Abstract

   A detailed understanding of the molecular mechanism of SARS-CoV-2
   pathogenesis is still elusive, and there is a need to address its
   deadly nature and to design effective therapeutics. Here, we present a
   study that elucidates the interplay between the SARS-CoV and SARS-CoV-2
   viruses' and host's miRNAs, an epigenetic regulator, as a mode of
   pathogenesis; and we explored how the SARS-CoV and SARS-CoV-2
   infections differ in terms of their miRNA-mediated interactions with
   the host and the implications this has in terms of disease complexity.
   We have utilized computational approaches to predict potential host and
   viral miRNAs and their possible roles in different important functional
   pathways. We have identified several putative host antiviral miRNAs
   that can target the SARS viruses and also predicted SARS
   viruses-encoded miRNAs targeting host genes. In silico predicted
   targets were also integrated with SARS-infected human cell microarray
   and RNA-seq gene expression data. A comparison between the host miRNA
   binding profiles on 67 different SARS-CoV-2 genomes from 24 different
   countries with respective country's normalized death count surprisingly
   uncovered some miRNA clusters, which are associated with increased
   death rates. We have found that induced cellular miRNAs can be both a
   boon and a bane to the host immunity, as they have possible roles in
   neutralizing the viral threat; conversely, they can also function as
   proviral factors. On the other hand, from over representation analysis,
   our study revealed that although both SARS-CoV and SARS-CoV-2 viral
   miRNAs could target broad immune-signaling pathways; only some of the
   SARS-CoV-2 miRNAs are found to uniquely target some immune-signaling
   pathways, such as autophagy, IFN-I signaling, etc., which might suggest
   their immune-escape mechanisms for prolonged latency inside some hosts
   without any symptoms of COVID-19. Furthermore, SARS-CoV-2 can modulate
   several important cellular pathways that might lead to the increased
   anomalies in patients with comorbidities like cardiovascular diseases,
   diabetes, breathing complications, etc. This might suggest that miRNAs
   can be a key epigenetic modulator behind the overcomplications amongst
   the COVID-19 patients. Our results support that miRNAs of host and
   SARS-CoV-2 can indeed play a role in the pathogenesis which can be
   further concluded with more experiments. These results will also be
   useful in designing RNA therapeutics to alleviate the complications
   from COVID-19.

   Keywords: SARS-CoV-2, COVID-19, miRNA-microRNA, viral pathogenesis,
   immune regulation

Introduction

   Coronavirus outbreaks have been reported over the past three decades,
   but the recent SARS-CoV-2 pandemic has reached more than 200 countries,
   has been the causative agent for the death of 58,392 people around the
   globe, and 1,087,374 coronavirus cases have been filed as of the date
   of writing this article (Worldometer, [29]2020). Among closed cases of
   SARS-CoV-2, 20% of the patients have died, and 5% of patients within
   active cases are in critical situations (Worldometer, [30]2020). The
   initial estimation of the SARS-CoV-2 death rate of 3.4%, as declared by
   the WHO ([31]2020), requires refreshing, as the global casualty is on
   the rise. This novel virus requires novel and in-depth studies to
   promote new strategies for the management of this pandemic.

   The coronavirus subfamily is a single-stranded positive-sense (+ssRNA)
   virus with a genome size of around 30 kb (Lu et al., [32]2020). The
   family is categorized into four subgenera as alpha, beta, gamma, and
   delta coronavirus (Cheng and Shan, [33]2020). SARS-CoV-2 is a beta
   coronavirus with a genome size of 29.9 kb (Accession no.
   [34]NC_045512.2), 11 genes being reported in NCBI-Gene ([35]2020).
   Phylogenetic analysis between SARS-CoV-2 and SARS-CoV showed ~79%
   similarity. Meanwhile, the distance is much greater for MERS-CoV (~50%
   similarity), but the closest relative to the SARS-CoV-2 is bat-derived
   SARS-like coronavirus (~90% similarity) (Jiang et al., [36]2020; Lu et
   al., [37]2020; Ren et al., [38]2020). Genomic analysis of SARS-CoV and
   SARS-CoV-2 has shown substitution of 380 amino acids and deletion of
   ORF8a, elongation of ORF8b (84 vs. 121 amino acid residues) and
   truncation of ORF3b (154aa in SARS-CoV whereas 22aa in SARS-CoV-2) (Lu
   et al., [39]2020).

