Abstract

   The ongoing COVID-19 pandemic caused by SARS-CoV-2 infections has
   quickly developed into a global public health threat. COVID-19 patients
   show distinct clinical features, and in some cases, during the severe
   stage of the condition, the disease severity leads to an acute
   respiratory disorder. In spite of several pieces of research in this
   area, the molecular mechanisms behind the development of disease
   severity are still not clearly understood. Recent studies demonstrated
   that SARS-CoV-2 alters the host cell splicing and transcriptional
   response to overcome the host immune response that provides the virus
   with favorable conditions to replicate efficiently within the host
   cells. In several disease conditions, aberrant splicing could lead to
   the development of novel chimeric transcripts that could promote the
   functional alternations of the cell. As severe SARS-CoV-2 infection was
   reported to cause abnormal splicing in the infected cells, we could
   expect the generation and expression of novel chimeric transcripts.
   However, no study so far has attempted to check whether novel chimeric
   transcripts are expressed in severe SARS-CoV-2 infections. In this
   study, we analyzed several publicly available blood transcriptome
   datasets of severe COVID-19, mild COVID-19, other severe respiratory
   viral infected patients, and healthy individuals. We identified 424
   severe COVID-19 -specific chimeric transcripts, 42 of which were
   recurrent. Further, we detected 189 chimeric transcripts common to
   severe COVID-19 and multiple severe respiratory viral infections.
   Pathway and gene enrichment analysis of the parental genes of these two
   subsets of chimeric transcripts reveals that these are potentially
   involved in immune-related processes, interferon signaling, and
   inflammatory responses, which signify their potential association with
   immune dysfunction leading to the development of disease severity. Our
   study provides the first detailed expression landscape of chimeric
   transcripts in severe COVID-19 and other severe respiratory viral
   infections.

   Keywords: chimeric transcripts, severe COVID-19, aberrant splicing,
   cis-SAGe, respiratory viral infections

1. Introduction

   The ongoing COVID-19 pandemic is caused by SARS-CoV-2, a recently
   identified member of the Coronaviridae family [[30]1,[31]2]. The
   clinical representation of COVID-19 is variable, and sometimes it can
   lead to the development of disease severity resulting in immune
   dysfunction, acute respiratory distress syndrome (ARDS), and
   multi-organ failure [[32]3,[33]4,[34]5]. Despite several studies, the
   molecular mechanisms behind the development of severe COVID-19 are
   still unclear. During the severe COVID-19 infection, SARS-CoV-2
   antagonizes interferon (IFN) induction and IFN signaling, which
   generates dysfunctional immune response
   [[35]6,[36]7,[37]8,[38]9,[39]10,[40]11,[41]12]. Recent studies
   demonstrated that SARS-CoV-2 suppresses the interferon (IFN) response
   by disrupting the host cell splicing [[42]13]. Aberrant splicing and
   dysregulation of transcripts were found to be strongly correlated with
   the clinical severity of COVID-19 [[43]14,[44]15].

   Aberrant splicing promotes the increase in non-canonical splicing
   events, resulting in the generation of novel chimeric transcripts
   [[45]16,[46]17,[47]18,[48]19]. In several cancers and complex diseases,
   non-canonical splicing was observed, which was reported to generate
   novel chimeric transcripts [[49]18,[50]20,[51]21]. Chimeric transcripts
   are formed by the fusion of the exon/intron from two separate genes
   [[52]22]. The generation of chimeric transcripts was found to be
   functionally involved with the development of several diseases
   [[53]23,[54]24]. Further, the appearance of chimeric transcripts in the
   cell is thought to be associated with the generation of phenotypic
   plasticity that helps the cell’s survival in response to specific
   stress [[55]25,[56]26]. Chimeric transcripts could modify cell
   functionality by altering the regulatory network and protein
   interaction network, and they can also switch pathways
   [[57]27,[58]28,[59]29].

