INTRODUCTION
Type I interferons (alpha and beta interferons [IFN-α/β]) are pleiotropic cytokines that were originally identified for their ability to interfere with viral replication (
1) and are now recognized for their potent immunomodulatory effects (
2–4). Engagement of their cognate heterodimeric receptor, comprised of IFNAR1 and IFNAR2, initiates signaling that culminates in the expression of interferon-stimulated gene (ISG)-associated proteins, critical for antiviral activity. Given the rapid replication of viruses, in the order of several hours (
5–8), the IFN-α/β response must be equally fast and robust, with rapid production of IFN-β and the subsequent activation of signaling cascades downstream of IFNAR1 and IFNAR2 within hours of infection (
9–12). IFNAR activation by IFN results in the induction of ISGs (
13–15). This rapid response initiated by IFN-αs and IFN-β is governed by a series of signaling effectors that are intermediates in the JAK/STAT, mitogen-activated protein kinase (MAPK), and phosphatidylinositol 3-kinase (PI3K)/mammalian target of rapamycin (mTOR) pathways, which coordinately regulate the transcriptional and translational expression of ISGs (
3,
16).
Previously, we and others have shown signaling effectors in the PI3K/mTOR pathway to be critical in governing an effective IFN-α/β-mediated antiviral response. Cells lacking p85α and -β (p85α/β) or Akt1 and -2 (Akt1/2) showed defective antiviral responses and reduced IFN-α/β-inducible ISG protein expression (
17–19). Pharmacological inhibition of PI3K, Akt, or mTOR inhibits IFN-β-mediated suppression of hepatitis C virus (HCV) in a cell-based replicon system (
20). Additionally, cells lacking repressors of IFN-α/β-mediated translational regulation, namely, TSC2 or 4E-BP1, show enhanced responsiveness to IFN-α/β and greater inducible expression of ISG proteins (
21,
22). In mice lacking the translational suppressor 4E-BP1, we also showed enhanced IFN-β antiviral potency in infection with coxsackievirus B3 (CVB3) (
22).
Since protein synthesis consumes a large proportion of cellular ATP, cellular processes are required to maintain energy homeostasis during the induction of translation. AMP-activated protein kinase (AMPK), an important sensor of cellular ATP flux, is invoked to balance energy-consuming pathways, mediated by regulation of mTOR and glucose uptake (
23). Indeed, various growth factors (insulin, platelet-derived growth factor [PDGF], insulin-like growth factor 1 [IGF-1], and vascular endothelial growth factor [VEGF]) and cytokines (interleukin-3 [IL-3], IL-5, IL-6, IL-7, granulocyte-macrophage colony-stimulating factor [GM-CSF], tumor necrosis factor-alpha [TNF-α], and CCL5) that signal through PI3K/Akt/mTOR have been shown to regulate glucose metabolism, specifically through the PI3K/Akt/mTOR pathway (
24–36). Cognizant that IFN-α/β engage PI3K/Akt/mTOR signaling to upregulate protein synthesis, we undertook studies to investigate any influence that IFN-β may exert on glucose metabolism in the context of protection from viral infection. Our data suggest IFN-β mobilization of metabolic events. Given the common signaling effectors between IFN-β and insulin, downstream from their respective cell surface receptors, we examined the effects of metformin, an insulin sensitizer, during an acute viral infection with CVB3. Our data reveal that IFN-β treatment engages mechanisms that meet the energy requirements of cells, thereby enabling a IFN-β-induced antiviral response, and that metformin enhances the antiviral effects of IFN-β.
DISCUSSION
Type I IFNs exert their immunomodulatory influence in a wide variety of cell types and, in the context of virus infections, do so rapidly to inhibit virus replication and limit virus spread. This antiviral activity is mediated by transcriptional and posttranscriptional signaling proteins, including STATs, MAPKs, and PI3K (
16). In recent years, the role of type I IFNs in regulating PI3K/mTOR-mediated posttranscriptional effects has become better defined, with a significant area of focus on translational regulation (
18–21,
37,
51–53). It has become increasingly apparent that mTOR is a central sensor of metabolic stresses and, in addition to translation, regulates processes such as autophagy and lipid and carbohydrate metabolism, thereby maintaining cellular energy homeostasis (
54). Here, we report on the influence of IFN-β on glucose metabolism in the context of virus infection.