   MicroRNAs are small ncRNA molecules that regulate
   post-transcriptional-level gene expression; it has already been
   established that viruses use host machinery to produce miRNAs (Ambros,
   [40]2001). Although miRNA can be an important antiviral tool (Trobaugh
   and Klimstra, [41]2017) that can stimulate the innate and adaptive
   immune system (Ambros, [42]2001; Trobaugh and Klimstra, [43]2017) but
   that can also be a back door for viral propagation; it is
   non-antigenic, thereby modulating cellular pathways without triggering
   the host immune response (Cullen, [44]2013; Głobinska et al.,
   [45]2014). For example, nucleocapsid protein of coronavirus OC43 binds
   miR-9 and activates NF-κB (Lai et al., [46]2014). Although host
   microRNAs are either utilized or regulated by viruses, viral miRNAs are
   another side of the coin; they regulate host gene expression, cellular
   proliferation, stress-related genes, and even viral gene expression
   (Cullen, [47]2010; Haasnoot and Berkhout, [48]2011; Lai et al.,
   [49]2014). A summary explored how a number of DNA and RNA viruses
   produce miRNAs known as viral miRNAs (v-miRNAs) to evade the host
   immune response (Mishra et al., [50]2020). Novel viral miRNAs have been
   predicted to play an important role in neurological disorders as well
   (Islam et al., [51]2019). Among RNA viruses, HIV-1-encoded miR-H1, for
   example, can cause mononuclear cells apoptosis; H5N1 influenza
   virus-encoded miR-HA-3p targets host PCBP2 and contributes to the
   “cytokine storm” and mortality; and KUN-miR-1 of the West Nile virus
   targets the host's GATA4, which facilitates virus replication (Li and
   Zou, [52]2019). The interactions of host miRNAs with the SARS-CoV
   genome and viral proteins have been elucidated to suppress viral growth
   and immune evasion (Mallick et al., [53]2009). Novel classes of ncRNAs
   have been also observed by studies as possibly playing a definitive
   role in pathogenesis and survival (Liu et al., [54]2018). Respiratory
   viral infections caused by influenza, rhinovirus, adenovirus, RSV, and
   coronaviruses can be related to aberrant host miRNA expression, and
   their effect on the host can results in cell apoptosis, inhibition of
   immunologic pathways, downregulation of host antiviral responses, etc.
   (Mallick et al., [55]2009; Bondanese et al., [56]2014; Islam et al.,
   [57]2019; Li and Zou, [58]2019; Mishra et al., [59]2020). Transmissible
   gastroenteritis virus (TGEV), though it induces significant IFN-I
   production after infection by inducing endoplasmic reticulum (ER), can
   evade antiviral effect of IFN-I by downregulating miR-30a-5p, which
   normally enhances IFN-I antiviral activity (Ma et al., [60]2018).

   On the other hand, host miRNA expression plays a major role in
   controlling viral pathogenesis by mediating T cells and antiviral
   effector functions (Dickey et al., [61]2016). The first-reported
   example of a cellular miRNA that targets a viral RNA genome is miR-32,
   which targets the retrovirus PFV-1 transcripts and results in reduced
   virus replication (Lecellier et al., [62]2005). Similarly, miR-24,
   miR-93 can target VSV virus L and P proteins (Otsuka et al., [63]2007);
   miR-29a targets HIV Nef proteins (Ahluwalia et al., [64]2008) to
   inhibit replication; and miR-1, miR-30, miR-128, miR-196, miR-296,
   miR-351, miR-431, and miR-448 targets HCV C and NS5A proteins to
   inhibit translation/replication by inducing IFN signaling (Pedersen et
   al., [65]2007). miRNAs can therefore provide a different perspective in
   explaining the pathogenesis and infectivity of the novel SARS-CoV-2.
   Although SARS-CoV is distantly related to SARS-CoV-2, there are some
   similarities in their signs and symptoms, and they might be similar in
   pathogenesis, but there are crucial differences between the two
   diseases too (Xu et al., [66]2020). SARS-CoV-2 has infected many
   countries, and this has resulted in a stable mutation rate and some
   degree of variation (Xu et al., [67]2020). There is evidence that viral
   pathogens can have a novel immune evasion role by utilizing host miRNAs
   (Islam et al., [68]2019; Mishra et al., [69]2020).