   SARS-CoV-2 induces the widespread alternation of host cell-splicing
   machinery and transcriptional dynamics in severe COVID-19 infection,
   which raises the possibility of the generation and expression of novel
   chimeric transcripts in severe COVID-19-infected cells. However, no
   study so far has attempted to check if SARS-CoV-2-induced aberrant
   splicing in severe COVID-19 infections could generate novel chimeric
   transcripts, nor the potential functional impact of these chimeric
   transcripts’ generation. In this study, we integrated publicly
   available RNA-seq datasets of blood from 85 severe COVID-19 patients,
   10 mild COVID-19 patients, 84 other severe respiratory viral infected
   patients, 54 healthy persons, and 199 samples from 32 different tissues
   and identified the chimeric transcripts unique to severe COVID-19
   infections.

2. Materials and Methods

2.1. Acquisition of the RNA-Seq Datasets

   RNA-seq data from several sources were collected for this study. A
   total of 44 blood samples of RNA-seq data from severe COVID-19 patients
   and 10 samples from healthy donors were obtained from the Gene
   Expression Omnibus (GEO) database ([60]GSE171110) [[61]30]. RNA-seq
   data for 24 severe COVID-19 and 34 healthy blood samples were obtained
   from the [62]GSE152418 dataset [[63]31]. We also collected single-cell
   RNA sequencing data of peripheral blood mononuclear cells (PBMCs) from
   8 severe COVID-19 patient samples and 6 healthy controls (PRJNA633393)
   [[64]32]. We downloaded the RNA-seq data of blood from EBI ArrayExpress
   (accession E-MTAB-10926) for 10 mild COVID-19 patients and 10 severe
   COVID-19 patients [[65]33]. From the [66]GSE157240 dataset, we
   downloaded the RNA-seq data for 84 blood samples of patients with
   severe respiratory viral infections caused by influenza,
   enterovirus/rhinovirus, human metapneumovirus, dengue virus,
   cytomegalovirus, Epstein–Barr virus, or adenovirus was also obtained
   [[67]34]. Further, we downloaded RNA-seq data from 199 samples from 122
   individuals representing 32 different healthy human tissues from EBI
   ArrayExpress (accession E-MTAB-2836). The downloaded raw sequence reads
   were converted to FASTQ using the SRA toolkit, version 2.10.7 [[68]35].

2.2. Identification of Chimeric Transcripts from the RNA-Seq Data

   Using our in-house reference-based method ChiTaH [[69]36], all the
   above samples were used to identify chimeric transcripts. Recently,
   ChiTaH [[70]36] was demonstrated to be the most efficient
   reference-based approach for chimeric transcript detections from
   RNA-seq data compared to other popular tools such as EricScript
   [[71]37], STAR-Fusion [[72]38], JAFFA [[73]39], and FusionCatcher
   [[74]40]. Furthermore, ChiTaH can also detect the chimeric transcripts
   from single-cell RNA (scRNA)-seq data. ChiTaH maps RNA-Seq datasets and
   predicts potential chimeric transcripts in each sample using 43,466
   non-redundant high-quality human chimeras from the ChiTaRS 5.0 database
   [[75]41]. The three rigorous criteria listed below were used to
   determine which transcripts were chimeric: (i) Reads should only map
   with human chimeric transcripts found in the ChiTaRS database and not
   with the human genome or transcriptome, (ii) at least five reads should
   cover the chimeric transcript junction length, and (iii) the mapping
   reads’ quality should be MAPQ > = 10.

2.3. Differential Gene Expression Analysis

   STAR aligner [[76]42] was used to map the RNA sequencing reads from all
   samples; the human reference genome (hg38) was used for alignment.
   Next, the featureCounts tool [[77]43] was used to determine the total
   amount of reads mapped to each human gene. Finally, DEseq2 [[78]44] was
   used to perform differential gene expression analysis. When
   log2foldchange ≥ 2 and Padj ≤ 0.05, a gene was deemed significantly
   upregulated, and when log2foldchange was negative and Padj ≤ 0.05, it
   was deemed significantly downregulated. The Benjamini–Hochberg method
   was used to adjust the p-values and the false discovery rate (FDR)
   [[79]45].

