Mitochondria-localized β-actin is essential for priming innate antiviral immune signaling by regulating IRF3 protein stability

Antiviral innate immunity acts as a defensive barrier to viral infection in both immune and nonimmune cells. The viral nucleic acids present in the cytoplasm after viral replication can activate sensors such as the RIG-I-like receptors RIG-I and MDA5 or the cytosolic DNA sensor cGAS to initiate signaling transduction.¹ Activation of this signaling cascade leads to phosphorylation of the protein kinase TBK1, which in turn phosphorylates and activates the transcription factor IRF3 to upregulate type-I interferons.² The production of the type-I interferons IFN-α and IFN-β induces the expression of interferon-stimulated genes (ISGs) that suppress viral replication.³ The stability of IRF3 is essential for sustained antiviral responses,⁴ and altered mitochondrial function can affect antiviral immune signaling.⁵ However, whether mitochondrial defects affect IRF3 protein stability remains unclear. In this study, we found that the mitochondrial pool of β-actin, which is essential for mitochondrial quality and membrane potential, is required for the maintenance of IRF3 stability and the activation of antiviral immune responses.

We initially compared the transcriptome profiles between β-actin^+/+ WT and β-actin^−/− KO mouse embryonic fibroblasts (MEFs). Biological processes, such as the defense response to viruses and immune system processes, were significantly enriched in genes downregulated in the KO cells (Fig. 1a). A large number of ISGs showed reduced expression in KO cells (Fig. 1b). We further analyzed the KEGG pathways and found that the RIG-I-like receptor pathway and the cytosolic DNA sensing pathway were also enriched in genes downregulated in the KO cells (Supplementary information, Fig. S1a). Indeed, multiple RNA and DNA sensors (MDA5, RIG-I, cGAS), signal transducers (MAVS, STING) and interferon response factors (IRFs, Irf1, 6, 7, 8, 9) were also downregulated in KO cells (Fig. 1c–e, Supplementary information Fig. S1b, c). Together, these data demonstrate the systematic downregulation of genes involved in innate antiviral immunity and ISGs in cells lacking β-actin.

Fig. 1.

The mitochondrial pool of β-actin is required for priming innate antiviral immunity by maintaining IRF3 stability. a Gene Ontology (GO) enrichment analysis identified “defense to virus” as enriched in genes downregulated in β-actin KO MEFs. Statistics are in brackets (p value of each GO term). b Heatmap of the relative expression of ISGs (interferon-stimulated genes) in WT (n = 4 biological replicates) and KO (n = 3 biological replicates) MEFs; scale bar: Log2 (counts per million reads). c Scheme of the canonical cytosolic nucleic acid sensing pathway in innate antiviral immunity. Genes labeled red were downregulated in KO cells in comparison to their expression in WT cells. d Heatmap of significantly downregulated (FDR < 0.05) sensors or signal transducers in KO cells. e Heatmap of the relative expression of all Irf genes in WT and KO cells. f qPCR quantification of the Ifna4, Ifnb1, and Cxcl10 (IP10) genes in WT and KO cells with or without poly(dA-dT) stimulation. g qPCR quantification of the Ifna4, Ifnb1, and Cxcl10 (IP10) genes in WT and KO cells with or without poly(I:C) stimulation. Statistics in f and g: mean ± SEM, n = 3 biological replicates, representative of three independent experiments). **P < 0.01, ***P < 0.001 by unpaired two-tailed Student’s t-test. h Heatmap of the expression levels of ISG genes in WT and KO cells with or without poly(dA-dT) stimulation by RNA-seq analysis. Three biological replicates of each condition were analyzed. The violin plot shows the distribution of the net increase in the expression of ISGs in poly(dA-dT)-treated and mock-treated KO and WT cells, respectively. Statistics: The Wilcoxon signed-rank test was used to show difference in the net increase of ISGs between WT and KO cells, ***P < 0.001. i Western blot showing the reduced IRF3 protein level in KO cells in mock- and poly(I:C)-stimulated conditions detected by two different antibodies from Abcam and Cell Signaling Technology (CST). j IRF3 protein decay assay: The same number of WT or KO cells were treated with cycloheximide (150 µg/ml) for 0, 2, 4 and 6 h. IRF3 and β-tubulin protein levels in total cell lysates at each time point were determined by western blotting. The band intensity was normalized to that at the 0-h time point. n = 3 independent experiments. Statistics: mean ± SEM, two-way ANOVA with Bonferroni post-hoc test: **P < 0.01, ***P < 0.001. k Control cells (DMSO-treated) or cells treated with 20 mM β-NAD for 18 h were stained with 0.5 µM TMRE. The mitochondrial membrane potential (MMP) shown by TMRE staining was measured by FACS. Data are the summary of three biological replicates representative of two independent experiments. Statistics: mean ± SEM, one-way ANOVA with Bonferroni post-hoc test: **P < 0.01, ***P < 0.001. l The relative IRF3 protein levels in control or β-NAD-treated cells were determined by western blotting. Data are the summary of three independent experiments. Statistics: mean ± SEM, one-way ANOVA with Bonferroni post-hoc test: *P < 0.05, **P < 0.01. m Schematics of retroviral vector containing GFP (negative control), Actb (β-actin gene), Actb with SV 40 nuclear localization signal (NLS) and Actb with COX4 mitochondrial targeting sequence (MTS). The GFP, Actb, ActbNLS, and ActbMTS constructs were introduced into KO cells by retroviral transduction, and cells expressing the rat CD8a surface transduction marker were sorted by FACS. IRES: internal ribosome entry site. n Relative IRF3 protein levels in KO cells stably expressing GFP (KO::GFP), Actb (KO::Actb), ActbNLS (KO::ActbNLS) and ActbMTS (KO::ActbMTS). Data are the summary of three independent experiments. Statistics: mean ± SEM, one-way ANOVA with Bonferroni post-hoc test: **P < 0.01. o qPCR quantification of the Ifna4, Ifnb1, and Cxcl10 (IP10) genes in mock cells or cells stimulated by poly(I:C). Data following poly(I:C) stimulation are the summary of four independent experiments, and mock data are the summary of two experiments. Statistics: mean ± SEM, one-way ANOVA with Bonferroni post-hoc test: *P < 0.05, **P < 0.01

