|Home | About | Journals | Submit | Contact Us | Français|
Astrocytes, together with microglia and macrophages, participate in innate inflammatory responses in the CNS. While inflammatory mediators such as interferons generated by astrocytes may be critical in the defense of the CNS, sustained unopposed cytokine signaling could result in harmful consequences. Interferon regulatory factor 3 (IRF3) is a transcription factor required for IFNβ production and antiviral immunity. Most cells express low levels of IRF3 protein and the transcriptional mechanism that upregulates IRF3 expression is not known. In the current study, we explored the consequence of adenovirus-mediated IRF3 gene transfer (Ad-IRF3) in primary human astrocytes. We show that IRF3 transgene expression suppresses proinflammatory cytokine gene expression upon challenge with IL-1/IFNγ and alters astrocyte activation phenotype from a proinflammatory to an anti-inflammatory one, akin to an M1 to M2 switch in macrophages. This was accompanied by the rescue of neurons from cytokine-induced death in glial-neuronal cocultures. Furthermore, Ad-IRF3 suppressed the expression of microRNA-155 and its star-form partner miR-155*, immunoregulatory miRNAs highly expressed in multiple sclerosis lesions. Astrocyte miR-155/miR155* were induced by cytokines and TLR ligands with a distinct hierarchy, and were involved in proinflammatory cytokine gene induction by targeting suppressor of cytokine signaling 1 (SOCS1), a negative regulator of cytokine signaling and potentially other factors. Our results demonstrate a novel pro-inflammatory role for miR-155/miR-155* in human astrocytes, and suggest that IRF3 can suppress neuroinflammation through regulating immunomodulatory miRNA expression.
Astrocytes in vivo react to pathogen/danger signals by cytoskeletal changes associated with an increase in glial fibrillary acidic protein (GFAP) and process extension, a hallmark of “reactive” astrogliosis (Lee et al, 2005; Carpentier et al, 2008). These morphologic changes are accompanied by alterations in innate inflammatory gene expression. Although astrocytes have traditionally been assigned a trophic role due to the production of neurotrophins and their critical role in regulating extracellular glutamate and potassium concentrations, astrocyte activation has also been linked to inflammation and neurodegeneration. While inflammatory mediators generated by activated astrocytes may be critical in the host defense against pathogens, sustained unopposed proinflammatory cytokine signaling could result in harmful consequences. Therefore, astrocytes also play a dual role depending on their activation phenotype, akin to the concept of classical (M1) and alternative (M2) activation phenotypes in macrophages and microglia (Gordon, 2003; Martinez et al, 2009; Hanisch and Kettenmann, 2007). In the mouse, macrophage activation phenotypes are determined by the expression of characteristic surface receptors and inflammatory molecules. For example, inducible nitric oxide synthase (iNOS) and arginase I are markers of M1 and M2 macrophages, respectively. However, in humans, iNOS is expressed by astrocytes rather than macrophages or microglia (Brosnan et al, 1994; Zhao et al, 2001; Liu et al, 2001).
Astrocytes are also important sources of many proinflammatory cytokines (Dong and Benveniste, 2001; John et al, 2004b). Indeed, stimulation of human or mouse astrocytes with the M1 and Th1 cytokines (IL-1 ± IFNγ) triggers the generation of a whole slew of inflammatory molecules similar to TLR-activated macrophages, with a phenotypic switch from a neurotrophic to a neurotoxic one (Downen et al, 1999; Thornton et al, 2006; Basu et al, 2004). Crucial in the cell signaling pathway underlying this proinflammatory and neurotoxic astrocyte phenotype is the recruitment of MyD88 to the toll-IL-1 receptor (TIR) domain of the IL-1 receptor leading to NF-κB and MAPK activation (Lee et al, 2005; Suh et al, 2009a; Carpentier et al, 2008). In addition, the transcription factor STAT1 binds to the IFNγ-activated sequence (GAS) element of many gene promoters, synergizing with NF-κB and MAPK to maximally induce proinflammatory and neurotoxic gene expression in astrocytes (Hua et al, 2002; Baker et al, 2009).
