The NS1 protein of influenza A virus binds not only to poly(A) and a stem-bulge region in U6 small nuclear RNA (snRNA), but also to double-stranded (ds) RNA. Binding assays with NS1 protein mutants established that the previously identified RNA-binding domain of the NS1 protein is required for binding to ds RNA as well as for binding to poly(A) and U6 snRNA. In addition, dsRNA competed with U6 snRNA for binding to the NS1 protein, consistent with both RNAs sharing the same binding site on the protein. As a consequence of its binding to dsRNA, the NS1 protein blocks the activation of the dsRNA-activated protein kinase (PKR) in vitro. This kinase phosphorylates the alpha subunit of eukaryotic translation initiation factor 2 (elF-2 alpha), leading to a decrease in the rate of initiation of translation. Assays using purified PKR and purified elF2 demonstrated that the NS1 protein blocks the dsRNA activation of PKR, and experiments using reticulocyte extracts showed that the NS1 protein blocks the inhibition of translation caused by dsRNA activation of PKR. The implications of these results for control mechanisms occurring in influenza virus-infected cells are discussed.

Eukaryotic viruses have devised numerous strategies to downregulate activity of the interferon-induced, double-stranded (dsRNA)-activated protein kinase (referred to as p68 on the basis of its Mr of 68,000 in human cells). Viruses must exert this control to avoid extensive phosphorylation of the alpha subunit of eukaryotic initiation factor 2 (eIF-2) by p68 and the resultant negative effects on protein synthesis initiation. To begin to define the molecular mechanisms underlying this regulation, we optimized expression of p68 in an in vitro transcription-translation system utilizing the full-length cDNA clone. The in vitro-expressed kinase was autophosphorylated in response to dsRNAs and heparin in a manner similar to that for the native p68 provided that the kinase inhibitor, 2-aminopurine, was present during the in vitro translation reaction. Further, the activated kinase efficiently phosphorylated its natural substrate, the alpha subunit of eIF-2. Binding experiments revealed that the expressed kinase complexed with the dsRNA activator, reovirus dsRNA, as well as the adenovirus-encoded inhibitor, VAI RNA. Interestingly, both the reovirus RNAs and VAI RNA also complexed with protein kinase molecules that lacked the carboxyl terminus and all catalytic domains. Deletion analysis confirmed that the p68 amino terminus contained critical determinants for reovirus dsRNA and VAI RNA binding. Further, reovirus dsRNA efficiently bound to, but failed to activate, p68 kinase molecules containing a single amino acid substitution in the invariant lysine 295 present in catalytic domain H. Taken together, these data demonstrate that this expression system permits a detailed mutagenic analysis of the regions of p68 required for interaction with virus-encoded activators and repressors.

Expression of the double-stranded RNA-activated protein kinase (PKR) is induced by interferons, with PKR activity playing a pivotal role in establishing the interferon-induced antiviral and antiproliferative states. PKR is directly regulated by physical association with the specific inhibitor, P58 IPK , a cellular protein of the tetratricopeptide repeat (TPR) family, and K3L, the product of the corresponding vaccinia virus gene. P58 IPK and K3L repress PKR activation and activity. To investigate the mechanism of P58 IPK -and K3L-mediated PKR inhibition, we have used a combination of in vitro and in vivo binding assays to identify the interactive regions of these proteins. The P58 IPK -interacting site of PKR was mapped to a 52-amino-acid aa segment (aa 244 to 296) spanning the ATP-binding region of the protein kinase catalytic domain. The interaction with PKR did not require the C-terminal DNA-J homology region of P58 IPK but was dependent on the presence of the eukaryotic initiation factor 2-␣ homology region, mapping to the 34 aa within the sixth P58 IPK TPR motif. Consistent with other TPR proteins, P58 IPK formed multimers in vivo: the N-terminal 166 aa were both necessary and sufficient for complex formation. A parallel in vivo analysis to map the K3L-binding region of PKR revealed that like P58 IPK , K3L interacted exclusively with the PKR protein kinase catalytic domain. In contrast, however, the K3L-binding region of PKR was localized to within aa 367 to 551, demonstrating that each inhibitor bound PKR in unique, nonoverlapping domains. These data, taken together, suggest that P58 IPK and K3L may mediate PKR inhibition by distinct mechanisms. Finally, we will propose a model of PKR inhibition in which P58 IPK or a P58 IPK complex binds PKR and interferes with nucleotide binding and autoregulation, while formation of a PKR-K3L complex interferes with active-site function and/or substrate association.

Recently we reported that introduction of catalytically inactive PKR molecules into NIH 3T3 cells causes malignant transformation and the development of tumors in nude mice. We have proposed that PKR may be a tumor suppressor gene possibly because of its translational inhibitory properties. We have now designed and characterized a number of PKR mutants encoding proteins that retain their catalytic competence but are mutated in their regulatory double-stranded RNA (dsRNA) binding domains (RBDs). RNA binding analysis revealed that PKR proteins either lacking or with point mutations in the first RBD (RBD-1) bound negligible amounts of dsRNA activator or adenovirus VAI RNA inhibitor. Despite the lack of binding, such variants remained functionally competent but were much less active than wild-type PKR. PKR variants completely lacking RBD-1 were largely unresponsive to dsRNA in activation assays but could be activated by heparin. To complement these studies, we evaluated the effects of point mutations in RBD-1 or the removal of either RBD-1 or RBD-2 on the proliferation rate of mouse 3T3 cells. We were unsuccessful at isolating stably transformed cells expressing RBD-1 point mutants or RBD-2-minus mutants. In contrast, NIH 3T3 cells, which constitutively expressed PKR proteins that lacked RBD-1, were selected. These cells displayed a transformed phenotype and caused tumors after inoculation in nude mice. Further, levels of endogenous eIF-2 alpha phosphorylation in RBD-1-minus cell lines were reduced, suggesting that such mutants act in a dominant negative manner to inhibit the function of endogenous PKR. These results emphasize the importance of RBD-1 in PKR control of cell growth and provide additional evidence for the critical role played by PKR in the regulation of malignant transformation.

P58IPK , a member of the tetratricopeptide repeat and J-domain protein families, was first recognized for its ability to inhibit the double-stranded RNA-activated protein kinase, PKR. PKR is part of the interferon-induced host defense against viral infection, and downregulates translation initiation via phosphorylation of eukaryotic initiation factor 2 on the ␣-subunit. P58IPK is activated in response to infection by influenza virus, and inhibits PKR through direct protein-protein interaction. Previously, we demonstrated that the molecular chaperone heat shock protein 40 (hsp40) was a negative regulator of P58 IPK . We could now report that influenza virus activates the P58 IPK pathway by promoting the dissociation of hsp40 from P58 IPK during infection. We also found that the P58 IPK -hsp40 association was disrupted during recovery from heat shock, which suggested a regulatory role for P58 IPK in the absence of virus infection. The PKR pathway is even more complex as we show in this report that the molecular chaperone, hsp/Hsc70, was a component of a trimeric complex with hsp40 and P58 IPK . Moreover, like other J-domain proteins, P58IPK stimulated the ATPase activity of Hsc70. Taken together, our data suggest that P58 IPK is a cochaperone, possibly directing hsp/Hsc70 to refold, and thus inhibit kinase function.