases in other viral respiratory infections with neurological complications such as a human respiratory syncytial virus (RSV) and In uenza. It can be considered as an indicator of their neurologic prognosis (Aiba, Mochizuki et al. 2001, Kawashima, Kashiwagi et al. 2012, Morichi, Morishita et al. 2017. HOXA5 was identi ed as another critical gene in our in-silico analysis. It is a member of the HOX family of transcription factors expressed throughout adulthood, especially in glutamatergic and GABAergic neurons. It regulates many genes associated with neuronal survival and synaptic function (Lizen, Moens et al. 2017). HoxA5 is reported to regulate the viral immediate-early (IE) gene expression in herpes simplex virus (HSV). Besides, the IE gene has an essential role in acute viral replication and its latency in neurons (Mitchell, De Santo et al. 1993, Mitchell 1995. HOXA5 expression is also reported to change signi cantly in some other neurotrophic viral infections like Cytomegalovirus, Coxsackievirus B3 (CVB3), and lymphocytic choriomeningitis virus (LCMV). The viruses can infect neurons and cause meningitis and encephalitis (Ester 2011, Puccini, Ruller et al. 2014).
POU2F2 (Oct-2) was another identi ed crucial gene. It has an essential role in virus replication and is a member of the POU family. It is predominantly expressed in B cells, activated T cells, and the nervous system (Luchina, Krivega et al. 2003). Some POU family members, such as POU2F1, are reported to be necessary for viral DNA replication and gene expression in other viruses such as herpes simplex virus (HSV) (Ryan and Rosenfeld 1997). It can be hypothesized that SARS-CoV-2 elevates the expression of HOXA5 and POU2F1 to increase its viral proliferation possibly.
The ITGB3 gene, coding for the integrin β3 subunit, is expressed and enriched at cortical, hippocampal, and midbrain synapses in the brain (Varney, Polston et al. 2015). The integrin β3 mediates in HSV-1 cell entry by relocating HSV receptor nectin1, and thus HSV to cholesterol-rich microdomains of the membrane where TLR2 presents. Therefore, integrin β3 plays a vital role in endocytosis of the virus and initiation of the innate immune response (NF-κB activation and production of IFNα, IFNβ, IL2, and IL10) (Gianni, Leoni et al. 2012, Gianni, Leoni et al. 2013. The ITGB3 was up-regulated in our in-silico results.
Therefore, we suggest that SARS-CoV-2 is probably utilizing this upregulation to increase its entry. However, further experimental studies are required to con rm the prediction.
The antiviral response is also activated by Yin Yang 1 (YY1). It is a multi-functional transcription factor that can activate or repress gene expression in various cell types, including neurons (He and Casaccia-Bonne l 2008, Chen and Chan 2019). Some Viral infections down-regulate the expression of YY1 since it can mediate the antiviral innate immune response and regulate the production of interferon-beta (IFN-β) (Zan, Zhang et al. 2017). YY1-1 can also repress the transcription of many retroviruses such as human immunode ciency virus type I (HIV-1). It also contributes to a neurological disorder caused by the human T lymphotropic virus type 1 (HTLV-1) (Coull, Romerio et al. 2000, Wang andGoff 2020). We also identi ed the YY1 as a downregulated gene in SARS-CoV-2 and OC-43 infections. Therefore, it can be postulated that the virus is probably bene ting from the YY1-1 downregulation in its neural pathogenesis in COVID-19. E2F6 expression is a known mechanism that slows down or exits the cells from S-phase. Some viral proteins can inactivate E2F6 to extend the S-phase in virus-infected cells, such as human papillomavirus Our results showed that SARS-CoV-2 infection down-regulates the expression of E2F6. The E2F6 downregulation presumably contributes to the replication of the virus.
