r1918 and Tx/91 induce distinct cellular microRNA expression patterns in mouse lungs.
In a study published in 2006 (17
), we reported a correlation between the extreme virulence of r1918 and an aberrant cellular gene expression profile in the lungs of infected mice. Compared with pathology observed in mice infected by the nonlethal influenza virus Tx/91, mice infected by r1918 exhibited more severe pulmonary lesions accompanied by excessive lung inflammation. However, we did not suggest a potential mechanism causing the aberrant gene expression. During the past few years, microRNAs have emerged as a group of key regulators of cellular gene expression. We therefore sought to determine whether gene expression changes might be influenced by the differential expression of cellular microRNAs during influenza virus infection. To test this hypothesis, we made use of archived lung samples from the previous study to assess cellular microRNA expression. We also utilized the gene expression data from the same study to assess the functional associations of the differentially expressed cellular microRNAs from the perspective of their differentially expressed targets.
We started by directly comparing the cellular microRNA expression profiles present in mouse lung tissues obtained from r1918- and Tx/91-infected animals on days 1, 3, and 5 p.i. Among 567 microRNAs present on the arrays, 310 were detected in mouse lung tissues and were included into our analyses (see Table S1 in the supplemental material). In order to select the microRNAs with differences in abundance upon r1918 and Tx/91 infections, we adopted a cutoff 1.5-fold change (P
< 0.01) at one or more of three time points. A 1.5-fold cutoff on microRNA expression changes was used based on previous microRNA profiling studies, confirming that this difference can have a significant impact on the biology of cells (13
). The expression patterns of differentially expressed microRNAs between r1918- and Tx/91-infected lungs relative to a common reference, a pooled mock sample, are shown in Fig. . Notably, the number of differentially expressed microRNAs between the two virus-infected samples increased over time. Twenty-nine microRNAs exhibited differential expression on day 1 p.i. The number increased to 106 on day 3 and increased further to 132 on day 5. Thus, our data indicate that the lethal r1918 and nonlethal Tx/91 infections induce distinct cellular microRNA expression patterns in mouse lungs.
FIG. 1. The infections of H1N1 influenza viruses r1918 and Tx/91 induce distinct cellular microRNA expression patterns in mouse lungs. (a) Distinct cellular microRNA expression patterns in r1918- and Tx/91-infected mouse lungs. The columns correspond to the expression (more ...)
To confirm these results, we measured the relative abundance of a few select microRNAs using TaqMan qRT-PCR assays. Eight cellular microRNAs with known functions, including mmu-miR-34a, mmu-miR-30a, mmu-miR-200a, mmu-miR-200b, mmu-miR-133a, mmu-miR-21, mmu-miR-1, and mmu-miR-223, were tested by qRT-PCR. The fold change of a particular microRNA in the r1918 sample relative to the time-matched Tx/91 sample was calculated (see Materials and Methods). The value of the coefficient of determination, R2, was used to indicate the correlation coefficient among the fold changes derived from microarrays and from qRT-PCR assays on the same pair of samples. The calculated R2 value of 0.865 indicates a high correlation coefficient between the results obtained from microRNA microarrays and those obtained from qRT-PCR (Fig. ).
Differences in microRNA expression during r1918 and Tx/91 infections.
Among more than 130 cellular microRNAs showing distinct expression patterns between the lethal r1918 and the nonlethal Tx/91 infections, we focused on 18 microRNAs for further analyses because of their high abundance in lung tissues (see Table S1 in the supplemental material). These microRNAs had different expression patterns between the r1918- and Tx/91-infected lungs, including directly opposite regulation (up in one and down in the other), regulation in the same direction but to differing degrees, or regulation during one infection but not the other (Fig. ). For example, miR-193 was strongly downregulated during r1918 infection, while it was upregulated during Tx/91 infection. In contrast, miR-709 was strongly upregulated during r1918 infection, while it was strongly downregulated during Tx/91 infection. miR-223 and miR-21, which were strongly upregulated in r1918 infection, were moderately upregulated only upon Tx/91 infection. On the other hand, while strongly downregulated in r1918 infection, miR-29a and miR-29b were moderately downregulated only upon Tx/91 infection. Finally, the expression levels of miR-200a, miT-34a, and miR-30a were downregulated in r1918 infection but were below the cutoff in Tx/91 infection. Taken together, our data indicated that some of the most abundantly expressed cellular microRNAs demonstrated distinct expression patterns between the infections of r1918 and Tx/91 in mouse lungs.