   The detailed miRNA-mediated epigenetic interplay between SARS-CoV-2 and
   its host is yet to be elucidated. It is not known what probable miRNAs
   produced by SARS-CoV-2 are affecting which human processes.
   Additionally, we do not know which antiviral miRNAs are taking part in
   host immunity. The genomic difference that, as a result, controls the
   host miRNA target sites and viral miRNAs might explain the difference
   between SARS-CoV and various isolated of SARS-CoV-2 in terms of
   pathogenesis and infectivity. Here in this study, we hypothesize on
   three potential effects of host and viral miRNA: (1) genomic
   differences between SARS-CoV and SARS-CoV-2 can lead to variations in
   host miRNA binding and differences in hence pathogenicity, signs and
   symptoms of these diseases and might explain the relatively longer
   incubation period of SARS-CoV-2; (2) on the other hand, there might be
   differences in viral miRNAs that can regulate expressions of different
   sets of host genes, which can in turn be advantageous to the virus or
   the host; and (3), due to a rapid mutation rate, observed variations
   among SARS-CoV-2 isolates in different regions of the world might
   result in variation in host capacities to target the virus with its
   miRNAs. This, in turn, might play a significant role in varying degrees
   of disease severity, symptoms, and mortality rate in different regions.
   In this study, we have carried out comparative analyses between
   SARS-CoV and SARS-CoV-2 with respect to host miRNA–viral genome
   interaction as well their differences based on region-specific isolates
   of SARS-CoV-2 and viral miRNA–host mRNA interactions to delineate the
   exclusive features of COVID-19 and their roles in viral survival and
   pathogenicity with respect to SARS-CoV ([70]Figure 1).

Figure 1.

   [71]Figure 1
   [72]Open in a new tab

   Overall workflow of the whole study.

Materials and Methods

Obtaining SARS-CoV and SARS-CoV-2 Genome Sequences

   The reference genomes of SARS-CoV (RefSeq Accession no.
   [73]NC_004718.3) and SARS-CoV-2 (RefSeq Accession no. [74]NC_045512.2)
   were retrieved from the NCBI RefSeq database (NCBI-RefSeq, [75]2020). A
   total of 67 whole-genome sequences of SARS-CoV-2 isolates covering 24
   different countries ([76]Supplementary File 1) were retrieved from
   NCBI-Virus ([77]2020) and Shu and McCauley ([78]2017).

Obtaining Human 3′UTR and Mature miRNA Sequences

   Human miRNAs were accessed from microRNA database miRBase (Kozomara et
   al., [79]2018), and 3′UTR sequences of human protein-coding genes were
   obtained from Ensembl-Biomart (Hunt et al., [80]2018) (release 99).

Prediction of Viral Pre-miRNA and Validation

   We used miRNAFold (Tav et al., [81]2016) for de novo prediction of all
   possible precursor-miRNAs from the obtained reference sequences of
   SARS-CoV and SARS-CoV-2 with a sliding window size of 150 and minimum
   hairpin size as 0. The results were validated using three different
   tools. First, RNAfold (Gruber et al., [82]2008) was used with minimum
   free energy (MFE) and partition function fold algorithm to find stable
   secondary structures. Second, a fixed-order Markov model-based
   algorithm namely FOMmiR (Shen et al., [83]2012) was used. Finally, a
   SVM-based tool iMiRNA-SSF (Chen et al., [84]2016) was used that
   calculates minimum free energy (MFE), p-value of randomization test
   (P-value), and the local triplet sequence-structure features. The
   common predictions from these three tools were utilized for further
   analysis.

Prediction of Mature miRNA

   A Naive Bayes classifier algorithm implemented in tool MatureBayes
   (Gkirtzou et al., [85]2010) was used to identify mature miRNA
   candidates within the miRNA precursor sequences.

RNA-RNA Interaction Analysis

   Three different tools were used to analyze RNA–RNA interactions for the
   host miRNA-viral genome and viral miRNA-host 3′UTR of coding sequences.
   IntaRNA 2.0 (Mann et al., [86]2017) was used considering sites with
   parameters –mode=H, –model=X, –outMode=C, and ΔΔG ≤ −10 kcal/mol, with
   seed 2–8, allowing for G:U base pairs. microRNA.org (Betel et al.,
   [87]2008) was used with a score cutoff ≥ 140, energy cutoff ≤ −20
   kcal/mol, gap opening = −9.0, and gap extension = −4.0; psRNATarget
   (Dai and Zhao, [88]2011) with the default parameter was also used to
   determine RNA-RNA interactions. Finally, the common predictions from
   these three tools were considered for downstream analysis.

Extraction of Targets of Host miRNAs

   Targets of host miRNAs were mainly obtained using Funrich software
   (Pathan et al., [89]2017), which curates the experimentally validated
   targets of host miRNAs from different databases. The targets were also
   cross checked with the experimentally validated targets from miRTarBase
   database (Huang et al., [90]2019).

