Type I interferons (IFNs; referred to as IFN-α and -β below) are well known for their antiviral and immune-modulating effects. They are crucial for the survival of the host upon various viral infections and are secreted within hours after pathogen contact.[1, 2] All type I IFNs bind to one common type I IFN receptor (IFNAR), which is present on basically all nucleated cells of the body. Under most conditions, positive feedback via the IFNAR is critically needed to induce robust type I IFN responses.[2, 4] Immune-activating properties of type I IFNs can lead to an overall proinflammatory cytokine milieu.[5, 6] They are potent inducers of dendritic cell (DC) maturation and activation,[7, 8] thus contributing to enhanced adaptive immune responses. Moreover, type I IFNs were shown to activate the cytolytic potential of natural killer (NK) cells as well as the expansion and survival of virus-specific T lymphocytes.[10, 11] There is an increasing body of evidence indicating that type I IFNs can also exert anti-inflammatory effects. In a murine collagen-induced model for rheumatoid arthritis, administration of IFN-β decreased inflammation-associated destruction of bone and cartilage. Analysis of the joint tissue revealed decreased levels of proinflammatory tumor necrosis factor and interleukin (IL)-1β and an increase in anti-inflammatory IL-10. Katakura et al. demonstrated that type I IFNs can protect mice from experimental colitis. A strong support for the anti-inflammatory or immunosuppressive potential of type I IFNs is the type I IFN treatment of patients suffering from multiple sclerosis (MS). Long-term treatment with IFN-β in patients with relapsing-remitting MS markedly attenuates the course and severity of the disease and can reduce the frequency of relapses in approximately one third of patients.[14, 15] Several studies point toward immune-modulating capacities of type I IFNs, particularly in the liver. In a hepatitis C virus (HCV) transgenic (Tg) mouse model, an impaired type I IFN signaling was shown to alter the hepatic acute-phase response. Furthermore, Petrasek et al. showed that deficiency in type I IFN signaling enhances hepatic damage in a Toll-like receptor (TLR)9/TLR2 ligand-induced model of liver injury. Moreover, it has been suggested that beneficial effects of IFN-α treatment on liver histology and function in patients with chronic viral hepatitis result, in part, from an anti-inflammatory effect of type I IFNs, in addition to reducing viral titers. Despite this increasing knowledge about anti-inflammatory effects of type I IFNs and the use of IFN-α in the therapy of chronic hepatitis C infection, the mechanisms underlying the protective effects of type I IFNs in the liver are poorly understood. Specifically, it is not clear which cell type needs to be type I IFN triggered to mediate protective and/or immune-modulating effects of type I IFNs.
Cell types and mechanisms involved in type I interferon (IFN)-mediated anti-inflammatory effects are poorly understood. Upon injection of artificial double-stranded RNA (poly(I:C)), we observed severe liver damage in type I IFN-receptor (IFNAR) chain 1-deficient mice, but not in wild-type (WT) controls. Studying mice with conditional IFNAR ablations revealed that IFNAR triggering of myeloid cells is essential to protect mice from poly(I:C)-induced liver damage. Accordingly, in poly(I:C)-treated WT, but not IFNAR-deficient mice, monocytic myeloid-derived suppressor cells (MDSCs) were recruited to the liver. Comparing WT and IFNAR-deficient mice with animals deficient for the IFNAR on myeloid cells only revealed a direct IFNAR-dependent production of the anti-inflammatory cytokine interleukin-1 receptor antagonist (IL-1RA) that could be assigned to liver-infiltrating cells. Upon poly(I:C) treatment, IFNAR-deficient mice displayed both a severe lack of IL-1RA production and an increased production of proinflammatory IL-1β, indicating a severely imbalanced cytokine milieu in the liver in absence of a functional type I IFN system. Depletion of IL-1β or treatment with recombinant IL-1RA both rescued IFNAR-deficient mice from poly(I:C)-induced liver damage, directly linking the deregulated IL-1β and IL-1RA production to liver pathology. Conclusion: Type I IFN signaling protects from severe liver damage by recruitment of monocytic MDSCs and maintaining a balance between IL-1β and IL-1RA production. (Hepatology 2014;59:1555-1563)
enhanced green fluorescent protein
hepatitis C virus
hematoxylin and eosin
type I IFN receptor
IL-1 receptor 1
interleukin-1 receptor antagonist
IFN-stimulated response element
myeloid-derived suppressor cell
retinoic acid-inducible gene 1
specific pathogen free
transforming growth factor beta
Materials and Methods
All mice were bred and kept under specific pathogen-free (SPF) conditions at the Zentrale Tierhaltung of the Paul-Ehrlich-Institut (Langen, Germany). Mouse experimental work was carried out using 8- to 12-week-old mice in compliance with regulations of German animal welfare. Details on the mouse strains used are given in the Supporting Information.
