RIG-I mediates nonsegmented negative-sense RNA virus-induced inflammatory immune responses of primary human astrocytes

Abstract

While astrocytes produce key inflammatory mediators following exposure to neurotropic nonsegmented negative-sense RNA viruses such as rabies virus and measles virus, the mechanisms by which resident central nervous system (CNS) cells perceive such viral challenges have not been defined. Recently, several cytosolic DExD/H box RNA helicases including retinoic acid-inducible gene I (RIG-I) have been described that function as intracellular sensors of replicative RNA viruses. Here, we demonstrate that primary human astrocytes constitutively express RIG-I and show that such expression is elevated following exposure to a model neurotropic RNA virus, vesicular stomatitis virus (VSV). Evidence for the functional nature of RIG-I expression in these cells comes from the observation that this molecule associates with its downstream effector molecule, interferon promoter stimulator-1, following VSV infection and from the finding that a specific ligand for RIG-I elicits astrocyte immune responses. Importantly, RIG-I knockdown significantly reduces inflammatory cytokine production by VSV-infected astrocytes and inhibits the production of soluble neurotoxic mediators by these cells. These findings directly implicate RIG-I in the initiation of inflammatory immune responses by human glial cells and provide a potential mechanism underlying the neuronal cell death associated with acute viral CNS infections.

FIGURE 1

Primary human astrocytes constitutively express the cytoplasmic viral RNA sensor, RIG-I, and such expression is elevated following viral challenge. Panel A: Cultured astrocytes (2 × 10⁶) were untreated (0) or infected with VSV (MOI of 0.01, 0.1, 1, and 10). At 4 and 8 hours p.i., the level of RIG-I and G3PDH (G3) mRNA expression was determined. For comparison purposes, RIG-I mRNA levels in a similar number of HeLa cells are shown (+). Representative results of three separate experiments are shown. Panel B: Cells (2 × 10⁶) were untreated (0) or infected with VSV, and RIG-I protein expression determined at 6 and 24 hours p.i. A representative immunoblot of protein isolates prepared at 6 hours p.i. is shown in the upper panel, and the average densitometric values of three separate experiments at 6 and 24 hours p.i. normalized to β-actin expression is shown below. Data is expressed as mean +/− SEM and an asterisk indicates a statistically significant difference from uninfected cells (p < 0.05). For comparison purposes, RIG-I protein expression in a similar number of resting HeLa cells is shown in the representative immunoblot (+). Panel C: Cells (2 × 10⁶) were untreated (0) or infected with VSV. At 12 hours p.i., nuclear and cytoplasmic extracts were prepared and RIG-I expression assessed. For comparison purposes, RIG-I protein expression in HeLa cells is shown (+). A representative immunoblot of three separate experiments is shown.

FIGURE 2

VSV induces the activation of RIG-I associated signaling pathways in primary human astrocytes. Panel A: Cells (2 × 10⁶) were untreated (0) or infected with VSV (MOI of 0.01, 0.1, 1, and 10). At 12 hours p.i., expression of the RIG-I downstream effector molecule IPS-1, was determined in whole cell protein isolates. IPS-1 expression in HeLa cells (+) is shown for comparison purposes. Panel B: Cells (2 × 10⁶) were untreated (0) or infected with VSV (MOI of 1 and 10). At 1 hour p.i., association of RIG-I with IPS-1 was assessed using co-immunoprecipitation techniques. RIG-I content in samples following the pre-clearing step (pc) is shown as a control and RIG-I expression in a similar number of resting HeLa cells is shown (+) for comparison purposes. Representative results are shown for one of three separate experiments. The relative level of RIG-I/IPS-1 complex formation in this experiment was determined by densitometric analysis normalized to total IPS-1 content in the precipitated material and is shown below the immunoblot. Panel C: Cells (2 × 10⁶) were untreated (0) or infected with VSV (MOI of 0.1, 1, and 10). At 1 and 2 hours p.i., nuclear (NUC) and cytoplasmic (CYT) extracts were prepared and RelA expression determined. A representative immunoblot of three separate experiments is shown.

FIGURE 3

5′ triphosphate single stranded RNA, a specific ligand for RIG-I, upregulates RIG-I expression but not IPS-1 in primary human astrocytes. Panel A: Cells (2 × 10⁶) were untreated (0) or exposed to 5′ triphosphate ssRNA (5′ppp-ssRNA: 4 or 8 ug/ml) in the presence of Tfx-20 transfection reagent. At 2 hrs following transfection, RIG-I and G3PDH mRNA levels were assessed. For comparison purposes, RIG-I mRNA levels in a similar number of resting HeLa cells are shown (+). Representative results of three separate experiments are shown. Panel B: Cells (2 × 10⁶) were treated with Tfx-20 alone (0), 5′ triphosphate ssRNA (ppp: 4 or 8 ug/ml), or unphosphorylated RNA (-p: 4 or 8 ug/ml). At 6 hours following transfection, RIG-I protein expression was assessed in whole cell lysates. For comparison purposes, RIG-I protein expression in a similar number of resting HeLa cells is shown (+). Representative results are shown for one of three separate experiments and densitometric analysis of this experiment normalized to β-actin expression is shown below the immunoblot. Panel C: Cells (2 × 10⁶) were untreated (0), or exposed to 5′ triphosphate ssRNA (ppp: 2, 4, or 8 ug/ml) or unphosphorylated ssRNA (-p: 2, 4, or 8 ug/ml) as a control in the presence of Tfx-20, or without the addition of the transfection reagent (−Tfx). At 6 hours following transfection, IPS-1 protein expression was determined in whole cell lysates. Representative results are shown of one of three separate experiments.