Interferon regulatory factor 3 (IRF3) is a 53 kDa transcription factor crucial in the TRIF (non-MyD88) pathway of TLR3 and TLR4 signaling (Lin et al, 1998; Sharma et al, 2003; Grandvaux et al, 2002; Fitzgerald et al, 2003). IRF3 plays an indispensible role in innate antiviral immunity. IRF3 is activated by carboxy terminal serine phosphorylation, downstream of TRIF and TANK-binding kinase (TBK). IRF3, in concert with NF-κB and the MAP kinases, transactivates IFNβ (primary response gene), which then acts to amplify the transcription of secondary IFN-stimulated genes (ISGs) in an autocrine and paracrine manner. In addition to TLR3/4, intracellular cytosolic dsRNA sensors RIG-I and related receptors can also activate IRF3 (Hiscott et al, 2006).
Evidence suggests that IRF3 expression might be cell type-dependent, but little information is available on IRF3 expression in normal or pathologic tissues. One recent study reports IRF3 expression in normal human lung tissue and its aberrant expression in lung cancer (Tokunaga et al, 2010). Moreover, IRF3 promoter polymorphisms associated with low IRF3 mRNA expression have been linked to increased incidence of autoimmune diseases (Akahoshi et al, 2008; Gutierrez-Roelens and Lauwerys, 2008). Synthetic cannabinoids, compounds that have shown promise as therapy for neuroinflammatory disorders (Cabral and Griffin-Thomas, 2008), may also produce their beneficial effects in part by regulating IRF3 (Downer et al, 2011). These results together suggest a broader role for IRF3 in autoimmunity, cancer and neurodegenerative diseases, in addition to its well-known role in antiviral immunity.
In the current study, we explored the consequence of adenovirus-mediated IRF3 gene transfer (Ad-IRF3) in human astrocyte cultures. Upon challenge with IL-1/IFNγ, IRF3 transgene becomes phosphorylated and participates in the induction of robust amounts of IFNβ. Most importantly, Ad-IRF3 suppresses astrocyte proinflammatory gene expression and changes the astrocyte activation phenotype from a proinflammatory (which we propose to be termed “A1”) to an anti-inflammatory one (which we propose to be termed “A2”), resembling a macrophage M1 to M2 switch. This was accompanied by the rescue of neurons in the co-culture from cytokine-induced death. Together, our findings support the notion that astrocytes could play an important role in modulating the cytokine balance in the CNS, and we propose that IRF3 gene therapy could predispose glial cells to express an alternatively-activated phenotype and help promote CNS repair.
Human CNS cell cultures were prepared from human fetal tissue. All tissue collection was approved by the Albert Einstein College of Medicine Institutional Review Board. Primary mixed CNS cultures were prepared by enzymatic and mechanical dissociation of the cerebral tissue followed by filtration through nylon meshes. Highly enriched human astrocyte cultures were generated by repeated passages of the mixed CNS cultures, as described previously (Lee et al, 1992; Liu et al, 1996). Mixed neuronal and glial cultures (“mixed cultures”) were generated by replating the initial CNS cell cultures once into 60-mm tissue culture dishes or 96-well tissue culture plates, after collecting microglia (Downen et al, 1999). All cultures were kept as monolayers in DMEM with 5% FBS and antibiotics.
Ad-IRF3 was generated by inserting the full-length IRF3 gene (gift from Dr. J. Hiscott, McGill University, Canada) into the human serotype 5 recombinant adenoviral vector using the Adeno-X Expression System 1, as previously described (Rivieccio et al, 2005). Cultures were inoculated with adenovirus at 500 multiplicity of infection (moi), unless indicated otherwise. After 48 h, cultures were activated with cytokines (IL-1β and IFNγ) (10 ng/ml each) or poly IC (PIC) (10 μg/ml) for an additional 30 min to 72 h, as specified in each individual experiment.
Mixed neuronal and glial cultures at an in vitro age of approximately 3-4 weeks were treated with IL-1β ± IFNγ both at 10 ng/ml in low serum medium (DMEM + 0.5% FCS). Seventy two (72) hours later, neuronal death was assayed by vital dye exclusion, as detailed (Downen et al, 1999).