Target Genes Functional Enrichment Analysis

Enrichment Analysis in Gitools

   The functional annotation of target genes is based on Gene Ontology
   (GO) (Ashburner et al., [91]2000), as extracted from the EnsEMBL
   (Hubbard et al., [92]2007) and KEGG pathway database (Kanehisa and
   Goto, [93]2000). Accordingly, all genes are classified into the
   ontology categories' biological process (GOBP) and pathways when
   possible. We have taken only the GO/pathway categories that have at
   least 10 genes annotated. We used Gitools for enrichment analysis and
   heatmap generation (Perez-Llamas and Lopez-Bigas, [94]2011). Resulting
   p-values were adjusted for multiple testing using the Benjamini and
   Hochberg's method of False Discovery Rate (FDR) (Benjamini and
   Hochberg, [95]1995).

Enrichment Analysis Using Web-Based Tools

   The host miRNAs targeting SARS-CoV and SARS-CoV-2 were used for
   functional over-representation analysis to visualize and predict the
   roles of these miRNAs in human diseases and find enriched pathways.
   Besides Gitools, functional enrichment analyses of the target human
   genes were conducted using EnrichR (Kuleshov et al., [96]2016), DAVID
   6.8 (Huang et al., [97]2009; Sherman and Lempicki, [98]2009),
   WebGestalt 2019 (Liao et al., [99]2019), and FunRich 3.1.3 (Pathan et
   al., [100]2017). The targeted genes are analyzed to determine their
   role in viral pathogenesis, infectivity, and immune evasion.

Microarray Expression Data Analysis

   Microarray data for changes in gene expression induced by SARS-CoV on
   2B4 cells, comparing the infected cells with SARS-CoV with the
   uninfected cells (for 12, 24, and 48 h), obtained from the Gene
   Expression Omnibus (GEO), ID [101]GSE17400
   ([102]https://www.ncbi.nlm.nih.gov/geo) (Barrett et al., [103]2012).
   Raw Affymatrix CEL files were background corrected and normalized using
   the “rma” algorithm of Bioconductor package “affy” (version 1.28.1).
   The quality of microarray experiment (data not shown) was verified by
   Bioconductor package “arrayQualityMetrics” (Kauffmann et al.,
   [104]2009) (version 3.2.4 under Bioconductor version 3.10; R version
   3.6.0). To determine genes that are differentially expressed (DE)
   between two experimental conditions, Bioconductor package Limma (Smyth,
   [105]2005) was utilized to generate contrast matrices and fit the
   corresponding linear model. Probe annotations of genes were done using
   the Ensembl gene model (Ensembl version 99) as extracted from Biomart
   (Flicek et al., [106]2007) and by using in-house python script. When
   more than one probe was annotated to the same gene, the highest
   absolute expression value was considered (maximizing). To consider a
   gene is differentially expressed, multiple tests corrected FDR
   (Benjamini and Hochberg, [107]1995) p-value ≤ 0.05 was used as a
   cut-off.

RNA-seq Expression Data Analysis

   RNA-seq raw read-count data on SARS-CoV-2 mediated expression changes
   in primary human lung epithelium (NHBE), and transformed lung alveolar
   (A549) cells were obtained from the GEO database ([108]GSE147507)
   (Barrett et al., [109]2012). For the differential expression (DE)
   analysis, we used the Bioconductor package DESeq2 (version 1.38.0)
   (Anders and Huber, [110]2010) with R version 3.6.0 (R Core Team,
   [111]2016) with a model based on the negative binomial distribution. To
   avoid false positives, we considered only those transcripts where at
   least 10 reads were annotated and a p-value of 0.01.

MicroRNA Clustering

   The hierarchal clustering of human miRNAs that could target SARS-CoV-2
   genomes (binary mode) obtained from various countries was done using
   Manhattan distance and complete linkage analysis with the Genesis tool
   (Sturn et al., [112]2002). The human death count (per million
   population) due to SARS-CoV-2 infection was obtained on the 2nd of
   April, 2020, from the “Worldometer” website (Worldometer, [113]2020).

Overlap Analysis

   Two- or three-way overlap analysis was done using online venn-diagram
   program Venny 2.1.0 (Oliveros, [114]2018). Multiple pairwise overlaps,
   correlation analyses, and heatmap generations were carried out using
   Gitools (Perez-Llamas and Lopez-Bigas, [115]2011).

Data Visualization

   We have visualized human miRNAs that bind to the virus genome in
   web-genome browser NCBI genome data viewer (NCBI's-Genome-Browser,
   [116]2020).