Given the high energy demands of IFN-inducible protein synthesis, we anticipated an effect on AMPK activation and cellular ATP synthesis accompanying treatment with IFN-β. Indeed, IFN-β treatment reduced AMPK phosphorylation at Thr172, with a concurrent increase in STAT1 phosphorylation at Tyr701, which is indicative of a IFN-β-inducible cell response. Since AMPK is a sensitive indicator of the cytosolic AMP/ATP ratio (
55), activated by phosphorylation in the presence of low ATP concentrations, we infer that the decrease in Thr172 phosphorylation we identified upon IFN-β treatment is associated with an increase in ATP production. Indeed, IFN-β treatment of MEFs resulted in an increase in ATP production. It is unlikely that IFN-β directly regulates AMPK phosphorylation; rather, it is likely that IFN-β induces an effect which indirectly influences AMPK activation through changes in the AMP/ATP ratio. IFN-β-mediated changes in ATP levels were abrogated in the presence of the nonmetabolizable glucose analog 2-DG. This inhibition of glycolytic-derived ATP provides evidence that IFN-β influences glucose metabolism. In support of this, we demonstrate that IFN-β promotes a dose-dependent uptake of
3H-2-DG by cells.
For IFNs to be most effective as antivirals, it is crucial that cells respond rapidly in terms of producing antiviral proteins that will inhibit viral replication. Accumulating data implicate IFN-α/β in the regulation of translation of host protein synthesis and the corresponding expression of antiviral proteins (
18,
19,
21). Our data suggest that there is a rapid and robust uptake of glucose by cells, within minutes of IFN-β treatment, consistent with meeting the energy demands of protein synthesis. Moreover, the nature of the biphasic response, whereby glucose uptake is initially increased, followed by a suppression, is in agreement with the paradigm of type I IFN-mediated antiproliferative effects (
56–71). Specifically, in uninfected cells, the early translation of antiviral proteins is followed by a progressive shutdown of protein synthesis that would disable cell growth and, upon infection, inhibit viral protein synthesis. Indeed, this biphasic response is consistent with a scenario where virus replicates rapidly and infection spreads. An infected cell produces and secretes IFN-β in response to viral replication prior to viral progeny egress, thereby activating the antiviral response in neighboring uninfected cells (
9–11). Transiently, uninfected cells rapidly increase their metabolism to support the synthesis of antiviral proteins, such as 2′-5′-oligoadenylate synthetase (2′-5′-OAS), protein kinase R (PKR), and RNase L, followed by the subsequent downregulation of metabolism. Upon viral spread, IFN-β-primed cells respond to viral RNA by secreting additional IFN-α, thereby inhibiting further viral replication and spread.
In contrast, when astrocytes are exposed to low concentrations of IFN-α2a, IFN-α2b, or IFN-β (<5 U/ml), no significant changes in glucose consumption are observed over 2 h, and yet chronic exposure to low-dose IFN reduces glucose uptake (
71). This model of low-dose, chronic IFN exposure was intended to reflect the systemically low plasma concentrations of type I IFN in HCV-infected individuals over the duration of a chronic infection. In contrast, our studies reflect a scenario of localized virus infection where cells in close proximity experience high concentrations of IFN-α/β produced by tissue-resident cells or plasmacytoid dendritic cells during an acute immune response to virus infection. In other studies, Navarro et al. examined the effects of type I IFN treatment on glucose metabolism in primary mesenteric and splenic lymphocytes after 48 h and likewise showed a suppression of glucose uptake (
72). Notably, in the earliest IFN experiments of Isaacs and Lindenmann, conducted in chicken embryo cells, they identified a modest IFN-inducible effect on lactate production after 4 h, an indicator of glycolysis (
73).