Western blot analysis was performed as previously described with some modifications (Suh et al, 2007). Thirty to fifty micrograms of protein (cell lysates) was separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and then transferred to polyvinylidene difluoride membrane (Bio-Rad, Hercules, CA). After blocking in 5% milk the blots were incubated with antibodies for 16 h at 4°C. Primary antibodies directed against IRF3 (Abcam, Cambridge MA), p-IRF3 (Upstate Biotechnology, currently Millipore: Billerica, MA), iNOS (Santa Cruz Biotechnology, Santa Cruz, CA) and SOCS1 (Cell Signaling, Danvers, MA) were used. Secondary antibodies were horseradish peroxidase-conjugated anti-mouse or anti-rabbit IgG (Pierce Biotechnology, Rockford, IL), and were used at 1:1,000 for 1 h at RT. Signals were developed using enhanced chemiluminescence (Pierce Biotechnology). All blots were reprobed with β-actin to control for protein loading. Densitometric analysis was performed using NIH ImageJ software.
Quantitative real-time reverse transcription-PCR (Q-PCR) was performed as described previously (Suh et al, 2007), using porphobilinogen deaminase (PBDA) as an endogenous control. Briefly, total RNA was extracted with TRIzol (Invitrogen Life Technologies), following the manufacturer’s instructions. PCR was performed using a SYBR green PCR mix and conducted with the ABI Prism 7900HT (Applied Biosystems). All values were expressed relatively to PBDA. The median value of the replicates for each sample was calculated and expressed as the cycle threshold (CT; cycle number at which each PCR reaches a predetermined fluorescence threshold, set within the linear range of all reactions). ΔCT was calculated as CT of endogenous control gene (PBDA) minus CT of target gene in each sample. The relative amount of target gene expression in each sample was then calculated as 2ΔCT. Fold change was calculated by dividing the value (2ΔCT) of test sample by the value (2ΔCT) of control sample (control = 1).
All reagents including specific and control primers for TaqMan® MicroRNA real-time RT-PCR were purchased from Applied Biosystems and the reactions were set according to the manufacturer’s protocol. Briefly, total RNA was purified by TRIzol (Invitrogen). For each reaction, 10 ng of total RNA was used for reverse transcription using TaqMan® MicroRNA Reverse Transcription Kit and reverse transcription primers for hsa-miR-155, miR-155* and the housekeeping gene RNU44. Real-time PCR quantification was performed using TaqMan® PCR primers and TaqMan Universal PCR Master Mix, No AmpErase UNG on a ABT PRISM 7900HT Fast PCR system (Applied Biosystems). The samples were measured in triplicate for a minimum of 3 different cases. The relative expression was determined as described above for real-time PCR.
Total RNA from astrocytes was extracted using the TRIzol reagent (Invitrogen). RNA concentration was determined on a Nanodrop ND-2000 and RNA quality was determined on a Bioanalyzer nanochip, using Agilent’s RNA Integrity Number (RIN) (Schroeder et al, 2006). MiRNA profiling was performed employing the Human MicroRNA expression profiling assay, V 2.0 panel, (Illumina, Inc.) (Chen et al, 2008). These assays allow high throughput profiling of 1,146 human probes, including known miRNAs (97% of miRNAs described in miRBase database) and putative miRNA sequences. Briefly, 200 ng of total RNA with RIN > 9.0 was used for miRNA profiling following manufacturer’s instructions (Ruiz-Mateos et al, 2008). The PCR-amplified fluorescently labeled probes, obtained for each sample, were hybridized onto 12 Beadchip arrays for 16 h at 4°C in a hybridization oven. The arrays were washed, coated and scanned onto a Beadarray reader. The scanned data was analyzed with the BeadStudio 3.3 software and exported into Microsoft Excel for further statistical analysis.
Anti-miR-155 inhibitor, anti-miR-155* inhibitor and control miR inhibitor (Negative Control #1) were purchased from Applied Biosystems/Ambion and were added to cells using the TransIT-TKO Transfection Reagents (Mirus BioLLC, Madison, WI) following the manufacturer’s protocol.
IFNβ levels were determined with VeriKine-HS Human IFNβ Serum ELISA kit (sensitivity: 2.3-150 pg/ml) from PBL Interferon Source (Piscataway, NJ), according to the manufacturer’s protocol. Luminex Multiplex ELISA was performed with a customized kit according to the manufacturer’s protocol (Millipore Corp. Billerica, MA). IL-8 ELISA was performed using the antibody pair purchased from the R&D Systems and as previously described (Suh et al, 2009b).