Results

Several Human miRNAs Are Found to Target SARS-CoV and SARS-CoV-2

   It is possible that, during viral infections, host-encoded miRNAs can
   modulate viral infections as a means of a host immune response (Girardi
   et al., [117]2018). To identify possible host miRNAs that can get
   induced during the SARS-CoV (R) and SARS-CoV-2 (R) infections, we have
   utilized a bioinformatics approach. From our rigorous analysis
   pipeline, which covers three different well-established algorithms
   (IntaRNA, miRanda, and psRNATarget) to predict RNA-RNA interactions, we
   have identified 122 and 106 host antiviral miRNAs against SARS-CoV (R)
   and SARS-CoV-2 (R), respectively ([118]Figures 2A,B, [119]Supplementary
   File 2). Amongst these, 27 miRNAs were found to be targeting both
   viruses ([120]Figure 2A). Whilst comparing these miRNAs with the
   antiviral miRNAs from VIRmiRNA (Qureshi et al., [121]2014), we have
   found four (hsa-miR-654-5p, hsa-miR-198, hsa-miR-622, and
   hsa-miR-323a-5p) and three (hsa-miR-17-5p, hsa-miR-20b-5p, and
   hsa-miR-323a-5p) host miRNAs against SARS-CoV (R) and SARS-CoV-2 (R),
   respectively, to exhibit experimental evidence of having antiviral
   roles during infections ([122]Figures 2A–C).

Figure 2.

   [123]Figure 2
   [124]Open in a new tab

   The host miRNAs targeting SARS-CoV and SARS-CoV-2. (A) Common host
   miRNAs and their target genes in SARS-CoV and SARS-CoV-2. (B) Host
   miRNAs and their target genes which uniquely target either SARS-CoV or
   SARS-CoV-2. (C) Venn diagram showing the common and unique host miRNAs
   targeting SARS-CoV and SARS-CoV-2 and host miRNAs that have exhibited
   experimental evidence as antiviral miRNAs.

   Moreover, we compared the miRNAs targeting the two reference genomes of
   SARS-CoV (R) and SARS-CoV-2 (R), and we found most of the host miRNAs
   can target the ORF1ab region, followed by the S region as the
   second-most targeted ([125]Figures 3A,B). Additionally, the M, N,
   ORF3a, ORF7a, ORF8 (ORF8a, ORF8b for SARS-CoV), 5′ UTR, and 3′ UTR
   regions of both viruses were targeted by host miRNAs. The significant
   variance was observed in the targeting positions of the host miRNAs
   between these two viruses ([126]Figures 3A,B).

Figure 3.

   [127]Figure 3
   [128]Open in a new tab

   Genome browser view of host miRNAs targeting the regions of (A)
   SARS-CoV (Reference) and (B) SARS-CoV-2 (Reference) genomes.

   Since the RNA virus mutates fast, it is conceivable that mutations in
   crucial genomic locations would lead to differences in host miRNA
   binding patterns. Therefore, the ability of the host miRNAs in
   targeting genomes of 67 SARS-CoV-2 isolates covering 24 different
   countries was also seen. Although, as expected, most of the identified
   host miRNAs' binding profiles across these isolates remained somewhat
   similar to that of SARS-CoV-2 reference sequence; interestingly, we
   have identified 24 host miRNAs that bind differentially across the
   isolates ([129]Figure 4A), which might have occurred due to the genomic
   variations between these isolates. Complete linkage agglomerative
   hierarchal cluster (HCL) analysis with Manhattan distance of these
   miRNAs (binary mode, bind or not bind) revealed two major clusters with
   a side cluster for one South Korean and two Singaporean isolates
   ([130]Figure 4B). As miRNA is crucial in both host defense and viral
   pathogenesis, to understand the significance of this cluster, we have
   also compared the host miRNA clusters with the death rate (normalized
   by per million population) from different countries. Surprisingly,
   relatively higher deaths are found to be more prominent in the European
   major clusters (right side cluster) compared to the other major cluster
   (left side), and we also found much lower deaths in side clusters
   ([131]Figure 4B). However, many more factors are there which can also
   play pivotal roles in the susceptibility of the patients, this suggests
   only one interesting aspect for this correlation between the miRNA
   binding pattern and host susceptibility.

Figure 4.

   [132]Figure 4
   [133]Open in a new tab

   Differences of host miRNA-binding profiles, (A) representing only
   uncommon miRNAs binding pattern in 67 different SARS-CoV-2 genomes from
   24 different countries, and (B) hierarchal clustering of all miRNAs
   binding in 67 genomes (upper panel, same country with same color code)
   and association of country-specific death rates (in color-coded scale)
   per million population (lower panel).

