Avian influenza remains a serious threat to poultry and human health. From December 2003 to April 2013, more than 600 human infections and 374 deaths have been reported to the World Health Organization [1]. Outbreaks of H5N1 in poultry swept from Southeast Asia to many parts of the world. To date, there is still no sign that the epidemic is under control.

While it has been well documented that infection with H5N1 results in high mortality in humans [2–5], the cellular pathway leading to such adverse outcome is unknown. The naive host immune system cannot be the sole explanation as infection of other avian influenza viruses, e.g. H9N2, only results in mild infections [6]. While the predilection of H5N1 towards cells in the lower respiratory tract contributes to the development of severe pneumonia [7], the available clinico-pathological evidence indicates that the infected patients progress to multi-organ failure early in the course of illness, and the degree of organ failure is out of proportion to the involvement of infection [8–10]. Cytokine storm and reactive haemophagocytic syndrome are the key features that distinguish H5N1 infection from severe seasonal influenza. These indirect mechanisms seem to play an even more important role than direct cell killing due to lytic viral infection.

MiRNAs, a new class of endogenous, 18–23 nucleotide long noncoding and single-stranded RNAs, were recently discovered in both animals and plants. They trigger translational repression and/or mRNA degradation mostly through complementary binding to the 3′UTR of target mRNAs. Studies have shown that miRNAs can regulate a wide array of biological processes such as cell proliferation, differentiation, and apoptosis [11–14].

Given the nature of viruses, being intracellular parasites and using the cellular machinery for their survival and replication, the success of the virus essentially depends on its ability to effectively and efficiently use the host machinery to propagate itself. This dependence on the host also makes it susceptible to the host gene-regulatory mechanisms, i.e. the host miRNAs may also have direct or indirect regulatory role on viral mRNAs expression.

Recently, several reports indicated that miRNAs can target influenza viruses and regulate influenza virus replication. In one report, 36 pig-encoded miRNAs and 22 human-encoded miRNAs were found to have putative targets in swine influenza virus and Swine-Origin 2009 A/H1N1 influenza virus genes, respectively [15]. In another report, results showed that miR-323, miR-491 and miR-654 could inhibit replication of H1N1 influenza A virus through binding to the conserved region of the PB1 gene [16]. These miRNAs could downregulate PB1 expression through mRNA degradation instead of translation repression [16]. Besides targeting influenza virus, cellular miRNAs were also implicated in the lethal infections of mice with a highly pathogenic 1918 pandemic H1N1 influenza virus [17]. A previous study on miRNA gene expression in avian influenza virus infected chicken showed that miR-146, which was previously reported to be associated with immune-related signal pathways in mammals, was found to be differentially expressed in infected tissues [18]. Moreover, a study of profiling cellular miRNAs of lung tissue from cynomolgus macaques infected with a highly pathogenic H5N1 avian and a less pathogenic 1918 H1N1 reassortant virus identified that 23 miRNAs were associated with the extreme virulence of highly pathogenic H5N1 avian virus [19]. Also, the predicted gene targets of the identified miRNAs were found to be associated with aberrant and uncontrolled inflammatory responses and increased cell death [19].

This study aimed at elucidating how avian influenza infection perturbs the human gene regulatory pathways leading to adverse pathological events, e.g. cytokine storm. We hypothesized that miRNAs could be involved in influenza virus infection response and began addressing this hypothesis using a microarray-based screening. The ultimate goal of this study is to generate essential information for further studies to identify novel intervention targets to ameliorate the adverse outcome of infection.

In this study we examined the connection between influenza A virus infection and the global patterns of cellular miRNA expression. The major observations from this work were that influenza A virus infection resulted in the altered regulation of cellular miRNAs. Avian influenza A virus can alter cellular miRNAs to a greater extent than that of seasonal human influenza A virus.

