Comprehensive Transcriptomic Analysis Identifies Novel Antiviral Factors Against Influenza A Virus Infection

Influenza A virus (IAV) has a higher genetic variation, leading to the poor efficiency of traditional vaccine and antiviral strategies targeting viral proteins. Therefore, developing broad-spectrum antiviral treatments is particularly important. Host responses to IAV infection provide a promising approach to identify antiviral factors involved in virus infection as potential molecular drug targets. In this study, in order to better illustrate the molecular mechanism of host responses to IAV and develop broad-spectrum antiviral drugs, we systematically analyzed mRNA expression profiles of host genes in a variety of human cells, including transformed and primary epithelial cells infected with different subtypes of IAV by mining 35 microarray datasets from the GEO database. The transcriptomic results showed that IAV infection resulted in the difference in expression of amounts of host genes in all cell types, especially those genes participating in immune defense and antiviral response. In addition, following the criteria of P<0.05 and |logFC|≥1.5, we found that some difference expression genes were overlapped in different cell types under IAV infection via integrative gene network analysis. IFI6, IFIT2, ISG15, HERC5, RSAD2, GBP1, IFIT3, IFITM1, LAMP3, USP18, and CXCL10 might act as key antiviral factors in alveolar basal epithelial cells against IAV infection, while BATF2, CXCL10, IFI44L, IL6, and OAS2 played important roles in airway epithelial cells in response to different subtypes of IAV infection. Additionally, we also revealed that some overlaps (BATF2, IFI44L, IFI44, HERC5, CXCL10, OAS2, IFIT3, USP18, OAS1, IFIT2) were commonly upregulated in human primary epithelial cells infected with high or low pathogenicity IAV. Moreover, there were similar defense responses activated by IAV infection, including the interferon-regulated signaling pathway in different phagocyte types, although the differentially expressed genes in different phagocyte types showed a great difference. Taken together, our findings will help better understand the fundamental patterns of molecular responses induced by highly or lowly pathogenic IAV, and the overlapped genes upregulated by IAV in different cell types may act as early detection markers or broad-spectrum antiviral targets.


INTRODUCTION
Influenza A virus (IAV) infection causes severe respiratory symptoms and persistent morbidity as well as mortality during annual seasonal or pandemic outbreaks, resulting in a severe threat to public health and safety, and even huge economic burden (1). Over the past decade, influenza outbreaks and pandemics have been caused by different subtypes of IAV, including H1N1, H3N2, swine-origin H1N1, and highly pathogenic avian influenza viruses (2)(3)(4), suggesting that the deeper biologic and epidemiologic mechanisms should be revealed to confidently and accurately predict the next influenza outbreak. Accumulative evidence has shown that IAV is capable of eliciting cellular immune response thought changing the expression of multiple genes in diverse types of cells, which in turn inhibit IAV infection. Airway epithelial cells are the preferred location for IAV replication and dissemination, and IAV infection induced toll-like receptors (TLRs)-related genes expression in responses to the pathogen (5). Moreover, other cell types, including endothelial cells, macrophages, monocytes, dendritic cells, and neutrophils, play important roles in response to IAV infection (6)(7)(8)(9)(10). During IAV infection, interferon, interferon-stimulated genes, and cytokines were secreted and activated in epithelial cells and immune cell types such as macrophages, monocytes, and neutrophils to facilitate antiviral responses. However, IAV could also elicit inflammation and cause various disorders of the respiratory system. Therefore, the systematic comparison of host responses of various types of cells to a range of strains of IAV still need to be further investigated.
Microarray technology with maturity is a powerful tool for the global view of gene expression levels, and enormous amounts of genome-wide gene expression microarray studies were distributed and archived in the gene expression omnibus (GEO) repository at the National Centre for Biotechnology Information (NCBI) in the last few decades, providing the chance for investigators revisiting these data to solve scientific questions. In this current study, we collected various transcriptomic datasets that were involved in diverse types of cells infected with subtypes of IAV, in order to examine common aspects of host cell responses to various subtypes of IAV infection. By integrating the global gene expression data, our results suggested that although the differentially expressed genes involved in host responses might not conform, the similar immune responses of diverse cell types were triggered by the infection of different subtypes of IAV.

Overlap Genes and Functional Enrichment Analysis
DEGs with abs (logFC) > 1.5 and P value < 0.05 from each data series were obtained to analyze the overlap genes. Then, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) biological pathway analyses were performed to predict the functionalities of differentially expressed overlap genes using the R package clusterProfiler (31). In this study, a Venn diagram, heatmap, and volcano plot were constructed using R language and R packages, including VennDiagram (32), ggplot2 (33), and pheatmap (34).