Host miRNAs Targeting SARS-CoV and SARS-CoV-2 Play Crucial Roles in
Neutralizing the Virus

   Though the primary action elicited by host miRNAs is to silence the
   viral RNA, they might also modulate some host factors which provide an
   edge to the viral pathogenesis. To find out if these particular
   pathways are also targeted by the host miRNAs induced by SARS-CoV and
   SARS-CoV-2 infections, we have performed miRNA pathway enrichment
   analysis. We have found that several such pathways might be deregulated
   by the host miRNAs to suppress the entry of the virus, prevent the
   spread of the virions, and to minimize the systemic symptoms resulting
   from the infection ([134]Figure 5A).

Figure 5.

   [135]Figure 5
   [136]Open in a new tab

   Enrichment analysis and comparison between host miRNA targets induced
   by SARS-CoV and SARS-CoV-2 infections. (A) Heatmap representation of
   enriched pathways involved in host defense obtained using Funrich
   software. (B) Enriched pathways that might act as proviral mechanisms
   obtained using Funrich software. Enrichment of downregulated host miRNA
   target genes in SARS-CoV and SARS-CoV-2 using gitools. (C) GO
   Biological Process module. (D) GO Molecular Function module. (E) KEGG
   pathway modules. The significance of enrichment in terms of adjusted
   p-value (<0.05) is represented in a color-coded P-value scale for all
   heatmaps. Color toward red indicates higher significance and color
   toward yellow indicates less significance, while gray means
   non-significant. Only selected significant enriched terms are shown.

   Host miRNAs might have a probable role in blocking the entry of the
   virus, as they are found to be targeting the pathways needed for viral
   entry- PDGF receptor-like signaling (Soroceanu et al., [137]2008),
   Arf-6 signaling (García-Expósito et al., [138]2011), PI3K-Akt signaling
   (Diehl and Schaal, [139]2013), EGFR signaling (Zheng et al.,
   [140]2014), signaling events mediated by focal adhesion kinase
   (Elbahesh et al., [141]2014), CDC42 signaling (Swaine and Dittmar,
   [142]2015), the EphrinB-EPHB pathway (Wang et al., [143]2019), Cadherin
   signaling (Mateo et al., [144]2015), RTK signaling (Haqshenas and
   Doerig, [145]2019), etc. ([146]Figure 5A).

   They can also block some machinery like-p38 MAPK signaling (Hirasawa et
   al., [147]2003), FAK signaling (Elbahesh et al., [148]2014), PI3K-Akt
   signaling (Diehl and Schaal, [149]2013), etc., which can be hijacked by
   viruses for their efficient replication, pre-mRNA processing, and
   translation ([150]Figure 5A). These host miRNAs might also try to
   reduce some host-induced inflammatory responses to prevent acute lung
   damage by targeting IGF1 signaling (Li et al., [151]2019), VEGF
   signaling (Alkharsah, [152]2018), PAR1 signaling (Heuberger and
   Schuepbach, [153]2019), integrin signaling (Teoh et al., [154]2015),
   TGF-beta signaling (Denney et al., [155]2018), TRAIL signaling (Cummins
   and Badley, [156]2009), etc. ([157]Figure 5A). Some signaling pathways,
   such as CXCR4 signaling (Arnolds and Spencer, [158]2014), TGF-beta
   signaling (Denney et al., [159]2018), mTOR signaling (Le Sage et al.,
   [160]2016), PI3K-Akt signaling (Diehl and Schaal, [161]2013), etc., can
   facilitate viral survival in infected cells by inhibiting apoptosis,
   autophagy, early immune responses, etc. Host miRNAs may function to
   downregulate these to invoke a proper immune response against the
   viruses ([162]Figure 5A).