The Forkhead box O transcription factors (FOXO1 and 3) were the next identi ed critical genes. They mediate the regulation of the cell cycle, apoptosis, autophagy, and DNA repair. They also are reported to regulate neural cell survival, neuronal signaling, and stress responses in the nervous system Paik 2018, Schäffner, Minakaki et al. 2018). Our results showed that SARS-CoV-2 down-regulated the FOXO proteins similar to other viruses such as the Japanese encephalitis virus (JEV). JEV induces cell apoptosis in neurons by inhibiting the FOXO-signaling pathway. Therefore, it can lead to severe viral encephalitis in humans and other animals (Guo, Yu et al. 2018). Furthermore, FOXO proteins have a validated role in the pathogenesis of HIV-I-infection and its associated neurological complications (Cui, Huang et al. 2009). ESR1 (Estrogen receptor α) is present in the hypothalamus and amygdala regions related to human emotion and cognitive functions (Ma, Tang et al. 2014). It is also involved in the pathogenesis of hepatitis viruses (HBV & HCV) and their complications (Deng, Zhou et al. 2004, Zhai, Zhou et al. 2006, Watashi, Inoue et al. 2007. Neurological complications of these viruses range from peripheral neuropathy to cognitive impairment (Mathew, Faheem et al. 2016). SARS-CoV-2 is also a positive-sense RNA virus similar to HCV. Therefore, we predict that the molecular mechanisms behind the two viruses' similar neurological complications are probably the same, and we recommend the prediction for further experimental investigations.
Myocyte enhancer factor 2 (MEF2, isoform A) is also highly expressed in the brain. MEF2A is a downstream protein of NMDA receptor-mediated excitotoxicity in which excessive stimulation of the NMDA receptor contributes to the death of neurons. It is also reported to mediate in the neuropathogenesis of some other single-strand RNA-viruses such as HIV-1 (Kaul andLipton 2004, Yndart, Kaushik et al. 2015). The MEF2A was identi ed as a critical gene and is recommended for further studies in the neuropathogenesis of SARS-CoV-2, similar to HIV-1.
Activating transcription factors (ATFs) are members of the ATF/cAMP response element-binding protein (CREB) family. ATF3 and ATF4 usually are low in neural cells, but their expression increases rapidly in response to pathological stress (Lange, Chavez et al. 2008, Hunt, Raivich et al. 2012. Activation of innate and adaptive immune systems can induce the expression of ATF3. However, ATF3 acts as a negative regulator of the immune response (Hunt, Raivich et al. 2012). Our analysis showed that SARS-CoV-2 down-regulated the ATF3. Therefore, it can be suggested that down-regulation of ATF3 possibly mediates in hyper-activation of the immune system and neural manifestations of some patients with severe COVID-19. ATF4 is a downstream protein of a cellular protective signaling pathway called the Unfolded Protein Response (UPR). UPR and Autophagy pathways are tightly interconnected and are reported to play a vital role in some viral infections such as hepatitis C virus, herpes simplex virus type 1 (HSV-1), In uenza virus, and severe acute respiratory syndrome (SARS) (Tardif, Mori et al. 2004, Versteeg, van de Nes et al. 2007, Burnett, Audas et al. 2012, Sen, Balakrishnan et al. 2014, Sims and Meares 2019. We hypothesize that SARS-CoV-2 probably modulates the UPR/Autophagy signaling pathways to produce large quantities of viral proteins. Besides, ATF4 (CERB-2) interacts with Human T-cell leukemia virus type 1 (HTLV-1) tax protein to regulate its transcription. The virus causes a severe neurologic disorder called HTLV-1 associated myelopathy/tropical spastic paraparesis (HAM/TSP) (Ahuja, Kampani et al. 2006, Barbeau andMesnard 2011).
Bcl2l2 (Bcl-w) is an Anti-apoptotic member of The B-cell lymphoma-2 (BCL-2) family (Hartman and Czyz 2020). Various viruses are reported to interact with the Bcl-2 family to regulate the cellular apoptosis, including In uenza A virus, hepatitis B virus, hepatitis C virus, Epstein-Barr virus, vesicular stomatitis virus, human immunode ciency virus, and SARS-CoV. (Tan, Tan et al. 2007, Nencioni, De Chiara et al. 2009, Pearce and Lyles 2009, Busca, Saxena et al. 2012, Geng, Huang et al. 2012, Park, Kang et al. 2012, Ghigna, Reineke et al. 2013. Some of the viruses have previously been reported with neural complications. The In uenza A virus and SARS-CoV are reported to suppress BCL-W and Bcl-XL (another anti-apoptotic member of the family) to induce cellular apoptosis and subsequent tissue damage (Tan, Tan et al. 2007, Nencioni, De Chiara et al. 2009, Guan, Shi et al. 2012. Our results showed that SARS-CoV-2 also down-regulates Bcl-w, Similar to SARS-CoV. The Bcl-w probably also mediates in SARS-CoV-2 subsequent tissue damage in a similar way. BMP4 is one of the Bone morphogenetic proteins (BMPs), which are members of the transforming growth factor-beta (TGF-b) superfamily (Higashi, Tanaka et al. 2018). TGF-beta and BMP signaling pathways are activated following some neurotropic viral infections, such as reovirus. The signaling pathways are part of the host immune response and have a neuroprotective effect (Beckham, Tuttle et al. 2009). Besides, the synergistic activation of BMP and alpha-interferon signaling pathways is also reported to reduce the hepatitis C virus [136]. Based on our results, the SARS-CoV-2 down-regulated the BMP4. Therefore, we suggest that the virus probably is bene tted from the BMP4 down-regulation in infected neurons.
