MicroRNA regulation of IFN-beta protein expression: rapid and sensitive modulation of the innate immune response

Abstract

IFN-beta production is an inaugural event in the innate immune response to viral infections, with relatively small fold changes in IFN-beta expression resulting in the activation of important antiviral signaling cascades. In our rapid SIV/macaque model of HIV encephalitis, the virus enters the CNS within 4 d of infection, accompanied by a marked IFN-beta response that wanes as SIV replication is controlled. The centrality of IFN-beta to the innate antiviral response in the CNS combines with the potential inflammatory damage associated with long-term activation of this pathway to suggest that IFN-beta may be subject to regulatory fine-tuning in addition to well-established transcriptional and message stability mechanisms of regulation. In this paper, we present for the first time evidence that microRNAs (miRNAs), including miR-26a, -34a, -145, and let-7b, may directly regulate IFN-beta in human and macaque cells. In primary primate macrophages, the main cell type implicated in HIV and SIV infection in the CNS, specific miRNAs reduce, whereas miRNA inhibitors enhance, IFN-beta protein production. The potential biologic significance of this regulation is supported by evidence of an apparent negative feedback loop, with increased expression of three IFN-beta-regulating miRNAs by primate macrophages exposed to recombinant IFN-beta or stimulated to produce IFN-beta. Thus, miRNAs may contribute significantly to the regulation of IFN-beta in innate immune responses.

FIGURE 1. Predicted miRNA recognition elements in the IFNβ 3′UTR

miRNA prediction algorithms were used to evaluate the likelihood of miRNA binding in the IFNβ 3′UTR. Over 200 hits (3′ ends represented as red diamonds along the ∼200 nt UTR) were filtered by algorithm overlap and expression analysis, with six miRNAs (including the four forming the duplexes shown, which were later validated) selected for further analysis. The 5′ half of the 3′UTR contains most of the predicted miRNA recognition elements (MREs), including those for the four selected miRNAs, which are further concentrated in two areas as indicated by black-outlined boxes. The 3′ half of the UTR is AU-rich (shaded region) and contains few predicted target sites. MREs for miRs -145, -26a, and -let-7b overlap in the first box, while the predicted -34a MRE starts at nt 50 of the 3′UTR. From human to macaque, only two nucleotide changes are found in the target-rich region (asterisks). Neither affects the seed binding regions though to be needed for miRNA interaction. Duplexes are shown with the target sequence above the miRNA sequence. G:U wobble (‘:’) and Watson-Crick pairing (‘|’) are indicated. A macaque to human transition in the -34a binding site (parentheses) makes the human pairing marginally more favorable.

FIGURE 2. miRNAs directly target sequences in the IFNβ 3′UTR

miRNA mimics reduce luciferase expression from an IFNβ 3′UTR-containing dual firefly/Renilla luciferase reporter vector (A). Normalized luciferase expression is reduced in HEK-293T cells transfected with specific miRNA mimics and equimolar mixtures of the four indicated miRNAs, but not in cells transfected with a control miRNA (CmiR). Each condition is presented relative to a no-miRNA control, set equal to one. Error bars are SEM from three independent experiments. Individual predicted IFNβ miRNA recognition elements cloned downstream of GFP suppressed GFP expression in HEK-293T cells, but three of four mutated MREs relieved this suppression (B). Fluorescence intensity is displayed as the ratio of mutant to wild-type MRE, normalized to red fluorescence (from pdsRed transfection control). Error bars represent SEM.

FIGURE 3. miRNA mimics inhibit IFNβ protein secretion by primary macrophages

A 10 or 40 nM equimolar mixture of four miRNAs (miRs -26a, -34a, -145, and let-7b) or control miRNA was transfected into primary human macrophages, which were then treated with 50 ug/ml poly I:C to stimulate IFNβ production. 24 hours post-treatment, supernatants were collected from no-miRNA controls as well as miRNA- (miR mix) and control miRNA-transfected (CmiR) macrophages; IFNβ levels were measured by ELISA. Levels from control miRNA-treated samples are depicted normalized to poly I:C-treated, no-miRNA controls. Error bars indicate standard deviation.

FIGURE 4. miRNA antagonists relieve native miRNA inhibition of IFNβ secretion

Primary human macrophages were transfected with specific or control antagomiRs, chemically modifed to enhance stability and to hinder recognition by intracellular RNA sensors and subsequent activation of the interferon pathway. Macrophages were treated with poly I:C (50 ug/ml). After 24 hours, supernatants were collected and secreted IFNβ protein levels were measured by ELISA. Results from four independent experiments with macrophages from three donors are shown relative to IFNβ levels of poly I:C treated, no-miRNA controls; error bars indicate standard deviation.

FIGURE 5. IFNβ treatment of primary macrophages modulates three of four IFNβ-regulating miRNAs

Primary macrophages from two pigtail macaque donors were treated with 100U/ml recombinant IFNβ, with RNA collected at 2, 8, and 24 hr post-treatment. miRNA levels for miRNAs -26a, -145, -34a, and let-7b were measured on all samples in triplicate by stem-loop qRT-PCR, including no reverse transcriptase and no template controls. The results were analyzed by ΔΔCt, with normalization to U6 snRNA levels and untreated controls. Error bars are standard deviation.

FIGURE 6. poly I:C treatment of macrophages induces IFNβ and miRNA production consistent with IFNβ-mediated miR upregulation

Primary human macrophages were treated with 50 ug/ml poly I:C. IFNβ response (A) was detectable by ELISA by three hours and increased through 24 hours post-treatment. Limit of detection (about 1 IU/ml) is shown as a line. Results are from two donors, measured in duplicate. miRs -26a, -34a, and let-7b were quantitated by stem-loop qRT-PCR (B). Results are fold change of treated over untreated macrophages at each time point, normalized to U6 snRNA. Error bars indicate standard deviation.