In Silico Investigation of Potential COVID-19-Associated MicroRNA Signatures

Background: The global pandemic COVID-19, caused by the coronavirus SARS-CoV-2, is persistent despite the increasing vaccination rates, with new cases being reported per week. MicroRNAs, non-coding RNA species that regulate gene expression at the post-transcriptional level, play a pivotal role in the SARS-CoV-2 life cycle, pathophysiology and host’s anticoronaviral responses. Materials and methods: In the present study, an integrative bioinformatics approach was employed – including database searching, gene set enrichment analysis, and network-based and microRNA target prediction methods – towards discovering epigenetic determinants of COVID-19. Results: An intricate microRNA-target gene network was constructed, and a set of eight highly interacting microRNAs that potentially co-target and co-regulate key COVID-19-related genes was detected. These miRNAs and their corresponding genes are likely involved in the response to SARS-CoV-2 infection. Conclusion: The eight functionally associated miRNAs could represent the components of a signature for COVID-19 diagnosis.

It has been demonstrated that miRNAs play a critical role in coronaviral life cycle, pathogenesis and host antiviral responses (Abedi et al., 2021;Arghiani et al., 2021).SARS-CoV-2 infection-induced changes in the expression patterns of the host miRNAs can lead to the down-regulation of key genes involved in the immune response and, consequently, to immunosuppression or attenuated host immune surveillance, thereby enabling the coronavirus to evade the host's immune system (Farr et al., 2021;Liang et al., 2023;Yang et al., 2022).SARS-CoV-2 can also co-opt host miRNAs implicated in immune response suppression (e.g., hsa-miR-939 and hsa-miR-146b), in order to facilitate coronaviral replication and subvert host immune responses (Arghiani et al., 2021;Panda et al., 2022).Moreover, miRNAs encoded by the coronaviral genome can target genes involved in host immune response/inflammation, like IFN1-mediated signaling (Khan et al., 2020;Singh et al., 2022).SARS-CoV-2 non-coding RNAs, by acting as miRNA sponges, can also deplete the host's miRNA pool (Bartoszewski et al., 2020;Li et al., 2022).
In this study, we have employed a bioinformatics approach towards the investigation of potential miRNA determinants of the epigenetic regulation of genes that are prominently associated with COVID-19, in order to decipher the co-regulation activities of miRNAs exerted upon those genes in coronavirus infection.

Protein-Protein Interaction Network
The physical and functional associations among the protein products of the retrieved COVID-19-related genes were investigated and visualized using STRING (Search Tool for Retrieval of Interacting Genes/Proteins) v11.5 (Szklarczyk et al., 2021), a database of both experimentally supported or predicted, functional and/or physical, association data among genes/proteins derived collected from diverse resources.A relatively high confidence interaction score (≥0.7) was set as a cutoff, and only those associations based on text mining, database and experimental evidence were considered, in order to enhance the reliability of the given interactions.The associations were further visualized and analyzed through the open-source platform Cytoscape (v.3.10.0)(Shannon et al., 2003).Moreover, the Cytoscape plugin cytoHubba (Chin et al., 2014) using the node-degree filter.

Functional Enrichment Analysis
Gene set enrichment analysis was conducted with the online tool WebGestalt (WEB-based GEne SeT AnaLysis Toolkit) (Kirov et al., 2014;Liao et al., 2019), for the identification of statistically significant over-represented terms in the COVID-19 genes under study.The WebGestalt parameters selected were "Organism of Interest": Homo sapiens, "Method of Interest": Over-Representation Analysis (ORA), "Functional database": pathway/Reactome for biological paths, "Select gene ID type": gene symbol, "Select Reference set": genome; the default advanced parameters were chosen, and only those pathways with a Benjamini-Hochberg-adjusted p-value (Benjamini and Hochberg, 1995) ≤ 0.05 were included in the analysis.Affinity propagation was applied to reduce the terms and cluster them into representative categories.

miRNA Regulators of COVID-19 Genes
The miRNAs potentially regulating the COVID-19 genes were investigated.(Chen and Wang, 2020;Liu and Wang, 2019).The miRNA-mRNA interactions were retrieved from each tool combined, and the duplicates were removed.To enhance the accuracy of the prediction, only those miRNA-gene targets predicted by more than three methods were considered in this study.

Pairwise miRNA associations
The miRNA-miRNA relationships were collected from two different sources, MiRGOFS (Y.Yang et al., 2018) and GOSemSim (Yu et al., 2010), which infer functional similarities between miRNA pairs based on the degree of functional relatedness of their corresponding target genes.The pairwise miRNA interactions from both sources were merged, and the duplicates were removed.Only those miRNA-miRNA interactions with a weight score above 0.85 (where "1.0" is the highest score) were chosen, so as to enhance robustness.

Independent Validation
The findings of this study were further compared against an independent transcriptomic dataset available in the comprehensive online resource COVID19db (Zhang et al., 2022), by using the "differential expression" module, which provides gene expression profiling in whole blood derived from COVID-19 patients and healthy controls.