FIGURE 4

5′ppp-ssRNA induces NF-kB activation and inflammatory cytokine production in primary human astrocytes. Panel A: Cells (2 × 10⁶) were untreated (0) or exposed to 5′ppp-ssRNA (2 or 4 ug/ml) in the presence of Tfx-20. At 1 hour following transfection, nuclear and cytoplasmic extracts were prepared and RelA levels assessed. Panels B and C: Cells (2 × 10⁶) were exposed to tri-phosphorylated (ppp) or dephosphorylated (-p) recombinant ssRNA (ssRNA; 2, 4, or 8 ug/ml) in the presence of Tfx-20, or were unstimulated in the presence (+) or absence (−) of Tfx-20 alone. At 6 hrs following transfection, levels of IL-6 (Panel B) and TNF-α (Panel C) secretion were determined. Data is expressed as mean +/− SEM (n = 6). An asterisk indicates a statistically significant difference from unstimulated cells, pound symbol indicates a statistically significant difference from the equivalent concentration of dephosphorylated RNA, and asperand indicates a statistically significant difference from the preceding lower dose of each stimulus (p < 0.05).

FIGURE 5

RIG-I knockdown attenuates VSV-induced inflammatory cytokine production by primary human astrocytes. Panel A: To determine knockdown efficiency, cells were untreated (0) or transfected with a combination of three different siRNA oligonucleotide pairs targeting RIG-I and cultured for 24, 48, or 72 hours prior to immunoblot analysis for RIG-I and β-actin. Panel B: To verify the effectiveness of the siRNA oligonucleotide pairs, cells were untreated (0) or transfected with each siRNA pair (1, 2, or 3), or a combination of the three (All), and cultured for 72 hours prior to analysis for RIG-I and β-actin. The relative level of RIG-I in astrocytes transfected with scrambled siRNA (−) is shown for comparison purposes. Panel C: Cells were untreated (0), or transfected with one siRNA pair (pair 3) targeting RIG-I (RIGI siRNA) or scrambled siRNA (Control siRNA). At 72 hours following transfection, cells were infected with VSV (MOI of 0.01, 0.1, 1, and 10) and TNF-α secretion measured at 12 hours p.i. Data is expressed as mean +/− SEM (n = 3). An asterisk indicates a statistically significant difference from unstimulated cells; a pound symbol indicates a statistically significant difference between cells transfected with siRNA directed against RIG-I versus scrambled siRNA (p < 0.05).

FIGURE 6

VSV induces the production of a soluble factor(s) by human astrocytes that elicit neuronal cell death in a RIG-I dependent manner. Panel A: Human astrocytes were untreated (0) or infected with VSV (MOI of 0.01, 0.1, 1, and 10). At 24 hours p.i., filtered conditioned media from these infected cells or media spiked with recombinant TNF-α (63 pg/ml; TNF) was placed on HCN-1A cells. Changes in viability of these human neuronal-like cells were assessed at 4, 8 and 12 hours by quantification of cell attachment and at 12 hours by trypan blue exclusion. Panel B: Human astrocytes were untreated (0) or transfected with siRNA targeting RIG-I (−R) or scrambled siRNA (+R). At 72 hours following the transfection protocol, cells were uninfected or infected with VSV (MOI of 1) and cultured for a further 24 hours. Filtered conditioned media from these cells was placed on HCN-1A cells. As a positive control, media was spiked with recombinant TNF-α (63 pg/ml; TNF) and placed on HCN-1A cells. Changes in viability of these human neuronal-like cells were assessed at 4, 8 and 12 hours by quantification of cell attachment and at 12 hours by trypan blue exclusion. For comparison purposes, cell death of HCN-1A cells was assessed following exposure to conditioned medium from untransfected astrocytes infected with VSV (1). Data is expressed as mean +/− SEM (n = 3). An asterisk indicates a statistically significant difference from unstimulated cells; a pound symbol indicates a statistically significant difference between cells transfected with siRNA directed against RIG-I versus scrambled siRNA (p < 0.05).

FIGURE 7

Proposed model by which astrocytes recognize neurotropic RNA viruses and elicit inflammatory CNS damage. Replicating RNA viruses generate phosphorylated single stranded RNA (5′ppp-ssRNA) that act as a ligand for RIG-I. RIG-I associates with IPS-1, which subsequently activates NF-kB by liberating the RelA subunit. RelA translocates to the nucleus and initiates the production of inflammatory mediators including TNF-α and IL-6. These soluble mediators, and others, could then promote inflammation, increase blood-brain-barrier (BBB) permeability, recruit leukocytes into the CNS, and directly or indirectly initiate neuronal cell damage and/or death.