Host Transcriptional Response to Influenza Virus Infection on Human Lung Epithelial Cells (A549)
To illustrate host cell response to influenza virus infection, the global gene expression profiles from four cell-based time-series gene expression datasets in A549, a lung epithelial cell known to be highly susceptible to IAV, were analyzed ( Table 1) (11)(12)(13)(14)(15).

Host Transcriptional Response to Influenza Virus Infection on Phagocytes
Phagocytes containing monocytes, macrophages, dendritic cells, and neutrophils are critical in the recognition, engulfment, and destruction of invading pathogens. To uncover the antiviral activities of phagocytes against different subtypes of influenza viruses, the transcriptome analysis of gene expression profiles was performed in human phagocytic cells (PBMC, monocytes, macrophages, pDC, and neutrophils) that were infected with different subtypes of influenza viruses, including H1N1, H5N1, and H7N7 (Table 5) (24,27,(35)(36)(37)(38)(39)(40)(41). Following the criteria of P<0.05 and |logFC|≥1.5, 12 genes (C19orf66, IL6, HESX1,  IFNB1, IFITM3, ISG20, ISG15, HERC5, IFIT1 monocyte-derived macrophages and primary alveolar macrophages ( Figure 7A). Moreover, BAFT2 was also significantly increased in multiple data series of macrophages, except GSE27702 where the BATF2 gene was not included in the microarray platform. On the side, 29, 54, and 35 overlapped DEGs were significantly regulated in H1N1-infected monocytes, monocyte-derived DCs, and neutrophils, respectively (Supplementary File 2), and further Gene Ontology (GO) and Venn diagram analysis showed that H1N1 infection triggered a stronger host response with an elevated expression of cytokines and interferons signaling molecules in phagocytic cells, which in turn blocked IAV infection ( Figures 7B-D). Five common genes (OASL, IFIT3, RSAD2, HERC5, and IFIT1) identified in macrophages were also found to be upregulated in monocytes, monocyte-derived DCs, and neutrophils infected with H1N1.
In addition, to illustrate whether there was a similar response of phagocytes to other subtypes of influenza A virus infection, the globe gene expression profiles of avian H5N1 and H7N7infceted monocytes and monocyte-derived macrophages were analyzed. The Venn diagram revealed that a total of 125 different expression genes were found to overlap in the avian flu-infected cells ( Figure 8A). Moreover, the results of GO and KEGG pathway analysis showed that the overlap genes were mainly concentrated on cytokine-mediated signaling, type I interferonmediated signaling, defense response to virus, and necroptosis ( Figures 8B, C). A similar antiviral response, including the type I interferon-mediated signaling pathway, cytokine signaling pathway, and antiviral defense in H1N1-infceted phagocytes was also induced in human phagocytes during highly or lowly pathogenic influenza virus infection, although an overlapped gene was not found in H1N1, H5N1, and H7N7-infceted phagocytes.