Influenza A virus affects the regulation of many cellular processes. In some cases, these changes are directed by the virus for its advantage and others are cellular defense responses to infection. Here, we found that influenza A virus infection led to altered regulation of cellular miRNAs. Given the number of genes that can be regulated by individual miRNAs and the number of miRNAs expressed in cells, this greatly expands the range of possible virus-host regulatory interactions. The complexity is underscored by there being no uniform global pattern of regulation; rather, it appears that individual (or groups of) miRNA are independently regulated, some positively and some negatively. Persistent and transient effects were seen, and changes in miRNA expression profiles were linked to the time course of infection. As a summary, miR-1246, miR-663 and miR-574-3p were up-regulated (>3-fold, p<0.05) at 24-hour post-infection with subtype H5 as compared with non-infected control cells. Moreover, miR-100*, miR-21*, miR-141, miR-1274a and miR1274b were found to be down-regulated (>3-fold, p<0.05) in infection with subtype H5, particularly at 18 or 24 hours post-infection as compared with non-infected control cells. Interestingly, many of the virally regulated miRNAs were predicted by TargetScan to target important biological pathways, immune-related signal pathways and have altered regulation in some cellular defense and some states of cellular differentiation [22].

In our study, we found that the expression of miR-141 was affected by influenza A virus infection. To validate the in silico findings empirically on the target of miR-141, we checked whether transient-transfection of anti- and pre-mir-141 into NCI-H292 cells resulted in TGF-β2 regulation. In our experiment, the transfection efficiency was an important factor affecting the degree of regulation on the target gene(s). In the case of higher transfection efficiency, as more miRNA would be transfected into the cells, the effect of gene(s) regulation by miRNA transfected would be greater. In our study, the transfection efficiency was about 78.2 ± 6.3% (mean ± SD), which was considered to be adequate for further functional analyses. During transfection, some oligonucleotide molecules were sequestered in internal vesicles and physically separated from their targets in the cytoplasm; and then released during cell lysis. Therefore monitoring miRNAs by qPCR after transfection would not be valuable. Previous researchers of this procedure had highly recommended investigating the target mRNAs and proteins instead of miRNA quantification. The time point of 24-hour post-transfection or post-infection was chosen for evaluation because miR-141 induction was observed at the early stage of virus infection, and sufficient time might be required for the miR-141 to have effect on its target(s), so we had chosen 24-hour post-transfection or post-infection for evaluation of the effect of this miRNA.

Indeed, upon detecting the TGF-β2 expression at mRNA and protein levels, we found that the altered miR-141 expression would affect the expression of the cytokine- TGF-β2. Literature search on the background of miR-141 confirmed that miR-141 is a member of the miR-200 family (miR-200a, miR-200b, miR-200c, miR-141 and miR-429). Previous studies of miR-141 were mainly on its role in cancer. It has been reported that miR-141 were markedly downregulated in cells that had undergone epithelial to mesenchymal in response to TGF-β. MiR-141 was also found to be overexpressed in ovarian and colorectal cancers [23, 24] and down-regulated in prostate, hepatocellular, renal cell carcinoma and in gastric cancer tissues [25–28] raising a controversial issue about the role of miR-141 in cancer progression. Furthermore, the miR-200 family members play roles in maintaining the epithelial phenotype of cancer cells [29]. A member of this family - miR-200a was also found to be differentially expressed in response to influenza virus infection in another study [17]. The targets of miR-200a are associated with viral gene replication and the JAK-STAT signaling pathway, which is closely related to type I interferon-mediated innate immune response [17]. However, the effect of miR-141 on virus infection was not known, except one recent report showing that enterovirus can induce miR-141 and contribute to the shutoff of host protein translation by targeting the translation initiation factor eIF4E [30].

In addition, evidence suggests that influenza A virus infection reduces or promotes the expression of the host miR-141 in a time dependent manner. We found that TGF-β2 mRNA was suppressed in miR-141 overexpressed cells. Our observation is in line with another study showing that the 3′UTR of TGF-β2 mRNA contained a target site for miR-141/200a and the expression of TGF-β2 was significantly decreased in miR-141/200a transfected cells [22]. Furthermore, miR-141 may not only work as translational repressors of target mRNAs, because it was observed that they also caused a decrease in TGF-β2 mRNA levels. These findings are similar to recent data demonstrating that some miRNAs can alter the mRNA levels of target genes [31]. This ability is probably independent of the ability of these miRNAs to regulate the translation of target mRNAs [14].