Infection Induced Host miRNAs Can Function as a Proviral Factor by Inhibiting
Host Immune Surveillance Pathways

   Host miRNAs can be like double-edged swords, as sometimes, they can
   facilitate viral immune evasion by targeting some important host immune
   responses (Bruscella et al., [163]2017). Our host miRNA enrichment
   analysis showed several significant pathways, such as IFN-gamma
   signaling (Kang et al., [164]2018), TGF-beta signaling (Mogensen and
   Paludan, [165]2001), Interleukin signaling (Kimura et al., [166]2013),
   IGF1 signaling (Li et al., [167]2019), TRAIL signaling (Cummins and
   Badley, [168]2009), etc. These are involved in important
   proinflammatory cytokine signaling during viral infections ([169]Figure
   5B). Interestingly, we have found out that host miRNAs induced during
   SARS-CoV-2 infection may particularly downregulate the signaling of
   different Toll-Like Receptors (TLRs) (Kimura et al., [170]2013), which
   are considered as the primary stimulatory molecules for producing host
   antiviral responses (i.e., production of interferons and other
   inflammatory cytokines) ([171]Figure 5B). Also, other receptor
   signaling involved in antiviral responses like- uPA-UPAR signaling
   (Alfano et al., [172]2003), TRAF6 signaling (Konno et al., [173]2009),
   S1P1 signaling (Oldstone et al., [174]2013), Estrogen receptor
   signaling (Kovats, [175]2015), Protease-activated Receptor (PAR)
   signaling (Antoniak et al., [176]2013), Bone morphogenetic protein
   (BMP) signaling (Eddowes et al., [177]2019), etc. can also be
   deregulated by the host miRNAs, leading to the host's immune
   suppression ([178]Figure 5B).

Host miRNAs' Targeted Downregulated Pathways Are Related to the Comorbidities
of COVID-19

   SARS-CoV-2-infected patients with comorbidities (i.e., cardiovascular
   diseases, diabetes, and renal problems) are found to be more
   susceptible to COVID-19. To find out whether host miRNAs play a role in
   these, we performed enrichment analyses using the downregulated targets
   genes of the host miRNAs using the expression data obtained from GEO
   dataset ([179]GSE17400 for SARS-CoV and [180]GSE147507 for SARS-CoV-2).
   These revealed that the downregulated targets of host miRNAs are
   involved in functions and pathways, such as heart development, kidney
   development, several neuronal processes, metabolic process, regulation
   of cellular ketone metabolism, insulin resistance, glucagon signaling
   pathway, fatty acid metabolism, and PPAR signaling ([181]Figures 5C–E).
   Aberrant regulation of these processes can overcomplicate the disease
   conditions of patients having existing disorders.

Viral miRNAs Encoded by SARS-CoV and SARS-CoV-2 Can Target Several Host Genes

   Many human viruses were found to produce miRNAs to assist in their
   overall pathogenesis by modulating host factors (Bruscella et al.,
   [182]2017). Previous study on SARS-CoV also suggests that viral small
   non-coding RNAs can help its efficient pathogenesis (Morales et al.,
   [183]2017). Our bioinformatics approach suggests that SARS-CoV and
   SARS-CoV-2 can also encode some viral miRNAs. The miRNAfold tool (Tav
   et al., [184]2016) yielded 529 and 519 putative pre-miRNAs from the
   genome of SARS-CoV and SARS-CoV-2, respectively. The RNAfold tool
   (Gruber et al., [185]2008) predicted 303 and 308 of these precursors of
   SARS-CoV and SARS-CoV-2, respectively are highly stable for forming
   hairpin structure which is a prerequisite of mature miRNA formation.
   Using FomMiR (Shen et al., [186]2012) and IMiRNA-SSF (Chen et al.,
   [187]2016), we then predicted which of these highly stable precursors
   can truly produce mature miRNAs. We have found 63 and 85 such
   precursors, respectively for SARS-CoV and SARS-CoV-2. Using the
   Maturebayes tool from these precursors, we identified 126 and 170
   mature miRNAs from SARS-CoV and SARS-CoV-2, respectively
   ([188]Supplementary File 3). We predicted the human target genes by
   utilizing three different target prediction tools, and, reducing false
   positives, we have taken only the common set. This returned 5,292 and
   6,369 human target genes for SARS-CoV and SARS-CoV-2, respectively
   ([189]Supplementary File 4). Out of these, 2,992 genes are found to be
   common in both, while 2,300 and 3,377 genes were found to be unique
   targets of SARS-CoV and SARS-CoV, respectively. An apparent difference
   of the coding regions of miRNAs between SARS-CoV and SARS-CoV-2 was
   observed ([190]Figures 6A,B).

Figure 6.

   [191]Figure 6
   [192]Open in a new tab

   Genome browser view of viral miRNAs transcribed from the regions of (A)
   SARS-CoV (Reference) and (B) SARS-CoV-2 (Reference) genomes.