Cell Isolation and Culture
Liver perfusion was performed according to Fang et al. Isolation of peritoneal exudate cells and splenocytes is given in the Supporting Information.
Quantification of Cytokine Production and Alanine Aminotransferase-Activity
To determine serum cytokine levels and serum alanine aminotransferase (ALT)-activity, commercially available kits were used (details are given in the Supportting Information).
For histological analysis, liver sections were fixed in 10% buffered formalin and embedded in paraffin. Tissue sections were affixed to slides, deparaffinized, cut into 2-μm sections, stained with hematoxylin and eosin (H&E; Mayer's Hämatoxylin; Sigma-Aldrich, München, Germany; and Eosin; Merck, Darmstadt, Germany), and examined by light microscopy.
Real-Time Polymerase Chain Reaction
Total RNA was prepared using Trizol (Invitrogen, Karlsruhe, Germany) and the RNeasy Mini Kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions. Further sample preparation and primer sequences are given in the Supporting Information.
In Vivo Applications
Poly(I:C) was purchased from Invivogen (Toulouse, France). It was prepared for injection by resuspending it in sterile, pyrogen-free phosphate-buffered saline (PBS) at a concentration of 2 μg/μL, heating it to 70°C to ensure complete solubilization, and allowing it to cool down to room temperature to ensure proper annealing of double-stranded (ds)RNA. Mice were injected intraperitoneally (IP) with 15 μg of poly(I:C) or intravenously (IV) with 10 μg of poly(I:C)/g body weight (BW) in a maximal volume of 200 μL. Recombinant human (rh)IL-1RA (Anakinra, Kineret; kindly provided by Swedish Orphan Biovitrum) was diluted in PBS and IP injected according to the following scheme: 6 hours before, simultaneously with, and 10 hours after poly(I:C) application at a concentration of 50 or 100 μg/g BW in a maximal volume of 200 μL. Neutralizing anti-IL-1β antibody (Ab; clone B122; eBioscience, San Diego, CA) was applied IP at a concentration of 0.5 μg/μL in a total volume of 200 μL. The application was started 3 hours before poly(I:C) administration and ongoing three times within the subsequent 9 hours.
Abs used for flow cytometry (FCM) are given in the Supporting Information.
Statistical analyses were performed with SAS/STAT software (version 9.3; SAS Institute Inc., Cary, SC), SAS System for Windows (SAS Institute) and Prism 5 for Windows (version 5.04; GraphPad Software, Inc., La Jolla, CA).
For microscopic images, the Axiophot microscope (Carl Zeiss AG, Jena, Germany) was used in combination with Pan-NEOFLUAR lenses (Zeiss) and the AxioCam camera (Zeiss). Acquisition software used was AxioVision (Zeiss).
Type I IFNs Protect From Poly(I:C)-Induced Liver Damage
Upon injection of artificial dsRNA (poly(I:C)), we observed massive changes in liver morphology of IFNAR−/− mice. On the liver surface, white areas were detectable already 10-18 hours after IP injection of IFNAR−/− mice, whereas livers of wild-type (WT) animals showed no changes (Fig. 1A). These macroscopic findings were supported by liver histology, indicating large necrotic areas (corresponding to the white areas in Fig. 1A) with infiltrating cells and hemorrhages in the liver of poly(I:C)-treated IFNAR-deficient mice (Fig. 1B). Accordingly, IFNAR−/− mice showed significantly elevated serum ALT-activity, indicating hepatocyte damage upon poly(I:C) treatment (Fig. 1C). Both WT and IFNAR-deficient mice did not show changes in body temperature or damage of other organs analyzed, such as spleen or kidney (Fig. 1D and data not shown), excluding major systemic inflammatory or degenerative processes induced by poly(I:C) treatment. Thus, type I IFNs protect from poly(I:C)-induced inflammation of the liver, whereas in the absence of a functional type I IFN system, poly(I:C) treatment leads to severe liver damage.