FIG. 2. A total of 18 microRNAs demonstrate various expression patterns between r1918-infected and Tx/91-infected lungs. All fold changes shown here are relative to mock-infected lungs. A red bar indicates an r1918 infection, and a blue bar indicates a Tx/91 (more ...)
Interestingly, these microRNAs have been implicated in multiple key functions. MiR-223 and Let-7 have been shown to be involved in immune responses, and miR-223 is a negative modulator of neutrophil activation and neutrophil-mediated killing (16
). A decreased expression level of Let-7 is associated with the activation of NF-κB in response to microbial challenge (12
). Upregulation of miR-21 is closely related to airway inflammation (21
), a symptom of lethal r1918 infection, and miR-34a is associated with tumorigenesis, as the mutual activation of MiR-34a and p53 has been shown both in a human cell line (40
) and in patients (24
). In addition, a stable expression of miR-200a is critical in maintaining the phenotype of epithelial cells (11
). Taken together, because these differentially expressed microRNAs are known to be related to important functions, their differential expression is likely to contribute to physiological changes in lung tissue during r1918 infection.
Expression changes of predicted mRNA targets are inversely correlated with the expression changes of their corresponding microRNAs.
To better understand the roles of these 18 microRNAs which were differentially expressed between r1918 and Tx/91 infections, we investigated the functional associations of the predicted cellular targets potentially regulated by these microRNAs. As microRNAs predominately function as repressors of target gene expression, we therefore focused only on the targets whose expression was inversely correlated with the expression of their corresponding microRNAs. The strategy for selection of inversely correlated targets is represented in Fig. . To make the analysis more straightforward, we directly compared microRNA and cellular gene expression between r1918 and Tx/91 infections to assess expression changes. In the following analyses, an upregulation means that a microRNA or a cellular gene was expressed more abundantly during r1918 infection relative to Tx/91 infection, while a downregulation means that a microRNA or a cellular gene was expressed less abundantly during r1918 infection relative to Tx/91 infection.
FIG. 3. Schematic representation of the strategy for assessing the functional associations of the differentially expressed microRNAs. A rectangular shape represents the type of microRNA, gene, or pathway. A diamond shape represents the type of database. An oval (more ...)
First, to select the differentially expressed microRNA targets, we overlaid the miRanda microRNA target database onto the cellular genes differentially expressed between r1918 and Tx/91 infections. We chose the miRanda database because the miRanda algorithm predicts more microRNA targets by scanning the microRNA binding sites in both the 3′ untranslated region (UTR) and the protein-coding region. Then, we selected the predicted targets, whose expression was inversely correlated with their corresponding microRNAs. Finally, we evaluated the enrichment of inversely correlated targets using a hypergeometric (HG) test. If the enrichment of inversely correlated targets of a particular microRNA was statistically significant, we analyzed the functional associations of these targets using GeneGo, a commercial tool for functional analysis.
Importantly, the enrichment of inversely correlated targets of all 18 microRNAs was statistically significant (P ≤ 0.05) at one time point at the least (Table ). This suggests that the inverse correlation between a particular microRNA shown in Table and its targets was not due to chance but was more likely due to microRNA-mediated regulation. The results obtained from the HG test also indicate an increasing impact mediated by microRNAs on cellular gene expression while the infection progressed. On day 1 p.i, only 3 microRNAs had P values of ≤0.05 among the 18 microRNAs. The number of microRNAs increased to 13 and 12 on days 3 and 5, respectively. Moreover, the HG test results suggested that microRNA-mediated regulation may be time specific. For example, although miR-223 was upregulated during r1918 infection at all three time points, only the HG test results from day 1 had a P of ≤0.05. These results suggest that miR-223-mediated regulation of gene expression was predominately elicited on day 1, an early stage of infection. In contrast, although miR-29a was significantly downregulated in r1918 infection at all three time points, the HG test results from days 3 and 5, but not day 1, had P values of ≤0.05; by inference, miR-29a-mediated regulation on target gene expression occurred predominately at a later stage of infection. Furthermore, although miR-34a was downregulated in r1918 infection at all three time points, the HG test results from only day 3 had a P of ≤0.05, suggesting that miR-34a-mediated regulation on gene expression may be transient. Taken together, our statistical analysis indicates a significant enrichment of the inversely related targets of 18 highly abundant microRNAs during r1918 infection; these microRNAs may regulate target gene expression at different stages of infections.