2.4. Gene Ontology and Pathway Enrichment Analysis of the Parental Genes of
Chimeric Transcripts

   Metascape [[80]46] was used to perform the Gene ontology (GO)
   biological process and pathway enrichment analyses. The p-value cutoff
   < 0.05 was set to select the enriched processes and pathways. The KEGG
   [[81]47] and REACTOME [[82]48] pathway databases were used during the
   pathway enrichment analysis.

3. Results

3.1. Identification of Chimeric Transcripts in the RNA-Seq Data of Blood
Samples from Severe COVID-19, Mild COVID-19, and Other Severe Respiratory
Virus-Infected Patients

   We identified a total of 957 chimeric transcripts from 78 bulk RNA-seq
   and 7 single-cell RNA (scRNA)-seq data from severe COVID-19 patients.
   In total, 1175 chimeric transcripts were identified in mild COVID-19
   patients from 10 bulk RNA-seq samples, and 1281 chimeric transcripts
   were identified in various severe respiratory viral infections from 84
   bulk RNA-seq samples. Moreover, we also identified 1061 chimeric
   transcripts from 54 healthy samples, while 2066 chimeric transcripts
   were identified in 199 samples from 32 normal tissues. Chimeric
   transcripts were observed to be abundant in all samples analyzed
   ([83]Table S1 ([84]Supplementary Materials)). However, from a
   pathological perspective, it would be interesting to identify the
   chimeric transcripts that are specific to severe COVID-19 infections,
   as well as those that are commonly expressed across multiple severe
   viral infections. We have identified that 44.3% (424 out of 957) of
   chimeric transcripts identified in the severe COVID-19 samples are
   unique to severe COVID-19 infections ([85]Figure 1). Furthermore, we
   have identified 189 chimeric transcripts, which are common to severe
   COVID-19 infection and other severe respiratory viral infections. These
   results indicate that a significant number of chimeric transcripts
   detected in individuals are common to severe COVID-19 and other severe
   respiratory viruses.

Figure 1.

   [86]Figure 1
   [87]Open in a new tab

   The table represents the number of chimeric transcripts detected in the
   different samples. The Venn diagram represents the common and unique
   chimeric transcripts detected in each sample.

3.2. Identification of Severe COVID-19 Specific Recurrent Chimeric
Transcripts and Functional Analysis of Their Parental Genes

   Chimeric transcripts expressed in multiple severe COVID-19 patient
   blood samples could be associated with the development of the severity
   of COVID-19. We have identified 42 chimeric transcripts as recurrent,
   which are expressed in at least 3 severe COVID-19 patient samples
   ([88]Table S2). The generation of chimeric transcripts could impact on
   the functionality of their parental genes [[89]27,[90]49]. For example,
   chimeric transcripts could translate into chimeric proteins that could
   replace the interactions of their parental proteins in the PPI network
   and alter the functionality of the cell [[91]27,[92]29,[93]50].
   Furthermore, chimeric transcripts could act as long non-coding RNA
   (lncRNA), which could regulate and alter the functionality of their
   parental genes [[94]49,[95]51,[96]52]. We have found 75 parental genes
   generating recurrent chimeric transcripts in severe COVID-19 patients.
   So, it can be assumed that severe COVID-19 specific, recurrent chimeric
   transcripts could impact on the functionality of their parental genes,
   which may lead to the development of disease severity. We performed the
   GO functional and pathway enrichment analysis to understand the
   potentiality for alteration of chimeric transcripts mediated by
   cellular functionality. A Metascape analysis of the most enriched
   biological processes revealed that the parental genes of chimeric
   transcripts are involved in a broad range of important processes such
   as localization, cellular processes, immune system processes, metabolic
   processes, regulation of biological processes, etc. ([97]Figure 2A).
   This observation indicates that chimeric transcripts generated from
   these genes could play a potential role in alternations of
   immune-related processes and alter the regulation of the diverse
   biological and metabolic processes that led to the development of
   disease severity. Further, the top-most enriched pathway and process
   analysis revealed that cellular response to stress, phagocytosis, and
   interferon alpha/beta signaling are the most important processes that
   could be altered by these chimeric transcripts ([98]Figure 2B).