IFNAR Triggering of Myeloid Cells Prevents Poly(I:C)-Induced Liver Damage
To assess which cell type has to be IFNAR triggered to confer protective effects of type I IFNs, we analyzed mice with an IFNAR deletion on selected immune cells (deletion of the IFNAR on T cells [CD4Cre IFNARfl/fl; IFNAR T], B cells [CD19Cre IFNARfl/fl; IFNAR B], myeloid cells [LysMCre IFNARfl/fl; IFNAR Mye], or on CD11c+ conventional DCs [CD11cCre IFNARfl/fl; IFNAR DC], whereas other cells remained IFNAR competent).[10, 19-21] Upon poly(I:C) treatment, mice lacking the IFNAR on myeloid cells (IFNAR Mye) developed severe liver damage, as indicated by elevated serum ALT-activity and changes in liver morphology (Fig. 2A,B) comparable to the liver damage observed in IFNAR−/− mice. Although no macroscopic changes in liver morphology were observed in IFNAR DC mice (data not shown), these mice also showed increased serum ALT-activity upon poly(I:C) treatment. However, serum ALT-activity in IFNAR DC mice was less pronounced, when compared to completely IFNAR-deficient animals or IFNAR Mye mice (Fig. 2A). Poly(I:C) treatment of mice with a selective IFNAR deletion on B and T cells resulted in serum ALT-activity and liver morphology indistinguishable from healthy WT mice (Fig. 2A and data not shown). Of note, upon poly(I:C) treatment, IFN-α responses in IFNAR Mye and IFNAR DC mice were comparable to those in WT animals (Supporting Fig. 1B). Taken together, the presented data indicate that IFNAR triggering of myeloid cells is necessary to protect from poly(I:C)-induced liver damage.
Myeloid Cells With a Monocytic Myeloid-Derived Suppressor Phenotype Are Recruited to the Liver After Poly(I:C) Treatment
Because myeloid cells had to be IFNAR triggered to protect from poly(I:C)-induced liver damage (Fig. 2), we next addressed the question of whether myeloid cells were present in or recruited to the liver, respectively, in WT animals upon poly(I:C) treatment. Analyses of liver perfusates revealed a population of large cells infiltrating the liver of WT, but not IFNAR-deficient animals upon poly(I:C) treatment (Fig. 3A,B). This liver-infiltrating population showed a typical myeloid phenotype, i.e., was positive for CD11b, F4/80, Ly6C, and Gr-1, but negative for Ly6G and other lineage markers, such as CD3, CD19/B220, NK1.1, and CD11c (Fig. 3C and data not shown). The Gr-1+Ly6C+Ly6G− phenotype argues strongly against a classical macrophage population, but is fully compatible with a suppressor macrophage population termed monocytic myeloid-derived suppressor cells (MDSCs).[22-24] Thus, after poly(I:C) treatment of WT animals, myeloid cells with a monocytic MDSC surface phenotype infiltrate the liver in a type I IFN-dependent manner. Poly(I:C) treatment of reporter mice expressing enhanced green fluorescent protein (eGFP) under the control of an IFN-stimulated response element (ISRE-eGFP) confirmed that these liver-infiltrating F4/80+CD11b+ myeloid cells largely expressed eGFP and therefore were indeed type I IFN-triggered (Fig. 3D).
Type I IFNs Induce IL-1 Receptor Antagonist Production by Liver-Infiltrating Myeloid Cells
To analyze how IFNAR-triggered, liver-infiltrating myeloid cells might contribute to the protection of the liver upon poly(I:C) treatment, liver homogenates of treated WT and IFNAR-deficient animals were analyzed for the induction of cytokines, which previously have been related to anti-inflammatory or liver-protective effects.[17, 25-28] No differences in IL-6, transforming growth factor beta (TGF-β), IL-10, and IL-22 production between poly(I:C)-treated WT and IFNAR-deficient animals were detected (Fig. 4A-D). However, we found significantly enhanced levels of the anti-inflammatory IL-1 receptor antagonist (IL-1RA) in the liver of WT animals upon poly(I:C) treatment, when compared to untreated mice and poly(I:C)-treated IFNAR−/− mice (Fig. 4E). Conversely, the levels of proinflammatory IL-1β were low in the liver of WT mice upon poly(I:C) injection, but high in treated IFNAR−/− animals (Fig. 4F). To assign the production of IL-1RA to liver-infiltrating cells, cells were isolated from untreated and poly(I:C)-injected WT and IFNAR−/− mice by liver perfusion. Analysis of the IL-1RA release during subsequent 24-hour in vitro culture clearly indicated that liver-infiltrating cells from poly(I:C)-injected WT mice produced IL-1RA, whereas cells taken from the liver of IFNAR-deficient mice did not (Fig. 4G). Thus, IFNAR signaling is essential for induction of anti-inflammatory IL-1RA by liver-infiltrating cells.