The expression of endogenous and transgenic IRF3 protein was examined by immunocytochemistry using a rabbit anti-IRF3 antibody (Abcam) as previously described (Cosenza-Nashat et al, 2011).
For multiple comparisons, one-way ANOVA with Bonferroni post test was performed. For comparison of two groups, Student’s t-test was used. Fold induction or inhibition by Ad-IRF3 from multiple experiments was compared to control (Ad-GFP) using single sample t-test. All statistics were run using the GraphPad Prism 5.0 software. Two-sample t tests were used to identify differentially-expressed genes with fold changes > 2 or fold changes < 0.5. Multiple comparisons were adjusted by false discovery rates. A gene was declared significantly differentially expressed if adjusted p < 0.05.
We first determined the amount of IRF3 protein expression in Ad-IRF3-transduced astrocytes by western blot and immunocytochemistry. At 2 days post-inoculation (p.i.) at 0 to 5,000 multiplicity of infection (moi), Ad-IRF3 dose dependently increased the amount of IRF3 protein expression in astrocytes (Figure 1A). The amount of endogenous IRF3 in control cultures was negligible. By immunocytochemistry, the vast majority of astrocytes were IRF3+ at 500 moi, with variable intensity from cell to cell (Figure 1B). Phosphorylation of IRF3 was confirmed using an antibody against pIRF3 (Ser 396) (Figure 1C), which showed that the IRF3 transgene becomes activated upon stimulation with cytokines (IL-1/IFNγ).
Our previous studies suggested that over-expression of IRF3 gene might modulate microglial gene expression in favor of anti-inflammatory and antiviral responses (Suh et al, 2009b; Lee, 2010). In order to determine whether IRF3 had similar effects on astrocytes, we examined human astrocyte cultures for key inflammatory and antiviral gene expression after transduction with Ad-IRF3. Astrocytes were transduced with 500 moi of Ad-IRF3 or Ad-GFP or medium alone (Ctr) for 48 h, and then treated with IL-1/IFNγ or medium for 24 h. Q-PCR was performed to determine the level of gene expression. Little or no inflammatory gene induction was seen without cytokine treatment (see Figure 3B below, for example), and Ad-IRF3 appeared to have differential effects on cytokine-induced inflammatory gene expression compared to Ad-GFP or Ctr. To confirm these findings, Q-PCR analysis was performed using multiple cultures derived from different astrocyte cases, and the data were grouped into Ad-IRF3-upregulated or Ad-IRF3-downregulated genes, following statistical analysis using single sample t-test. Figure 2A demonstrates Ad-IRF3-upregulated genes (IFNβ, IFIT1, IL-13, IP-10 and IRF7) presented as fold-induction over Ad-GFP in the presence of IL-1/IFNγ in a log 10 scale (0.1 = no induction). Figure 2B shows Ad-IRF3-downregulated genes (IL-8, GROα, IL-1R, TNFα, A20 and iNOS) presented as % inhibition.
The regulation of select astrocyte cytokine gene expression by IRF3 was confirmed by Luminex Multiplex bead assays. These included TNFα and VEGF which showed suppression in cultures treated with Ad-IRF3 compared to Ad-GFP (55 vs. 224 pg/ml; 241 vs. 631 pg/ml, respectively, mean values from duplicate cultures). The results for IFNβ and IL-8 are detailed below. Suppression of iNOS induction by Ad-IRF3 was examined in detail by a combination of Q-PCR, western blot, and the Griess reaction (Figure 3). By all measures, iNOS induction in human fetal astrocytes was inhibited by Ad-IRF3.
Given the consistent and potent suppression of many proinflammatory molecules by Ad-IRF3, we next asked whether Ad-IRF3 expression might rescue neurons from cytokine-induced death in co-cultures of neurons and glia. Cultures were infected with adenovirus for 48 h, and then further treated with IL-1/IFNγ for an additional 72 h. Neurotoxicity was assessed by vital dye exclusion, as described (Downen et al, 1999). The results show that in the absence of Ad-IRF3, large numbers of neurons underwent apoptosis following cytokine treatment (Figure_4A and C). By contrast, neuronal death was greatly reduced by Ad-IRF3 (Figure 4B and C). Neurotoxicity data compiled from multiple independent experiments using different brain cases are presented in Figure 4C and show that Ad-IRF3 confers consistent and statistically significant neuroprotection in cytokine-treated CNS cultures. Furthermore, the expected astrocyte cell shape change from round/polygonal (control) to process-bearing (“reactive”) type, following cytokine treatment (Liu et al, 1994; John et al, 2004a), was also inhibited by Ad-IRF3 (Figure 4A and B). The inhibition of cytokine-induced astrocyte morphology change by Ad-IRF3 was confirmed in pure astrocyte cultures stained for astrocyte-specific marker, glial fibrillary acidic protein (GFAP) (Figure 4 D, E and F).