To confirm defects in the activation of antiviral immune responses, we treated WT and KO cells with the viral nucleoside analogs poly(I:C) and poly(dA-dT). The induction of the type-I interferons Ifna4 and Ifnb1 and the chemokine Cxcl10 was significantly impaired in KO cells following treatment with both nucleoside analogs (Fig. 1f, g). RNA-seq analysis showed that the overall induction of ISGs was largely impaired in KO cells stimulated with poly(dA-dT) (Fig. 1h). Gene ontology enrichment analysis further showed that multiple processes related to defense against viral infection were enriched in genes downregulated in KO cells stimulated by poly(dA-dT) compared to their levels in WT cells (Supplementary information, Fig. S2a). Apart from the upregulation of type-I interferons and ISGs, antiviral signaling activation can also lead to the upregulation of sensors and transcription factors in the antiviral response pathways in a positive feedback loop.⁶ The upregulation of many of these sensors and transcription factors was also impaired in poly(dA-dT)-stimulated KO cells compared to their expression in WT cells (Supplementary information, Fig. S2b–e). Similar results were observed when poly(I:C)-stimulated KO cells were compared to WT cells (Supplementary information, Fig. S3a–d and Fig. S4a). In addition, the induction of proapoptotic genes (Trail, Xaf1, Casp4) known to be upregulated by poly(I:C)⁷ was suppressed in KO cells (Supplementary information, Fig. S4b). Taken together, our data reveal the globally impaired induction of antiviral response genes in KO cells upon the activation of antiviral signaling by viral mimics.

A previous study found that mitochondrial DNA (mtDNA) released into the cytoplasm is involved in priming innate antiviral immunity.⁸ Since our recent study also revealed defects in the mitochondrial membrane potential and mtDNA morphology in cells lacking β-actin,⁹ we wondered whether a change in the cytoplasmic mtDNA level in KO cells can account for the defects in antiviral immune responses. We, therefore, selectively permeabilized the plasma membrane without damaging the mitochondrial and nuclear membranes, purified nuclear DNA and mtDNA in the cytoplasmic fraction, and quantified the nuclear DNA and mtDNA by qPCR (Supplementary information, Fig. S5a, b). However, no significant change in the amount of nuclear DNA and mtDNA in the cytoplasm was observed between WT and KO cells. Therefore, the defect in the antiviral immune pathway in KO cells is not caused by changes in the levels of cytoplasmic mtDNA or nuclear DNA.