DISCUSSION
Host defense responses elicited by IAV are critical to protect the host against IAV. However, the mechanisms underlying how the host response is activated among IAVs was not fully  GSM335851  GSM330315  GSM335396  GSM335904  GSM330314  GSM335395  GSM335859  GSM335906  GSM330316  GSM336559   OASL  HERC5  USP18  IFIH1  IFIT1  XAF1  OAS1  IFIT2  IFI44  IFIT3  IFI44L  OAS2  CXCL10   understood. In this study, we collected and analyzed a large amount of gene expression data of different host cells infected with different subtypes of IAV. We found that IAV infection could affect host response and generate special differentially expressed genes in the distinct cell lines. And then we identified some upregulated genes (IFIT2, HERC5, and BATF2) that may exhibit antiviral activities in different cell types against IAV infection. Moreover, with the available data collections from distinct types of cells with different strains of IAV infection, we compared gene expression difference and the potential response difference of distinct types of cells to IAV stains. Epithelial cells are the primary targets of IAV infection and can produce a protective environment by continuously secreting antiviral substances to initiate defense responses against infection. Many studies have shown that epithelial cells could produce many host cellular restriction factors in response to IAV infection (42). We summarized gene expression profiles of different human epithelial cells including human lung epithelial cells (A549), human airway epithelial cells (Calu-3), primary human bronchial epithelial cells (HBEC), well-differentiated human bronchial epithelial cells (wd-NHBE), human primary airway epithelial cells, human type I-like alveolar epithelial cells, and human type II-like alveolar epithelial cells. The results showed that various subtypes of IAV infection affected the amount of differentially expression genes, mainly involved in the interferon signaling pathway and cytokine and chemokine signaling pathway in different epithelial cells. But some overlapped genes (IFIT2, HERC5, and BATF2) were found to upregulate during different subtypes of IAV infection.
Interferon-induced protein with tetratricopeptide repeats 2 (IFIT2) is an interferon-stimulated gene (ISG) with a possible RNA-binding capacity and acts as an important restriction factor for many viruses, including rabies virus, Sendai virus, mouse hepatitis virus, hepatitis B virus, West Nile virus, and influenza virus (43)(44)(45)(46)(47)(48). However, recent genome-wide knockout screens provided a novel proviral function of IFIT2, and knockout of IFIT2 remarkedly reduced diverse influenza viruses infection by decreasing the translational efficiency for viral mRNA and IFIT2-bound mRNAs. The influenza virus hijacked IFIT2 to preferentially bind viral mRNAs and prevent ribosome pausing for increasing viral replication (49).
HERC5, an interferon-induced HECT E3 enzyme was identified to upregulate in distinct cell types infected with different subtypes of IAV and might potentially play an important role for IAV replication. Existing research has revealed the antiviral activity of HERC5 against IAV; knockdown of HERC5 weakens IFN-beta-induced antiviral activities against IAV. HERC5 activated the ISGylation system by catalyzing conjugation of ISG15 onto IAV-NS1 proteins, leading to ISG15 modification of NS1 protein and blocking of the nuclear import of the NS1A protein (50,51). In addition, our result that HERC5 acts as a key antiviral factor in IAV infection was in accordance with the result of a previous report, which identified HERC5 as a potential novel biomarker for the treatment of IAV by employing weighted gene co-expression network analysis (WGCNA) (52). Moreover, we found that although some microarray platforms did not include the BATF2 gene owing to the design of the array, the mRNA expression of BATF2, a member of AP-1 family transcription factor (53), was still significantly upregulated in multiple cells infected with different subtypes of IAV. BATF2 was proved as an antibacterial gene and was able to induce inflammatory responses in lipopolysaccharides and mycobacterium tuberculosis infection (54). IFNg induced high levels of BATF2 mRNA expression to downregulate trypanosoma cruzi-induced IL-23 production in innate immune cells by blocking the recruitment of the c-JUN-ATF-2 heterodimer to the IL23a promoter and preventing the formation of the c-JUN-ATF-2 complex, and IFN-g-induced BATF2 expression plays a key role for controlling Th17-mediagted immune responses during trypanosoma cruzi infection (54). In addition, BATF2 was broadly highly expressed in multiple tissues, including the spleen, lung, small intestine, cecum, and large intestine, and IFN-g-induced BATF2 also disturbed T cell-mediated intestinal inflammation through the regulation of the IL-23/IL-17 axis that was associated with intestine inflammation (55). Previous studies reported that BATF2 was a proapoptotic gene and overexpression of BATF2 could lead to inhibition of DNA binding activation protein (AP1), which causes growth inhibition and induces apoptosis particularly in cancerous cells (53). Additionally, BATF2 could dephosphorylate phosphor-STAT3 to promote DUSP2 expression and upregulate of NF-kB activity, and could also be modified by N6-methyladenosine (m6A) to suppress its expression (56,57). During feline infectious peritonitis virus (FIPV) infection, BATF2 showed continuous high expression and might be an important regulation factor of the death stages of infected cells (58). Our transcriptome analysis also showed that BATF2 was significantly increased in distinct cell types with IAV stains infection, indicating that it will be increasingly interesting to illustrate the role of BATF2 during IAV infection.
In conclusion, we demonstrated the gene expression pattern and molecular responses of distinct cells types among different subtypes of IAV infection. In general, IAV strains triggered a similar defense response among distinct cells types via the production of various antiviral cytokines and interferon-related genes, although few overlapped genes were present in distinct cells types. We identified that IFIT2, HERC5, and BATF2 might act as key antiviral factors to regulate IAV infection, but the molecular regulatory mechanisms of IFIT2, HERC5, and BATF2 involved in IAV infection still need to be validated.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

AUTHOR CONTRIBUTIONS
All authors contributed to the article and approved the submitted version. AZ designed this project, analyzed and interpreted data, and contributed to the writing of the manuscript. XD collected the data and modified the manuscript. BT analyzed and interpreted the data. ML drew the figures. BT and AZ supervised the financial support.