Figure 2.

   [99]Figure 2
   [100]Open in a new tab

   (A) GO biological process enrichment analysis of the parental genes of
   severe COVID-19 specific recurrent chimeric transcripts; (B) Top-most
   enriched processes and pathways of the parental genes of severe
   COVID-19 specific recurrent chimeric transcripts detected by Metascape
   (p-value cutoff < 0.05).

3.3. Genomic Neighborhood Analysis and Differential Gene Expression Analysis
of the Parental Genes of Severe COVID-19 Specific Recurrent Chimeric
Transcripts

   Several chimeric transcripts could be generated by the non-canonical
   splicing-based mechanisms, such as the trans-splicing and cis-splicing
   of adjacent genes (cis-SAGe) [[101]25]. Trans-splicing is the mechanism
   by which two transcripts from two different genes could fuse and
   generate a novel chimeric transcript [[102]53,[103]54,[104]55,[105]56].
   However, the mechanism of trans-splicing events is not well understood
   in humans, and no such computational method exists that can predict
   whether a chimeric transcript is generated by a trans-splicing
   mechanism. cis-SAGe is a frequent mechanism in human cancers and other
   complex diseases in which this mechanism generates a significant number
   of chimeric transcripts [[106]17,[107]18,[108]57,[109]58,[110]59].
   However, a recent study demonstrated that cis-SAGe is the common
   mechanism for generating chimeric transcripts in healthy human cells
   [[111]17,[112]22]. cis-SAGe is the intergenic splicing of directly
   adjacent genes with the same transcriptional orientation, which can
   generate chimeric transcripts [[113]60]. We performed the genomic
   neighborhood analysis of the parental genes from severe COVID-19
   specific recurrent chimeric transcripts, and we found seven confirmed
   cases where those chimeric transcripts were generated by cis-SAGe
   analysis ([114]Table S2). Integrative Genomics Viewer (IGV) [[115]61]
   was used to visualize the adjacent genes. [116]Figure 3 and [117]Figure
   S1 support the close proximity of the two genes from different
   recurrent severe COVID-19 specific chimeric transcripts and highlight
   the potentiality of cis-SAGe mechanism.

Figure 3.

   [118]Figure 3
   [119]Open in a new tab

   Genomic neighborhood analysis of the parental genes of severe COVID-19
   specific recurrent chimeric transcripts. The two parental genes of
   chimeric transcripts adjacent to each other indicate the potentiality
   of cis-SAGe.

   If the emergence of chimeric transcripts can regulate the functionality
   of their parental genes, then we might expect the differential
   expression of at least one of their parental genes. Interestingly, we
   observed in several instances that at least one of the parental genes
   of recurrent severe COVID-19 specific chimeric transcripts from
   different datasets were differentially expressed ([120]Figure 4). This
   finding could suggest that chimeric transcripts generated in severe
   COVID-19 could significantly regulate their parental genes and alter
   the pathways and biological processes where their parental genes are
   involved.

Figure 4.

   [121]Figure 4
   [122]Open in a new tab

   Differential gene expression analysis of the parental genes of severe
   COVID-19 specific recurrent chimeric transcripts.

3.4. Identification of Common Chimeric Transcripts Expressed in Severe
COVID-19 and Other Severe Respiratory Viral Infections