In order to assess whether neuroprotection occurs through glial or neuronal transduction by Ad-IRF3, we examined mixed cultures for IRF3 transgene expression and the neuronal marker MAP2 by double-label immunofluorescence (Figure 4G, H and I). The results show that adenovirus-mediated gene transfer (green) was limited to glial cells in these cultures. Since neurons were not transduced by adenovirus, these results indicate that the effect of IRF3 on neuroprotection was indirect through modulation of glial (inflammatory) activation.
As IFNβ is a crucial M2 cytokine in macrophages and is also the primary response gene in the IRF3 signaling pathway (Suh et al, 2009a; Suh et al, 2007; Kawai and Akira, 2007), we next examined the role of Ad-IRF3 in IFNß production, using a sensitive ELISA with a low limit of detection ~ 2.3 pg/ml. We first compared IL-1/IFNγ and PIC (TLR3 ligand, an endogenous activator of IRF3) in their induction of IFNβ. LPS (TLR4 ligand) was not included in this study as human astrocytes respond minimally to LPS (Lee et al, 1993; Lee et al, 1995). Figure 5A shows a representative experiment. As expected, PIC induced IFNβ in astrocytes, whereas IFNβ protein production was minimal in response to IL-1/IFNγ (Figure 5A). These results indicate that in vitro human astrocytes do express IRF3 which can be activated by PIC. Notably, Ad-IRF3 significantly increased IFNß production in response to both stimuli, by ~ 3-fold in PIC cultures and by ~ 40-fold in IL-1/IFNγ cultures (Figure 5B). These results indicate that IRF3 transgene expression translated into an increase of IRF3 signaling and IRF3-dependent gene expression, when cells were exposed to inflammatory stimuli.
We also compared the production of IL-8 protein by ELISA in astrocyte cultures stimulated with IL-1/IFNγ or PIC for 24 h. In contrast to IFNβ, IL-8 production was much more robustly induced by IL-1/IFNγ than by PIC. Furthermore, Ad-IRF3 suppressed IL-8 production in both conditions. Together, these ELISA results demonstrate that proinflammatory cytokines (IL-1/IFNγ) and the TLR3 ligand (PIC) differentially induce A1 and A2 cytokines in astrocytes and further confirm the findings with Q-PCR that IRF3 transgene differentially modulates astrocyte cytokine gene expression.
Recent studies have shown that microRNAs (miRNAs) are involved in the regulation of inflammation and innate immune responses (Baltimore et al, 2008). Furthermore, several miRNAs including miR-155 have been detected in multiple sclerosis lesions and in cytokine-treated astrocyte cultures (Junker et al, 2009). We therefore explored the possibility that Ad-IRF3 might modulate cytokine gene induction by regulating miRNA(s). We first profiled miRNA expression in IL-1/IFNγ-treated astrocytes (vs. untreated astrocytes) employing the Human MicroRNA expression profiling assay from Illumina. Four different astrocyte cases were analyzed. A total of 13 miRNAs were found to be significantly modulated by IL-1/IFNγ (11 upregulated and 2 down-regulated), by comparison with untreated astrocytes (Table 1). Importantly, miR-155 was one of the most significantly upregulated miRNAs in IL-1/IFNγ-activated astrocytes. The star-form partner (miR-155*) was also increased (see Discussion). Other significantly upregulated miRNAs included miR-27a*, miR-23a*, miR-147, miR-147b and miR-146a. Of these, miR-155, miR-23a and miR-147b have been reported previously in cytokine-activated human astrocytes and in multiple sclerosis lesions (Junker et al, 2009). Two miRNAs were significantly downregulated; miR-296-3p and miR-767-3p.