We also noted that antagomiR-141 moderately increased the accumulation of TGF-β2 protein during influenza virus infection. This might be because, by the use of anti-miR miR-141 inhibitor, which decreases the cellular pool of miR-141, the translation control of the TGF-β2 mRNA was subsequently released and caused the TGF-β2 protein to express and accumulate during virus infection. However, it was also observed that when there was an increase in TGF-β2 mRNA level, the corresponding TGF-β2 protein expression level would be increased, except in the case of non-miR-141-inhibitor treated H5N1 infected cells. In this case, there was a decrease in TGF-β2 mRNA level, while the TGF-β2 protein was increased. This might be explained by the fact that TGF-β2 mRNA degradation induced by miR-141 might be much faster than that of the corresponding protein degradation.

Recently, we had also reported that H1N1 was the only subtype that could induce a sustained increase in TGF-β2 at protein level [21]. That observation coincides with our results in this study, showing that H1N1 infection induced a little amount of miR-141 expression, while H5N1 infection induced a higher amount of miR-141 expression at the early phase of infection. As a consequence of the higher amount of miR-141 in H5N1 infection, TGF-β2 expression might be more greatly reduced than that in H1N1 infection. Since TGF-β2 can act as both an immunosuppressive agent and a potent proinflammatory molecule through its ability to attract and regulate inflammatory molecules, it plays a vital role in T-cell inhibition. Furthermore, it has been reported that TGF-β2 inhibits Th1 cytokine-mediated induction of CCL-2/MCP-1, CCL-3/MIP-1α, CCL-4/MIP-1β, CCL-5/RANTES, CCL-9/MIP-1γ, CXCL-2/MIP-2, and CXCL-10/IP-10 [32]. Moreover, the pro-inflammatory responses during influenza A virus infection are tightly controlled by anti-inflammatory mediators, such as TGF-β2, to protect the easily damageable lung tissue from destructive side effects associated with virus induced inflammation. Therefore, the downregulation of TGF-β2 protein by miR-141 may be an important step in the excessive inflammation progression during influenza A virus infection, particularly in H5N1 infection. However, whether the recovery of TGF-β2 expression by anti-miR miR-141 inhibitor could resolve the hypercytokinemia stage of H5N1 infection needs to be further studied.

Although our findings were obtained from an in vitro model, we could apply these to the real situation of an in vivo model or tissue comprised of different cell types. In real bronchial environments, lung epithelial cells are the key target of influenza viruses [33, 34]. After these cells are infected, they will activate an inflammatory cascade which launches a quick antimicrobial reaction and directs adaptive immunity to mount a protective response. Bronchial epithelial cells therefore modulate the activation of monocytes, macrophages, dendritic cells (DC), and T lymphocytes through cytokines and chemokines. Cytokines and chemokines generally function in an autocrine (on the producing cell itself) or paracrine (on nearby cells) manner. These mediators will contribute to the generation of a specific bronchial homeostatic microenvironment that affects the way in which the body copes with the viruses. This homeostatic “circuit” can inhibit excessive inflammatory response in lung tissues [35]. For example, TGF-β had been reported to mediate a cross-talk between alveolar macrophages and epithelial cells [36]. However, our findings show that, during highly pathogenic H5N1 avian virus infection, miR-141 would be induced shortly after infection. With high level of miR-141, the expression of TGF-β would be suppressed from the lung epithelial cells. Without sufficient TGF- β, the pro-inflammatory response might not be tightly controlled in cases of highly pathogenic H5N1 avian virus infection. This might explain the mechanism concerning bronchial infiltration of inflammatory cells, particularly lymphocytes and eosinophils, and the subsequent hyperresponsiveness of the bronchial wall induced by viral infection.

Our study has some limitations that will need to be addressed in future studies. Firstly, we did not assess the roles of other miRNAs whose expression were also altered after infection. The miRNA microarrays that we used did not contain probes for every known miRNA; thus it is possible that influenza A virus infection affects the expression of some other miRNAs not yet covered by the kit used in the current study. Secondly, the virus may interact with miRNA regulatory pathways differently in other cell or tissue types, or in other physiological status.

In conclusion, based on the broad-catching miRNA microarray approach, we found that dysregulation of miRNA expression is mainly observed in highly pathogenic avian influenza infection. We identified that miR-141 was induced at early time points upon influenza A virus infection. The induction was higher in H5N1 infection than that of seasonal H1N1 infection. Moreover, TGF-β2, which plays an important role in regulating inflammatory processes, was identified as a target of miR-141 binding. As a result, influenza A virus infection, in particular highly pathogenic H5N1, could affect the inflammatory processes via miR-141 induction.