Abstract
Sensing invading pathogens early in infection is critical for establishing host defense. Two cytosolic RIG-like RNA helicases, RIG-I and MDA5, are key to type I interferon (IFN) induction in response to viral infection. Mounting evidence suggests that another viral RNA sensor, protein kinase R (PKR), may also be critical for IFN induction during infection, although its exact contribution and mechanism of action are not completely understood. Using PKR-deficient cells, we found that PKR was required for type I IFN induction in response to infection by vaccinia virus lacking the PKR antagonist E3L (VVΔE3L), but not by Sendai virus or influenza A virus lacking the IFN-antagonist NS1 (FluΔNS1). IFN induction required the catalytic activity of PKR, but not the phosphorylation of its principal substrate, eIF2α, or the resulting inhibition of host translation. In the absence of PKR, IRF3 nuclear translocation was impaired in response to MDA5 activators, VVΔE3L and encephalomyocarditis virus, but not during infection with a RIG-I-activating virus. Interestingly, PKR interacted with both RIG-I and MDA5; however, PKR was only required for MDA5-mediated, but not RIG-I-mediated, IFN production. Using an artificially activated form of PKR, we showed that PKR activity alone was sufficient for IFN induction. This effect required MAVS and correlated with IRF3 activation, but no longer required MDA5. Nonetheless, PKR activation during viral infection was enhanced by MDA5, as virus-stimulated catalytic activity was impaired in MDA5-null cells. Taken together, our data describe a critical and non-redundant role for PKR following MDA5, but not RIG-I, activation to mediate MAVS-dependent induction of type I IFN through a kinase-dependent mechanism.