We validated the results of miR-155 and miR-155* using TaqMan® miRNA Q-PCR assays, and also determined the relative potency of cytokines and TLR ligands in the induction of astrocyte miR-155 and miR-155*. Data pooled from three separate astrocyte cases are shown in Figure 6. Our data demonstrate an increase in the expression of miR-155 after treatment by IL-1 or TNFα but not by IFNγ. IFNγ increased the level of miR-155 induced by IL-1. PIC induced less miR-155 than IL-1 or TNFα (Figure 6A). Although the amount was much higher, the pattern of miR-155* induction by cytokines and TLR ligand was identical to that of miR-155 (Figure 6B). These data indicate that miR-155 and miR-155* are co-regulated in astrocytes. Their induction pattern also suggests that they are NF-κB-dependent miRNAs. Importantly, the expression of both miR-155 and miR-155* was decreased in astrocytes transduced with Ad-IRF3 (Figure 6C, D), which suggested that Ad-IRF3 controls inflammatory gene expression in part through modulation of miR-155 and miR-155* expression (see below).
miR-155 has been shown to modulate immune response gene expression in macrophages, playing both proinflammatory and anti-inflammatory roles (Androulidaki et al, 2009; Louafi et al, 2010). Since no information is available on the role of miR-155 (or miR-155*) in astrocyte cytokine expression, we examined their role using specific oligonucleotide inhibitors. Astrocytes were treated with anti-miR-155, anti-miR-155* or control anti-miR for 2 days, then stimulated with IL-1/IFNγ and cytokine expression determined by Q-PCR. Pooled data from 3 separate astrocyte cases are shown in Figure 7. Using TaqMan® Q-PCR, we demonstrated that anti-miRNA inhibitors were highly effective in reducing miR-155 and miR-155* expression in astrocytes (~ 100% inhibition) (Figure 7A, B). Q-PCR analysis of astrocyte inflammatory gene expression showed that both anti-miR-155 and anti-miR-155* suppressed astrocyte proinflammatory gene expression (TNFα, IL-6 and IL-8) induced by IL-1/IFNγ (Figure 7C, D, E). These results indicate that miR-155 and it star-form partner miR-155* play a positive role in the induction of proinflammatory (A1) genes by IL-1/IFNγ in astrocytes.
A single miRNA has on average ~ 100 mRNA targets, and many miR-155 targets have been found in cell types ranging from T cells to B cells and to macrophages (O’Connell et al, 2009; Tili et al, 2009; Androulidaki et al, 2009; Tsitsiou and Lindsay, 2009; O’Connell et al, 2009). These include Src homology-2 domain-containing inositol 5-phosphatase 1 (SHIP1), SOCS1, the transcription factor PU.1 and activation-induced cytidine deaminase (AID). Astrocytes belong to the neuroepithelial cell lineage and do not express many of the hematopoietic lineage-specific proteins. Therefore, we examined SOCS1 as a potential miR-155 target in astrocytes. Q-PCR and western blot analyses were performed to determine the level of SOCS1 expression in the presence of specific anti-miR155 or control anti-miR. In addition, the effect of Ad-IRF3 on SOCS1 expression was examined. Results shown in Figure 8 demonstrate that astrocyte SOCS1 induced by IL-1/IFNγ was increased in the presence of anti-miR155, and Ad-IRF3 increased the expression of SOCS1. Together, these results suggest that Ad-IRF3 reduces astrocyte proinflammatory cytokine gene expression in part by suppressing miR-155, which normally acts to increase proinflammatory gene expression by targeting SOCS1 and potentially other genes.
A schematic of our results and hypothesis are shown in Figure 9. We hypothesize that astrocyte IRF3 gene transfer can change neuroinflammatory responses by switching the astrocyte activation phenotype from a classic proinflammatory one (A1) to an alternative, anti-inflammatory one (A2). Astrocyte inflammatory activation by products of activated macrophages and T cells such as IL-1/IFNγ results in activation of MyD88-dependent cell signaling with induction of NF-κB-dependent proinflammatory genes such as TNFα and iNOS. IRF3 is not activated under these circumstances and there is little or no induction of IFNβ (A1 >> A2). IRF3 gene therapy can reverse the CNS cytokine environment by suppressing astrocyte NF-κB signaling and miR-155 expression thereby increasing miR-155 target genes such as SOCS1. SOCS1 suppresses the expression of astrocyte proinflammatory genes. IRF3 transgene expression in the presence of IL-1/IFNγ robustly enhances the production of IFNβ from astrocytes by increasing the amount of activated IRF3. Together, IRF3 gene transfer can result in suppression of inflammation resulting in neuroprotection (A1 << A2).