Summary of 18 microRNAs with enrichment of inversely correlated targets
Inversely correlated microRNA targets are involved in key functions.
To further characterize the roles of these 18 differentially expressed microRNAs during r1918 infection, we interrogated their inversely correlated targets for functional associations. By using the functional analysis tool GeneGo, we were able to assign top gene ontology (GO) terms to the selected microRNAs (Table ).
Summary of functional associations of 18 microRNAs
GO analysis indicated that many of the inversely correlated targets could be related to influenza virus pathogenesis. For example, the inversely correlated targets of four microRNAs are associated with the immune response. The targets of Let-7f are associated with lymphocyte-mediated immune response, while the targets of miR-200a are associated with viral gene replication and the JAK-STAT signaling pathway, which is closely related to the type I IFN-mediated innate immune response. The inversely correlated miR-34a targets are associated with calcium ion homeostasis, which is critical for immune cell activation. In addition, the targets of miR-27a are associated with regulation of the immune response. The inversely correlated targets of three microRNAs, including miR-652, miR-27a, and miR-27b, are associated with apoptosis and cell death. Intriguingly, atypical expression of immune response-related and cell death-related genes was previously shown to be related to the extreme virulence of r1918 in mouse and macaque models of infection (17
). Furthermore, the inversely correlated targets of miR-223, miR-29a, miR-29b, and miR-709 are related to cell division and the cell cycle. Finally, the inversely correlated targets of four microRNAs, including miR-200b, -30a, -30d, and -429, are associated with the regulation of fever, a key sign of illness during influenza virus infection. To better understand how specific microRNAs affected these biological functions, we analyzed two related pathways comprising their inversely correlated targets.
Type I interferon pathway.
Type I IFN plays a key role in the host immune response to virus infection. Influenza virus triggers the activation of the type I IFN signaling pathway, which in turn represses virus replication. However, the activation of IFN also activates the damaging inflammatory response, a contributing factor to the lethal infection of r1918 in vivo
. Our analysis of microRNA targets indicated that the type I IFN pathway was subject to microRNA-mediated regulation, since key genes in this pathway, such as IFNAR1 and STAT2, are direct targets of miR-200a, and their expression was inversely correlated with the expression of this microRNA (Fig. ). Our data suggest that the downregulation of miR-200a in r1918-infected lungs may induce the upregulation of key genes in the type I IFN signaling pathway. Indeed, many IFN-stimulated genes demonstrated increased expression levels in r1918-infected lungs compared with those in the Tx/91-infected lungs (17
FIG. 4. miR-200a is implicated in the type I interferon signaling pathway. (Top) Type I interferon signaling pathway comprised of miR-200a targets; (bottom) expression changes of miR-200a and its targets associated with the type I interferon pathway in r1918-infected (more ...) CREB pathway.
The cyclic AMP (cAMP) responsive element binding protein, CREB, is a transcription factor that regulates the expression of hundreds of genes. CREB-null mice die immediately after birth from respiratory distress (33
), as CREB is involved in critical functions, including T-cell development (33
) and cell survival (6
). The activity of CREB is regulated by multiple upstream pathways, including insulin-like growth factor, Ca2+
, and G protein-coupled receptor signaling pathways. Many key genes in the CREB upstream pathways are miR-233 targets, and concomitant with the strongly increased expression of miR-233 in r1918-infected lungs, the miR-223 targets in the CREB upstream pathways were significantly downregulated (Fig. ). Our data therefore suggest that upregulation of miR-223 may repress the activity of CREB.
FIG. 5. miR-223 is implicated in the CREB signaling pathway. (Top) CREB pathway comprised of miR-223 targets; (bottom) expression changes of miR-223 and its targets associated with the CREB pathway in r1918-infected samples. Red represents microRNA or targets (more ...)
Taken together, our study reveals that influenza virus infection induces changes in the cellular microRNAome and that unique patterns of differential expression of microRNAs may contribute to the extreme virulence of r1918 influenza virus infection by regulating the expression of cellular targets involving immune response and other critical cellular functions.