Deregulated IL-1β and IL-1RA Production Upon Poly(I:C) Treatment Mediate Liver Damage in IFNAR−/− Mice
To directly link the deregulated IL-1β and IL-1RA production as observed in IFNAR−/− mice upon poly(I:C) treatment to liver pathology, we either depleted mice of IL-1β or treated them with rIL-1RA, respectively. Treatment of IFNAR−/− mice with Anakinra, a preparation of rhIL-1RA, which was previously shown to be biologically active also in mice, reduced or prevented poly(I:C)-induced liver damage in a dose-dependent manner (Fig. 5A). Likewise, depleting IL-1β using a neutralizing Ab prevented poly(I:C)-induced liver damage efficiently (Fig. 5B). From these data, we conclude that the severe liver damage observed in poly(I:C)-treated IFNAR−/− mice is causally linked to an imbalanced IL-1β and IL-1RA production.
IL-1RA Production Is Dependent on Direct IFNAR Triggering of Myeloid Cells
Our data suggest that direct type I IFN signaling within the recruited myeloid cells is necessary for their release of IL-1RA and the subsequent protection from poly(I:C)-induced liver damage (Figs. 2 and 4G). Hence, in IFNAR Mye mice, we expected that infiltrating myeloid cells should not produce IL-1RA. To prove this experimentally, we analyzed liver-infiltrating cells in poly(I:C)-treated IFNAR Mye mice. Similar amounts of F4/80+CD11b+ liver-infiltrating cells were found in WT and IFNAR Mye mice (Fig. 6A-C). Importantly, unlike their WT counterparts, IFNAR Mye mice failed to produce IL-1RA upon poly(I:C) treatment (Fig. 6E). Thus, direct IFNAR triggering of myeloid cells is indispensable for local IL-1RA production after poly(I:C) treatment of mice.
Upon application of artificial dsRNA (poly(I:C)), we observed severe, acute liver damage in IFNAR−/− mice, which was characterized by enhanced serum ALT-activity and morphological as well as histological changes. In contrast, WT mice did not develop such signs. Of note, we did not observe generalized inflammation (as might be indicated by body temperature changes upon treatment) or obvious damage of any other organs analyzed (Fig. 1). This is in line with data by Hou et al. using a poly(I:C)/D-GalN-induced model of hepatic injury, in which organ damage was restricted to the liver as well. Whether natural RNA, such as that derived from a viral genome or as an intermediate of viral replication, would induce liver damage as well is not yet investigated. Analyzing the cytokine milieu directly in the liver, our data show a disrupted balance of IL-1β/IL-1RA expression in IFNAR−/− mice (Fig. 4). Depleting IL-1β and exogenously adding IL-1RA both completely prevented poly(I:C)-induced liver damage in IFNAR−/− mice (Fig. 5). This finding strongly suggests that increased IL-1β and the lack of IL-1RA cause the liver injury. IL-1β and IL-1RA share a common IL-1 receptor, IL1R1 (IL-1 receptor 1). Binding of IL-1β to IL1R1 mediates inflammation, whereas binding of IL-1RA prevents signaling through its receptor. Hence, excessive IL1R1 signaling is most likely involved in mediating poly(I:C)-induced liver damage in IFNAR-deficient mice. This notion is indirectly supported by the successful treatment of various inflammatory diseases (e.g., rheumatoid arthritis, inflammatory bowel disease, and autoinflammatory syndromes) with recombinant IL-1RA or anti-IL-1β in humans. Guarda et al. showed that IFN-β suppresses both pro-IL-1β availability and IL-1β maturation by inhibiting inflammasome activity. Two other groups reported that poly(I:C) itself can induce IL-1β through activation of the NLRP3 inflammasome in vivo and in vitro.[34, 35] However, Guarda et al. did not find any evidence for direct inflammasome activation by poly(I:C). In turn, poly(I:C) induces type I IFN responses through both the TLR or the retinoic acid-inducible gene 1 (RIG-I)-like helicase pathway.[36, 37] Using bone marrow–derived DC subsets, we identified myeloid DCs as the main source of IFN-α upon poly(I:C) stimulation, induced by cytosolic RIG-I-like helicases rather than the TLR pathway (Supporting Fig. 2). Experiments in a model of TLR9-induced liver injury suggested that the protective effects of type I IFNs observed in this model might be associated with induction of IL-1RA expression. However, using a CpG-oligonucleotide or lipopolysaccharide instead of poly(I:C) did not induce liver damage in IFNAR−/− mice (Supporting Fig. 3). Importantly, none of these earlier studies identified the exact regulatory cell population mediating the protective effect of type I IFN. In the present study, we showed that IFN signaling in myeloid cells is crucial to prevent poly(I:C)-induced