This study was designed to investigate the therapeutic potential of IRF3 overexpression during inflammation. Data in primary human astrocyte and mixed neuronal and glial cultures showed that adenovirus-mediated overexpression of IRF3 changes the cytokine production profile from proinflammatory (A1) to anti-inflammatory (A2), associated with neuroprotection. Since neurons were not transduced with adenovirus in these cultures, the neurotrophic effect of IRF3 was strictly mediated by glial (mostly astrocytic) cells. Ad-IRF3-upregulated genes included IFNβ, IFN-induced protein with tetratricopeptide repeats 1 (IFIT1, aka ISG56, often the highest induced) and IP-10, all known IRF3 target genes (Grandvaux et al, 2002), the transcription factor IRF7 which synergizes with IRF3 in the induction of IFNα and ISGs, and the Th2 cytokine IL-13 (Figure 2).
Unexpectedly, the expression of many proinflammatory genes was suppressed by IRF3 and these included iNOS, TNFα, IL-1 receptor (IL-1RI), IL-8, CXCL1 (GROα), and A20. iNOS and TNFα induction in human astrocytes requires stimulation with IL-1, with IFNγ providing synergistic effects due to the presence of IFNγ-activated sequence (GAS) in their promoters (Hua et al, 2002; Hua and Lee, 2000; McManus et al, 2000). We have shown previously that IFNβ suppresses these genes by preventing STAT1 binding to GAS sequences (Hua et al, 2002). However, Ad-IRF3-suppressed astrocyte genes also included chemokine genes such as IL-8 and GROα that bear no known GAS or IFN-stimulated response element (ISRE). In addition, A20, an NF-κB-dependent gene involved in feedback inhibition of macrophage innate immunity (Turer et al, 2008; Lin et al, 2006), was also suppressed by Ad-IRF3. A20 mRNA suppression in IRF3-overexpressing human cell lines has been previously observed, in direct (inverse) proportionality to the amount of cellular IRF3 expression (Elco et al, 2005). Furthermore, the IL-1 receptor (IL-1RI) expression was also downregulated by Ad-IRF3, suggesting that receptor downregulation may also participate in the suppression of IL-1 (NF-κB) signaling by IRF3. These results together suggest that the mechanism by which Ad-IRF3 suppresses proinflammatory genes in astrocytes is probably multifaceted and not simply explained by over-production of anti-inflammatory cytokines such as IFNβ.
We also find that IRF3 overexpression is associated with a change in balance in M1 and M2 cytokines in microglia (for example, IL-1 receptor antagonist > IL-1)1. This is highly significant since IL-1 is a major proinflammatory cytokine expressed in several neurodegenerative disorders, and also is a prime inflammatory activator of astrocytes that acts through the MyD88 pathway (Lee, 2010; Burger et al, 2009; Simi et al, 2007). IL-1 and TLRs share the same receptor component (the toll/IL-1 receptor “TIR” domain) that signals through the MyD88 pathway or the non-MyD88 (TRIF) pathway. The TRIF pathway is triggered exclusively by TLR3 or TLR4 ligation and converges on the activation of IRF3. Although IL-1 is capable of activating IRF3 in astrocytes (Rivieccio et al, 2005), a direct comparison with PIC in this study shows that IL-1/IFNγ induces very little IFNβ expression (Figure 5). Consistent with these findings, our previous studies have shown that human astrocytes activated with PIC conferred effective antiviral immunity against HIV and HCMV in an IRF3-dependent manner, while IL-1 did not (Suh et al, 2007). Importantly, we observe robust increase in IFNβ production by IRF3 transduction (+ IL-1/IFNγ), resembling PIC-activated astrocytes. These results suggest that while cytokines alone do not elicit significant IRF3-dependent gene expression, they do so in the presence of increased amounts of IRF3 protein, as can be induced therapeutically by viral vector-mediated gene transfer.