In this study, we investigated our 259 repurposed medicines indications and mechanisms using the DrugBank database. Our emphasis was on drugs that affect the nervous system; since SARS-CoV-2 is a neurotropic virus, and the immune system's hyperactivation has a vital role in its complications (Table 8). We also reported the drug candidates that were categorized as in ammatory drugs or antiviral drugs. We have reported forty-four drugs with immunomodulatory or immunosuppressive functions that have validated effects in treating in ammatory or autoimmune diseases such as rheumatoid arthritis, psoriatic arthritis, ankylosing spondylitis, Chron's disease, ulcerative colitis, systemic lupus erythematosus, and Behcet's disease (Suzuki Kurokawa and Suzuki 2004, Allison 2005, Potthast, Dressman et al. 2005, Klotz, Teml et al. 2007, Mease 2007, Lucas 2016, Corbett, Chehadah et al. 2017, Laurence L. Brunton 2018, Yang, Wu et al. 2018, Afra, Razmi et al. 2019). Ten antiviral medications (mostly anti-HIV-1 and hepatitis) were also among our repurposed drugs. For example, Hydroxychloroquine and chloroquine were two drugs reported to inhibit the viral entry by changing the endosomal PH and disrupting glycosylation of ACE2 (Vincent, Bergeron et al. 2005, Plantone and Koudriavtseva 2018, Devaux, Rolain et al. 2020).
Melatonin seems a suitable candidate in treating COVID-19 since it shows excellent anti-oxidative properties by directly scavenging free radicals and stimulating antioxidant enzymes. Furthermore, Melatonin has anti-in ammatory effects by reducing pro-in ammatory cytokines and balancing innate immune response's over-activation while promoting adaptive immunity [158,159]. The published reports related to Melatonin used in the animals with deadly viral infections such as Venezuelan equine encephalomyelitis virus (VEEV), Semliki Forest virus (SFV), and West Nile virus (WNV) showed that Melatonin is not viricidal. However, somewhat it reduces the severity of viral infections [158,159].
Adequate Pyridoxine supplementation can prevent polyneuropathy, which is the most common neurological complication associated with HIV [160,161]. Perhaps, Pyridoxine can be helpful in such kinds of pains in COVID-19 patients. We investigated the number of shared medicines between our list and the list of all registered drugs for COVID-19 clinical trials to verify our repurposed medications for their possible use in treating neural manifestations of COVID-19. Fifty-four of our repurposed medications were previously registered for investigations against COVID-19 (Table 8).

Conclusion
The current study revealed signi cant regulatory motifs and clusters of protein-protein interactions that play essential roles in the pathogenesis of HCoV-OC43 & SARS-CoV-2. With the accessible information on neuronal cells infected with HCoV-OC43, we predicted crucial genes, biological processes, and pathways in SARS-CoV-2 neurological manifestations that were mostly linked to activation of the innate immune system in CNS. This bioinformatics analysis can help shed light on molecular mechanisms and interactions involved in the neurological aspects of COVID-19. This study recommends some identi ed crucial genes and medications for further investigations in vitro and in vivo.

Declarations Funding
No funding was received for conducting this study.

Con ict of interest
The authors have no con icts of interest to declare that are relevant to the content of this article.
Availability of data and material All authors con rm that all data and materials as well as software application or custom code support their published claims and comply with eld standards.      The intersection of the merged motifs between the HCoV-OC43 and SARS-CoV-2 treated cells upregulated DEGs networks Drug-gene interactions of the most crucial shared up-regulated genes (a) and down-regulated genes (b) of the intersection of merged motifs in the two GRN networks, involved in neurological processes