SARS-CoV and SARS-CoV-2 Can Evade Host's Immune Surveillance Pathway by
Utilizing Its miRNAs

   Many viruses use their miRNAs to suppress or escape host's immune
   responses (Mishra et al., [193]2020). To identify which pathways are
   associated with SARS-CoV and SARS-CoV infection, we have performed the
   gene ontology (GO) and pathway functional enrichment of the targeted
   genes using different tools. This revealed a myriad of significant
   functions and pathways involved in host immune responses, such as Wnt
   signaling (Ljungberg et al., [194]2019), MAPK signaling (Kimura et al.,
   [195]2013), T-cell-mediated immunity (Channappanavar et al.,
   [196]2014), autophagy (Yordy and Iwasaki, [197]2011), FGF receptor
   binding (van Asten et al., [198]2018), TGF-beta signaling (Denney et
   al., [199]2018), VEGF signaling (Alkharsah, [200]2018), ErbB signaling
   (Zheng et al., [201]2014), mTOR signaling (Le Sage et al., [202]2016),
   and TNF-alpha signaling (Kimura et al., [203]2013) are particularly
   targeted by SARS-CoV-2 ([204]Figures 7A–E).

Figure 7.

   [205]Figure 7
   [206]Open in a new tab

   Enrichment analysis and comparison between the SARS-CoV and SARS-CoV-2
   encoded viral miRNAs' target human genes. Functional enrichment using
   gitools: the (A) GO Biological Process module and (B) GO Molecular
   Funtion module. Enriched pathways obtained from (C) the Webgestalt
   (KEGG and Wikipathways) tool, (D) DAVID (KEGG pathways) tool, and (E)
   EnrichR (KEGG, Wikipathways, BioPlanet pathways) tool. Color codes are
   as in [207]Figure 5. Only selected significant enriched terms are
   shown.

   Functions and pathways, such as heart development, brain development,
   and the insulin signaling pathway, etc. ([208]Figures 7A–E), were also
   enriched for SARS-CoV-2 only, which can be targeted by the viral
   miRNAs, making the patients with previous complications more
   susceptible to COVID-19, and it can also lead to several signs uniquely
   found in SARS-CoV-2-infected patients.

   We have also identified the downregulated target genes by curating the
   GEO expression datasets ([209]GSE17400 for SARS-CoV and [210]GSE147507
   for SARS-CoV-2) and found 120 and 35 downregulated target genes in
   SARS-CoV and SARS-CoV-2, respectively ([211]Supplementary File 5).
   These downregulated target genes are found to be involved in different
   immune-signaling pathways as well as different pathways related to
   organ-specific function ([212]Figure 8).

Figure 8.

   [213]Figure 8
   [214]Open in a new tab

   Enrichment analysis and comparison between the enriched pathways of (A)
   SARS-CoV and (B) SARS-CoV-2 encoded viral miRNAs' downregulated target
   genes, obtained using the EnrichR (KEGG, Wikipathways, and BioPlanet
   pathways) tool. –Log[10] (P-value) scale is utilized for all in a bar
   graph. The higher the bar height, the more significant an enriched term
   is. Only selected significant enriched terms are shown.

Discussion

   Cellular miRNAs play a crucial role during the viral infection to
   strengthen host immunity by targeting virus genes as well as pathways
   that viruses utilize for their survival and immune evasion (Girardi et
   al., [215]2018). Viruses themselves can encode their miRNAs to target
   these immune-signaling pathways (Bruscella et al., [216]2017). COVID-19
   has become a serious public health issue, though the complete molecular
   mechanism of pathogenesis is not fully understood yet. In this context,
   we have carried out this whole study to investigate the miRNA-mediated
   interactions between the host and the SARS-CoV-2 virus, which might
   shed some light on the tug-of-war between host's immune responses and
   virus's circumvention strategies. Though the disease conditions caused
   by SARS-CoV and SARS-CoV-2 are more or less similar, several unique
   features [i.e., long incubation, enhanced latency, asymptomatic
   infection, intense pain, severe lung damage, etc. (Ceccarelli et al.,
   [217]2020)] of SARS-CoV-2 make it more challenging to manage compared
   to SARS-CoV. We also sought to find out if there are any existing
   differences between SARS-CoV and SARS-CoV-2 in the context of
   miRNA-mediated regulation of host responses.

   As host miRNAs are one of the key sources of immune protection against
   viral infections, we have tried to find out which cellular miRNAs can
   target SARS-CoV and SARS-CoV-2 genes. Due to differences in the genome
   sequences between these two viruses, there was a significant difference
   between cellular miRNAs and their targeting viral genes. Likewise, some
   of the commonly found cellular miRNAs were showing differential binding
   preferences for these viral genes ([218]Figure 2A). Previous study by