MicroRNAs (miRNAs) are small non-coding RNAs important in regulation of gene expression and immune responses. Among these, miR-155 has emerged as a multifunctional miRNA involved in the regulation of inflammation and antiviral responses in macrophages (Baltimore et al, 2008; O’Connell et al, 2007). In addition, miR-155 has been shown to be highly expressed in reactive astrocytes in multiple sclerosis lesions (Junker et al, 2009), and furthermore, miR-155-deficient mice are resistant to the development of experimental autoimmune encephalitis, an animal model for multiple sclerosis (O’Connell et al, 2010). The positive role of miR-155 in autoimmunity has been largely attributed to its ability to drive Th17 differentiation of T cells (O’Connell et al, 2010), and its role in endogenous CNS cells such as astrocytes have not been considered. Our microarray profiling of IL-1/IFNγ-activated astrocytes demonstrates that several miRNAs are significantly upregulated, confirming previous results in cytokine-activated human astrocytes (Junker et al, 2009). These include miR-155, miR-147, miR-147b and miR-146a, miRNAs that are shown to be induced in activated macrophages and involved in immune responses (Taganov et al, 2006; Baltimore et al, 2008). Our study using a specific miR-155 inhibitor oligonucleotide showed that miR-155 is involved in astrocyte proinflammatory gene expression. Interestingly, we find that the star-form partner miR-155* is the most highly induced miRNA in cytokine-activated astrocytes. The star-form partner miRNAs are derived from the same precursor as a passing strand but their roles have not been systemically studied (Yang et al, 2011). Although a recent study reported opposite roles that miR-155 and miR-155* play in dendritic cell cytokine production (Zhou et al, 2010), our own study of astrocytes show that miR-155 and miR-155* are co-regulated by cytokines and TLR ligand, and that they have the same proinflammatory function.
Our results in astrocytes agree with the proinflammatory role of miR-155 generally reported in TLR-activated macrophages (Androulidaki et al, 2009). We show that miR-155 plays an “M1-like” role in astrocytes (which we propose to be termed “A1” for astrocytes), and that the immune modulatory effect of IRF3 transgene may in part be mediated through inhibition of miR-155 transcription, thereby suppressing proinflammatory cytokine production, while preserving anti-inflammatory cytokine production (this astrocyte phenotype we propose to be termed “A2”). One of the many discovered targets of miR-155 is SOCS1. SOCS1 is an important negative regulator of cytokine and TLR signaling (Baker et al, 2009). The classic function of SOCS1 is to inhibit IFN signaling through interaction with p-JAK, thereby limiting activation of STAT proteins. SOCS1 can target additional signaling components such as NF-κB p65 (ibid). In astrocytes, we find that SOCS1 is induced by IL-1/IFNγ and this is further increased by anti-miR155 inhibitor. Furthermore, Ad-IRF3 increases SOCS1 expression, while suppressing miR-155 (and miR-155*). Together, these results demonstrate that IRF3 transgene reduces the “A1” gene expression by suppressing miR-155, which, in turn, increases the expression of miR-155 target genes such as SOCS1, a negative regulator of cytokine signaling (Figure 9).
The serendipitous discovery that overexpression of IRF3 suppresses some of the key proinflammatory molecules is particularly important to our understanding of glial biology. We believe IRF3 gene transfer will predispose glial cells to become an “A2” (and M2) phenotype by coordinately modulating the expression of various gene groups upon exposure to proinflammatory stimuli (such as IL-1/IFNγ). Since transduced IRF3 protein is dormant, there will be fewer undesirable effects originating from the transgene expression per se. The activating signals could be provided by intercurrent systemic infections or stress, conditions known to trigger CNS inflammation These results provide rationale for IRF3 gene therapy for CNS diseases.
This study was supported by NIH RO1 MH55477, T32 NS007098, KO1 MH084705, and the Einstein CFAR grant P30 AI051519. We thank Dr Brad Poulos of the Einstein Human Fetal Tissue Repository for tissue and Drs Meng-Liang Zhao and Namjong Choi for providing assistance with cell culture and cell staining. Dr Liise-anne Pirofski and Sarah Weber provided assistance with the Luminex Multiplex assays. We are grateful to Dr Yungtai Lo for statistical analyses and Dr Celia F Brosnan for critical reading of the manuscript and the proposal of the terms, A1 and A2.
1The detailed aspects of IRF3 transgene effects on microglial activation are the subject of a separate paper.
The authors declare no conflict of interest.