A number of studies have confirmed the roles of PI3K and Akt signaling in regulating glucose uptake induced by growth factors or cytokines in adipocytes, skeletal muscle cells, and lymphocytes (
24–35). Our strategy was to examine the contribution of different effector intermediates in the PI3K/Akt/mTOR signaling cascade to the IFN-β-inducible regulation of glucose uptake that we observed, specifically, by using MEFs with targeted disruption of certain genes (
Fig. 5). A striking effect was observed in cells null for either p85α/β or Akt1/2. The lack of either of these two signaling effectors was sufficient to completely ablate IFN-β-inducible glucose uptake. Consistent with the negative regulatory role that TSC2 exerts on mTOR activity, IFN-β-inducible glucose uptake in TSC2
−/− cells was unaffected. MEFs lacking mLST8, a nonessential component of mTORC1, exhibited a partial reduction in IFN-β-inducible glucose uptake, suggestive of a role for mTORC1 in regulating glucose uptake. Surprisingly, in cells lacking AMPKα1/2, an upstream negative regulator of mTOR through TSC2 (
74), we observed only a partial reduction in responsiveness to IFN-β-inducible glucose uptake. This may be attributed to the other role that AMPK has in influencing GLUT4 translocation to the cell surface (
49,
75). Consistent with our findings of IFN-β regulation of glucose uptake, the surface expression of GLUT4 was also increased upon treatment with IFN-β. PI3K and Akt activation are associated with GLUT4 translocation to the cell surface (
31,
48,
76), providing further support for a potential mechanism whereby IFN activation of these effectors enhances the expression of glucose transporters required for glucose uptake.
Previous publications have identified that treatment of cells with 2-DG reduces the replication of a variety of viruses, including CVB3 (
77–83). Limiting the energy supplies in an infected cell would affect protein synthesis and the assembly of viral progeny. In contrast, a rapid burst of energy will enable an early robust IFN response, as we show, and yet the biphasic nature of the effect we observe supports the subsequent inhibition of cell growth and viral replication.
Clinical studies have drawn attention to a correlation between insulin and IFN sensitivities in individuals who are infected with hepatitis C virus (
84). The expression levels of TNF-α are often increased in HCV-infected livers. TNF-α upregulates the activity of the phosphatase, PTP-1B, which is responsible for the downregulation of insulin and type I IFN signaling (
85). In the same study, metformin, an inhibitor of PTP-1B, was used effectively to restore insulin and IFN sensitivities in mouse livers expressing high levels of TNF-α. Indeed, metformin is used to treat insulin resistance in patients with type 2 diabetes (
86). Moreover, earlier studies demonstrated the negative regulatory effects of PTP-1B on JAK/STAT signaling (
87–90). We therefore reasoned that metformin may be administered along with IFN-β to enhance antiviral potency during a virus infection. Coxsackieviruses encompass a group of cardiotropic viruses that can cause acute myocarditis and lead to dilated cardiomyopathy (
91). While it is not a standard treatment for viral myocarditis, the administration of IFN-α/β has been shown to improve cardiac function (
92,
93). Interestingly, patient TNF-α expression levels are measured in the serum and heart during acute virus myocarditis, reflective of an inflammatory response to infection (
94–97). Given our data, it is intriguing to speculate that this TNF may influence endogenous type I IFN signaling in the heart, exacerbating infection. In our study, we provide evidence that metformin enhances the antiviral effects of low-dose IFN-β treatment of MEFs challenged with CVB3. Similarly, treating mice with IFN-β and metformin prior to infection with CVB3 enhanced the antiviral effects of IFN-β, most notably reducing viral titers in the hearts, livers, spleens, and sera of infected mice. We speculate that the antiviral effects of metformin alone may be associated with the promotion of endogenous type I IFN activity.
Viewed together, our data provide new evidence that IFN-β modulates glucose metabolism through a PI3K/Akt-dependent mechanism and that this regulation of metabolism appears important for the induction of an effective antiviral response. Additionally, we provide evidence for the application of metformin to enhance the antiviral activity of IFN-β.