Results and Discussion
Collectively, 149 COVID-19-relevant genes were obtained from GeneCards (Table S1).A functional network of the products of those genes was generated, and 128 nodes appear to be highly interconnected (Figure 1), suggesting physical and functional associations among the corresponding proteins.By examining the topological properties of the network, we identified those key nodes that are more relevant to the overall function of the network and, therefore, biologically meaningful (Barabasi et al., 2011;Kontou et al., 2016).The 25 top nodes were detected in the protein-protein interaction network (Figure 1 and Table S1) based on the combined output of the twelve algorithms in cytoHubba.The terms over-represented in the genes, the products of which are involved in the protein association network (Figure 1), are mainly associated with immune signaling pathways (Figure 2), e.g., cytokine, interleukin, interferon, Toll-like receptor, MYD88, NF-kF, T-cell receptor (TCR)-mediated signaling cascades.Notably, 109 (out of 128) and all 25 top-ranking genes are found in the over-represented Reactome pathways, further highlighting their prominent role in COVID-19 pathophysiology.Among the 25 key genes, there are mostly genes encoding components of the immune system (CD4, CD8A, HLA-B, IRF3, MYD88 etc.), including pro-and anti-inflammatory factors, such as chemokines and cytokines (CXCL8, CXCL10, IL1B, IL2, IL4, IL6, IL10, IL17A, IL18, IFNG) and the tumor necrosis factor TNF.In many COVID-19 cases, exacerbated inflammatory responses ("cytokine storm") are observed, which result from the acute increase in the levels of circulating pro-inflammatory cytokines and chemokines, and their uncontrolled release, both at local and systemic levels (Coperchini et al., 2020;Hu et al., 2021;Montazersaheb et al., 2022).Several studies, though, suggest that the human host's immune system is rather compromised upon SARS-CoV-2 infection and is not capable of eliciting a sufficient immune response  (Ozbek et al., 2022;Remy et al., 2020).
The NF-kB subunits, NFKB1 and RELA (Table S1), are tightly connected to several pro-inflammatory agents.NF-kB can modulate the host's immediate innate immune response to SARS-CoV-2 infection.In particular, SARS-CoV-2-mediated activation of NF-kB was found to induce the expression of the genes IL1, IL2, IL6, IL8 and TNF (Hariharan et al., 2021).
Increased expression of TLR4 greatly affects heart failure in COVID-19 patients (Choudhury and Mukherjee, 2021;Mukherjee, 2022).TLR4 is known to trigger the activation of pro-inflammatory cytokines (Swanson et al., 2020), including TNF, which is also up-regulated in COVID-19 (Figure 4); increased expression of TNF was found to be a prognostic factor for mortality among COVID-19 patients with comorbidities and disease progression (Mohd Zawawi et al., 2023).MYD88, which plays a central role in TLR/IL1R-mediated signalling in innate and adaptive immunity (Chen et al., 2020), is overexpressed in the COVID-19 group (Figure 4); MYD88 polymorphisms were shown to be tightly associated with COVID19 severity (Martinez-Gomez et al., 2023).Finally, the expression level of the interferon-inducible gene OAS1 is markedly higher in the SARS-CoV-2 infected patients as compared to the healthy samples, in agreement with the findings of a recent study, wherein SARS-CoV-2 infection significantly increased the expression of OAS1 (Assou et al., 2023).
The causal genes of diseases/disorders usually share common regulatory mechanisms in order to ensure their coordinated regulation.Disease-related miRNAs act in a cooperative manner so as to exert their regulatory effect upon their corresponding target genes (Arshinchi Bonab et al., 2022;Zinani et al., 2022).
The protein-protein interaction and miRNA-gene networks constructed in this study provide a fundamental framework for detecting protein-coding genes and epigenetic regulators (e.g., miRNAs) that likely respond to SARS-CoV-2 infection in a coordinated way.In this study, by applying stringent criteria, we discovered eight highly interacting miRNAs that potentially co-target key COVID-19 genes.This panel of miRNAs could represent potential signature components for COVID-19.

Conclusion
Herein, by employing a bioinformatics pipeline, we detected eight functionally related miRNAs that likely co-regulate pivotal COVID-19-associated genes.The differential expression status of the predicted miRNAs, and their corresponding pairwise interactions, merit further experimental validation in the context of COVID-19.These miRNAs could be taken into consideration in the clinical setting for updating and complementing currently used biomarkers towards improving the accuracy of diagnosing SARS-CoV-2-infected patients.

Figure 1 .
Figure 1.Interaction network of the COVID-19 proteins.The nodes denote genes/gene products, and the edges represent functional associations.The nodes corresponding to the top-ranking gene products are indicated by light red fill colors.

Figure 2 .
Figure 2. Bar plot illustrating the over-represented Reactome pathways in the COVID-19 genes.The width of bar plots is proportional to the number of genes in each pathway.

Figure 3 .
Figure 3. MiRNA-target gene network in COVID-19.The miRNAs are represented by polygons, and the miRNA target genes are denoted by circles.