liver damage by IL-1RA expression. Mice deficient for the IFNAR selectively on myeloid, LysM+ cells (IFNAR Mye) showed necrotic areas on the liver upon poly(I:C) treatment associated with enhanced serum ALT-activity, indicating a severe, acute liver injury as observed in completely IFNAR-deficient animals (Fig. 2). These data are in line with a number of other studies in which macrophages or other myeloid cells, respectively, turned out to be the crucial cell type that needs to be IFNAR triggered to prevent immunopathology or inflammation. In an experimental autoimmune encephalomyelitis mouse model for MS, Prinz et al. showed that IFNAR triggering of myeloid cells was important for the protective, disease-limiting effects of type I IFNs. Mice deficient for the IFNAR on LysM+ cells showed a disease course comparable to IFNAR−/− mice, which was much more severe than in their WT counterparts. In a lymphocytic choriomeningitis virus infection model, liver-resident macrophages prevented a severe virus-induced immunopathology if cells were IFNAR triggered. Additionally, direct IFNAR triggering of myeloid cells and DCs was shown to be important to prevent lethal liver disease induced by highly cytopathic mouse hepatitis virus. In the present study, mice deficient for the IFNAR on DCs showed enhanced serum ALT-activity upon poly(I:C) treatment as well (Fig. 2A). However, ALT-activity was lower, when compared to IFNAR Mye or complete IFNAR-deficient animals treated with poly(I:C), and no macroscopic or histological changes were observed in IFNAR DC mice. Thus, whether there is a contribution of DCs to liver protection could be a matter of future investigations. Undoubtedly, the protective effect of myeloid cells affected from IFNAR ablation in IFNAR Mye mice is more pronounced. In our study, we identified cells infiltrating the liver of WT mice upon poly(I:C) treatment as being F4/80+CD11b+Gr-1+Ly6C+Ly6G− and negative for lineage markers such as CD11c, CD3, CD19/B220, and NK1.1 (Fig. 3A-C and data not shown). Based on this surface phenotype, the cells can be classified as monocytic MDSCs, which are well known for their immunosuppressive function.[22, 40] Time kinetics analyzing the presence of F4/80+CD11b+ cells in liver perfusates, the peritoneum, spleen, and peripheral blood indicate that these cells did not accumulate in the spleen or became more abundant in the peripheral blood of poly(I:C)-treated WT animals. However, F4/80+CD11b+ cells were present in the peritoneum of untreated mice, but were almost absent or reduced 18-24 hours post poly(I:C) application (Supporting Fig. 4), which is the time point at which they are most abundant in the liver. Whether cells present in the liver of WT mice upon poly(I:C) treatment infiltrate the liver from the peritoneum or proliferate/differentiate within the liver is not yet analyzed. However, it has been reported that during acute and chronic liver injuries, macrophage counts within the liver massively expand as a result of the influx of peripheral cells, but not as a result of the expansion of tissue-resident macrophages. Interestingly, in addition to the IP route, IV poly(I:C) application (inducing systemic IFN-α responses), but not oral poly(I:C) application (not inducing systemic IFN-α responses), resulted in the recruitment of MDSCs to the liver of WT, but not IFNAR−/− mice. Moreover, in WT, but not in IFNAR−/− animals, those infiltrating cells produced IL-1RA, and consequently, IFNAR−/−, but not WT mice developed a severe liver damage upon IV poly(I:C) application (Supporting Fig. 5 and data not shown). In summary, data presented here reveal an important anti-inflammatory and protective role of type I IFNs in poly(I:C)-induced hepatitis and liver damage and suggest the following cascade of events (Supporting Fig. 7): Poly(I:C)-induced type I IFN (Supporting Fig. 1A) triggers a myeloid cell population (phenotypically resembling monocytic MDSCs). Additionally, type I IFNs induce MCP-1 production in the liver (Supporting Fig. 6), which might mediate the recruitment of MDSCs to this organ. The liver-infiltrating MDSCs produce IL-1RA in a type I IFN-dependent manner. At the same time, the production of IL-1β is down-regulated by type I IFNs. In contrast, in the absence of a functional type I IFN system, expression of IL-1RA is abrogated and IL-1β prevails.
The authors thank Dorothea Kreuz, Marion Wingerter, and Stefanie Bauer for their expert technical assistance, Kay-Martin Hanschmann for his statistical analyses, and Eberhard Hildt for critically reading the manuscript. Anakinra was kindly provided by Susan Eckerle.