IRF3 and TBK1 phosphorylation is a key step in antiviral immune response activation.² We next examined the protein levels and phosphorylation statuses of IRF3 and TBK1 upon poly(I:C) stimulation. Surprisingly, the total IRF3 protein level (detected by two different antibodies) was much lower in KO cells than in WT cells under both mock and stimulated conditions (Fig. 1i). The level of phosphorylated IRF3 was also greatly reduced in stimulated KO cells. The total TBK1 protein level was similar between WT and KO cells, although its phosphorylation was relatively decreased in stimulated KO cells (Fig. 1i). The significantly lower IRF3 protein level was not due to a reduction in the number of IRF3 transcripts in mock and stimulated KO cells (Supplementary information, Fig S6a). This observation was also confirmed in untreated WT and KO cells (Supplementary information, Fig S6b, c), suggesting that the decreased IRF3 level in KO cells is due to protein instability. To examine IRF3 stability, we analyzed the rate of IRF3 protein decay after its translation was blocked with cycloheximide (CHX). Unlike β-tubulin, IRF3 was rapidly degraded after 4–6 h of CHX treatment (Fig. 1j). At each time point, the degree of IRF3 protein degradation was significantly higher in KO cells than that in WT cells, while the β-tubulin level remained unchanged within the 6 h timeframe in both groups of cells (Fig. 1j). Unlike other IRF members that are expressed in only specific cell types, IRF3 is widely expressed and it is the key transcription factor for the activation of antiviral responses.¹⁰ Systematic defects in the activation of antiviral signaling responses in β-actin KO cells can thus be attributed to the instability of IRF3.

The loss of β-actin seemed to affect the IRF3 protein indirectly, as we did not observe an interaction between β-actin and IRF3 (Supplementary information, Fig. S6d). Since we previously observed the severely impaired mitochondrial membrane potential (MMP) in KO cells,⁹ we treated WT and KO cells with β-NAD (β-nicotinamide adenine dinucleotide) to increase the MMP (Fig. 1k). In both WT and KO cells, β-NAD significantly increased IRF3 protein levels, and this effect was not due to increased IRF3 transcription (Fig. 1l, and Supplementary information, Fig. S7a). In contrast, when mitochondria were depolarized by carbonyl cyanide 3-chlorophenylhydrazone, IRF3 protein levels were reduced in WT cells (Supplementary information, Fig. S7b, c). Taken together, these data suggest that the state of the MMP can influence IRF3 stability, providing a possible explanation for the impaired IRF3 stability in KO cells.

To directly show that the effect of β-actin on IRF3 protein levels is mediated through mitochondria, we used the following recently generated cell lines: KO::GFP (KO cells expressing GFP), KO::Actb (KO cells expressing exogenous β-actin), KO::ActbNLS (KO cells expressing exogenous β-actin tagged with a nuclear localization signal), and KO::ActbMTS (KO cells expressing exogenous β-actin tagged with a mitochondrial-targeting signal) (Fig. 1m). Although the expression levels of β-actin constructs reintroduced in KO cells were much lower than those in WT cells, we recently demonstrated that only mitochondria-targeted β-actin can rescue MMP defects in KO cells.⁹ Consistent with this finding, we detected significantly increased IRF3 protein levels in only the KO::ActbMTS cell line, which was not due to altered IRF3 transcription (Fig. 1n and Supplementary information, Fig. S7d). IRF3 also showed increased stability in KO::ActbMTS cells compared with its stability in KO::GFP cells (Supplementary information, Fig. S7e). Furthermore, mitochondria-targeted β-actin increased the induction of the Ifna4, Ifnb1, and Cxcl10 genes upon poly(I:C) stimulation (Fig. 1o). Therefore, our data demonstrate that the mitochondrial pool of β-actin is required for priming the antiviral signaling response by regulating IRF3 stability, which is likely mediated through controlling mitochondrial quality by maintaining the MMP.

In summary, we identified a novel role of mitochondria-targeted β-actin in priming innate antiviral immune signaling. The lack of β-actin not only causes the systematic downregulation of genes involved in antiviral innate immune pathways but also impairs the induction of antiviral response genes upon viral mimic stimulation. This effect seems to result from the instability of the key transcription factor IRF3 in β-actin KO cells. Specifically, we showed that the mitochondrial pool of β-actin, which is essential for mitochondrial quality through maintaining the MMP, is required for the stability of IRF3. Indeed, reintroduction of β-actin into mitochondria of KO cells rescues IRF3 stability and leads to activation of antiviral genes. We, therefore, propose that mitochondria-targeted β-actin is essential for IRF3 protein stability and the effective activation of antiviral immune responses by controlling mitochondrial quality. Future studies should focus on the mechanistic basis of the link between mitochondrial quality and IRF3 stability.

Author contribution

XX and PP conceived the project and designed the experiments. XX, MEC, RM, RJ, and TG performed the experiments. XX and PP analyzed the data and wrote the manuscript.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version of this article (10.1038/s41423-019-0269-2) contains supplementary material.