   Common chimeric transcripts from multiple severe respiratory viral
   infections and severe COVID-19 could be associated with the immune
   response common to the development of disease severity caused by most
   respiratory viral infections. With this aim, we have identified 189
   common chimeric transcripts unique to both severe COVID-19 and multiple
   other severe respiratory viral infections ([123]Table S3). Then, we
   analyzed the functional and pathway enrichment analysis of the parental
   genes of these common chimeric transcripts. In line with the GO
   biological process enrichment of the parental genes of recurrent severe
   COVID-19 specific chimeric transcripts, here we observed the parental
   genes of common chimeric transcripts of severe COVID-19 infections and
   other severe respiratory viral infections are involved in important
   processes related to immune system processes, localization, biological
   processes involved in interspecies interaction, cellular processes,
   metabolic processes, regulation of biological processes, etc.
   ([124]Figure 5A). This observation suggests that the chimeric
   transcripts mostly generated in severe viral infections are crucial for
   immune dysfunctions and alternations of various important cellular,
   metabolic, and regulatory processes and pathways leading to the
   development of disease severity. From the top-most enriched biological
   process and pathway analysis, we found the adaptive immune response is
   the most important process ([125]Figure 5B), which indicates that these
   chimeric transcripts generated from these genes could significantly
   associate with the adaptive immune response. Furthermore, the
   inflammatory response was found to be another enriched process which is
   an important factor in viral infection-mediated disease severity
   ([126]Figure 5B). Next, we observed the significant enrichment of the
   vesicle-mediated transport pathway ([127]Figure 5B), which suggests
   that the chimeric transcripts could be associated with the transport of
   viral particles into the membrane-bounded vesicles, which is an
   efficient mechanism for viruses to enter host cells and evade the host
   immune response [[128]62,[129]63,[130]64]. This line of observations
   supports our hypothesis that the chimeric transcripts common to
   multiple severe respiratory viral infections, including COVID-19, could
   impact the various important biological processes and pathways that
   drive the stress responses leading to the development of disease
   severity.

Figure 5.

   [131]Figure 5
   [132]Open in a new tab

   (A) GO biological process enrichment analysis of the parental genes of
   common chimeric transcripts associated with multiple severe respiratory
   viral infections and severe COVID-19; (B) Top-most enriched processes
   and pathways of the parental genes of common chimeric transcripts
   associated with multiple severe respiratory viral infections and severe
   COVID-19 detected by Metascape (p-value cutoff < 0.05).

4. Discussion

   In this study, we analyzed several publicly available RNA-seq datasets
   containing 85 severe COVID-19 patients’ blood samples, 10 mild COVID-19
   patients’ blood samples, 84 other respiratory viral infected patients’
   blood samples, and 253 healthy human samples from 32 different tissues,
   including blood. We identified 424 severe COVID-19 specific chimeric
   transcripts; among these, 42 chimeric transcripts were recurrent which
   were mapped in at least 3 samples. The important enriched biological
   processes and pathways of the parental genes of these recurrent severe
   COVID-19 specific chimeric transcripts highlighted that they are
   involved in cellular stress response, phagocytosis, and interferon
   signaling. The genes associated with the cellular stress response
   pathway could be involved in the activation of innate and adaptive
   immune responses and play a pivotal role in host defenses and
   inflammation [[133]65]. Next, we observed the second most enriched
   process/pathway is phagocytosis. Recent studies demonstrated that
   alternations of the phagocytosis response are strongly associated with
   COVID-19 severity [[134]66,[135]67,[136]68]. However, the molecular
   mechanisms behind the phagocytosis alternations and COVID-19 severity
   development were not clearly understood. Our findings suggest that the
   recurrent severe COVID-19 specific chimeric transcripts could alter the
   functionality of their parental genes which are involved in the
   phagocytosis process. Further, we observed that parental genes of the
   recurrent severe COVID-19-specific chimeric transcripts were
   significantly involved in interferon signaling. Recent studies showed
   that the dysregulated interferon response is crucial for the
   development of severe COVID-19 [[137]69,[138]70]. Interferon signaling
   induces the IFN-stimulated gene (ISG) expression by phosphorylating
   STAT1, and STAT1 phosphorylation was found to increase in severe
   COVID-19 cases, indicating an imbalanced JAK/STAT signaling and lack of
   ISG-induced transcription [[139]69]. In this study, we observed that
   the cis-SAGe mechanism STAT1 gene generated a recurrent severe COVID-19
   specific chimeric transcript. Altogether, these findings could suggest
   the potential involvement of recurrent severe COVID-19 specific
   chimeric transcripts in the generation of a stress response associated
   with dysregulated interferon signaling during severe COVID-19
   infections.

