The PA protein of Influenza A virus (IAV) encoded by segment 3 acts as a specialized RNA endonuclease in the transcription of the viral genome. The same genomic segment encodes for a second shorter protein, termed PA-X, with the first 191 N-terminal aminoacids (aa) identical to PA, but with a completely different C-ter domain of 61 aa, due to a ribosomal frameshifting. In addition, it has been shown that several IAV isolates encode for a naturally truncated PA-X variant, PAXΔC20, missing the last 20 aa. The biochemical properties of PA-X and PAXΔC20 have been poorly investigated so far. Here, we have carried out an enzymatic characterization of PA-X and its naturally deleted form, in comparison with PA from the human IAV strain A/WSN/33 (H1N1). Our results showed, to the best of our knowledge for the first time, that PA-X possesses an endonucleolytic activity. Both PA and PA-X preferentially cut single stranded RNA regions, but with some differences. In addition, we showed that PAXΔC20 has severely reduced nuclease activity. These results point to a previously undetected role of the last C-ter 20 aa for the catalytic activity of PA-X and support distinct roles for these proteins in the viral life cycle.

NMR structure calculation using NOE-derived distance restraints requires a considerable number of assignments of both backbone and sidechains resonances, often difficult or impossible to get for large or complex proteins. Pseudocontact shifts (PCSs) also play a well-established role in NMR protein structure calculation, usually to augment existing structural, mostly NOE-derived, information. Existing refinement protocols using PCSs usually either require a sizeable number of sidechain assignments or are complemented by other experimental restraints. Here, we present an automated iterative procedure to perform backbone protein structure refinements requiring only a limited amount of backbone amide PCSs. Already known structural features from a starting homology model, in this case modules of repeat proteins, are framed into a scaffold that is subsequently refined by experimental PCSs. The method produces reliable indicators that can be monitored to judge about the performance. We applied it to a system in which sidechain assignments are hardly possible, designed Armadillo repeat proteins (dArmRPs), and we calculated the solution NMR structure of YM4A, a dArmRP containing four sequence-identical internal modules, obtaining high convergence to a single structure. We suggest that this approach is particularly useful when approximate folds are known from other techniques, such as X-ray crystallography, while avoiding inherent artefacts due to, for instance, crystal packing.

High protein stability is an important feature for proteins used as therapeutics, as diagnostics, and in basic research. We have previously employed consensus design to engineer optimized Armadillo repeat proteins (ArmRPs) for sequence-specific recognition of linear epitopes with a modular binding mode. These designed ArmRPs (dArmRPs) feature high stability and are composed of M-type internal repeats that are flanked by N-and C-terminal capping repeats that protect the hydrophobic core from solvent exposure. While the overall stability of the designed ArmRPs is remarkably high, subsequent biochemical and biophysical experiments revealed that the N-capping repeat assumes a partially unfolded, solvent-accessible conformation for a small fraction of time that renders it vulnerable to proteolysis and aggregation. To overcome this problem, we have designed new N-caps starting from an M-type internal repeat using the Rosetta software. The superior stability of the computationally refined models was experimentally verified by circular dichroism and nuclear magnetic resonance spectroscopy. A crystal structure of a dArmRP containing the novel N-cap revealed that the enhanced stability correlates with an improved packing of this N-cap onto the hydrophobic core of the dArmRP. Hydrogen exchange experiments further show that the level of local unfolding of the N-cap is reduced by several orders of magnitude, resulting in increased resistance to proteolysis and weakened aggregation. As a first application of the novel N-cap, we determined the solution structure of a dArmRP with four internal repeats, which was previously impeded by the instability of the original N-cap.

In the original publication, the Figure 2 has been published incorrectly. The correct version is given below.

The PA subunit of the RNA dependent RNA polymerase of influenza virus is an endonuclease required for the cap‐snatching mechanism leading to the synthesis of viral mRNA. Recently, the PA‐X protein has been discovered in human influenza, which is translated from an alternative open reading frame within the PA gene coding sequence, resulting from a ribosomal frameshifting. The PA‐X retains the N‐terminal endonucleolytic domain of PA, but has a different and shorter C‐terminal part. It has been proposed that PA‐X degrades cellular mRNA, contributing to the attenuation of viral pathogenesis in the host. In human isolates from the 2009 pandemic H1N1 virus, a naturally truncated form of PA‐X is present. Here we express the recombinant PA and PA‐X proteins, the latter both in its full length and deleted form, from human seasonal influenza virus and show that their endonucleolytic activities have distinct substrate specificities, which depend on the structure of the RNA to be cleaved. Collectively, our results support distinct roles of PA and PA‐X in the viral life cycle.