   If the chimeric transcripts alter the functionality of their parental
   genes, then we should expect at least one of their parental genes could
   be differentially expressed. Interestingly, we observed in several
   instances that one of the parental genes of recurrent severe COVID-19
   specific chimeric transcripts was differentially expressed. We observed
   consistent differential gene expression patterns for some chimeric
   transcript’s parental genes in all datasets; however, the chimeric
   transcripts mapped in some datasets, but not all ([140]Figure 4). We
   chose strict criterion (Materials and Methods (2.2)) for chimeric
   transcripts detection to reduce the chances of obtaining false
   positives. So, as the quality of the RNA-seq data is different in each
   dataset, we could have missed some chimeric transcripts because fewer
   reads were mapped, or mapping read qualities were low. Similarly, the
   number of differentially expressed genes in each dataset depends on the
   quality of sequencing data and the p-value criteria selected to screen
   the differentially expressed genes. We chose the strict p-value, p
   <0.05, as the criterion for detection of differential gene expression,
   so there is a chance that we could have missed some differentially
   expressed genes in some datasets because of the quality of the
   sequencing data and read count bias. Interestingly, instead of the
   difference of sequence quality in different datasets, we mapped all the
   recurrent severe COVID-19 specific chimeric transcripts detected in the
   E-MTAB-10926 samples in the single-cell high-quality (scRNA)-seq
   dataset from the PRJNA633393 samples. Therefore, the production of the
   chimeric transcripts could be the important signature behind the
   generation of stress responses leading to COVID-19 severity
   development. Furthermore, we found 189 common chimeric transcripts
   unique to both severe COVID-19 and other severe respiratory viral
   infections whose parental genes were enriched in adaptive immune
   response and inflammatory response. This could signify these chimeric
   transcripts might be associated with stress responses associated with
   dysfunctional immune responses common to multiple respiratory viral
   severe infections.

   Our study first identified the chimeric transcripts unique to severe
   COVID-19 infections and demonstrated their potential functional
   importance ([141]Figure 6). During a severe COVID-19 infection,
   SARS-CoV-2 targets the host cell, splicing to alter the transcriptional
   dynamics of the host cell so that they can efficiently translate the
   viral proteins and evade the host cell’s defense mechanisms. The
   aberrant splicing during severe COVID-19 infections could promote the
   generation of chimeric transcripts that can alter the function of their
   parental genes, which are mainly associated with immune-related
   processes that could lead to the development of immune dysfunction and
   disease severity development. Therefore, further exploring the
   functional role of individual recurrent severe COVID-19 specific
   chimeric transcripts could aid in understanding the individual’s immune
   responses to COVID-19 infection, which might help to design
   personalized treatment strategies.

Figure 6.

   [142]Figure 6
   [143]Open in a new tab

   Schematic representation of the generation of chimeric transcripts in
   severe COVID-19 and their potential functional impact.

5. Conclusions

   The findings of this present study suggest the potential role of
   recurrent severe COVID-19 specific chimeric transcripts in alternations
   of immune-related processes, activating the processes that favor the
   viral transcription, gene expression, and alternations of various
   biological and metabolic processes that could lead to the development
   of severe disease. Furthermore, we identified common chimeric
   transcripts found in severe COVID-19 and other severe respiratory viral
   infections. The parental genes of these common chimeric transcripts are
   functionally enriched for adaptive immune response and inflammatory
   response. These observations suggest these chimeric transcripts could
   be associated with different stresses associated with the dysfunctional
   immune response generated during multiple severe respiratory viral
   infections, including COVID-19. The limitation of our study was the
   lack of experimental validation, as we relied on publicly available
   RNA-seq data from patients. Further, the study’s use of publicly
   available datasets had varying experimental designs, and the patient
   samples came from diverse populations which were collected at different
   time points during the development of COVID-19 severity. Despite the
   limitations inherent to computational analysis, this study successfully
   uncovered the hidden layers of the blood transcriptome from the various
   respiratory viral infected patients and identified several potential
   functionally relevant chimeric transcripts unique to severe COVID-19
   and common to multiple severe respiratory virus infections. Further
   experimental studies are necessary to elucidate the specific functions
   of these chimeric transcripts in COVID-19 pathogenesis.

Acknowledgments