Emmanuel Dos Santos Afonso

Genome segmentation facilitates reassortment and rapid evolution of influenza A virus. However, segmentation complicates particle assembly as virions must contain all eight vRNA species to be infectious. Specific packaging signals exist that extend into the coding regions of most if not all segments, but these RNA motifs are poorly defined. We measured codon variability in a large dataset of sequences to identify areas of low nucleotide sequence variation independent of amino acid conservation in each segment. Most clusters of codons showing very little synonymous variation were located at segment termini, consistent with previous experimental data mapping packaging signals. Certain internal regions of conservation, most notably in the PA gene, may however signify previously unidentified functions in the virus genome. To experimentally test the bioinformatics analysis, we introduced synonymous mutations into conserved codons within known packaging signals and measured incorporation of the mutant segment into virus particles. Surprisingly, in most cases, single nucleotide changes dramatically reduced segment packaging. Thus our analysis identifies cis-acting sequences in the influenza virus genome at the nucleotide level. Furthermore, we propose that strain-specific differences exist in certain packaging signals, most notably the haemagglutinin gene; this finding has major implications for the evolution of pandemic viruses.

Strong determinants of the host range of influenza A viruses have been identified on the polymerase complex formed by the PB1, PB2, and PA subunits and on the nucleoprotein (NP). In the present study, molecular mechanisms that may involve these four core proteins and contribute to the restriction of avian influenza virus multiplication in human cells have been investigated. The efficiencies with which the polymerase complexes of a human and an avian influenza virus isolate assemble and interact with the viral NP and cellular RNA polymerase II proteins were compared in mammalian and in avian infected cells. To this end, recombinant influenza viruses expressing either human or avian-derived core proteins with a PB2 protein fused to the One-Strep purification tag at the N or C terminus were generated. Copurification experiments performed on infected cell extracts indicate that the avian-derived polymerase is assembled and interacts physically with the cellular RNA polymerase II at least as efficiently as does the human-derived polymerase in human as well as in avian cells. Restricted growth of the avian isolate in human cells correlates with low levels of the core proteins in infected cell extracts and with poor association of the NP with the polymerase compared to what is observed for the human isolate. The NP-polymerase association is restored by a Glu-to-Lys substitution at residue 627 of PB2. Overall, our data point to viral and cellular factors regulating the NP-polymerase interaction as key determinants of influenza A virus host range. Recombinant viruses expressing a tagged polymerase should prove useful for further studies of the molecular interactions between viral polymerase and host factors during the infection cycle.

The influenza A virus RNA-dependent RNA polymerase complex consists in three subunits, PB2, PB1 and PA, that perform transcription and replication of the viral genome through very distinct mechanisms. Biochemical and structural studies have revealed that the polymerase can adopt multiple conformations and form oligomers. However so far it remained unclear whether the available oligomeric crystal structures represent a functional state of the polymerase. Here we gained new insights into this question, by investigating the incompatibility between non-cognate subunits of influenza polymerase brought together through genetic reassortment. We observed that a 7:1 reassortant virus whose PB2 segment derives from the A/WSN/33 (WSN) virus in an otherwise A/PR/8/34 (PR8) backbone is attenuated, despite a 97% identity between the PR8-PB2 and WSN-PB2 proteins. Independent serial passages led to the selection of phenotypic revertants bearing distinct second-site mutations on PA, PB1 and/or PB2. The constellation of mutations present on one revertant virus was studied extensively using reverse genetics and cell-based reconstitution of the viral polymerase. The PA-E349K mutation appeared to play a major role in correcting the initial defect in replication (cRNA -> vRNA) of the PR8xWSN-PB2 reassortant. Strikingly the PA-E349K mutation, and also the PB2-G74R and PB1-K577G mutations present on other revertants, are located at a dimerization interface of the polymerase. All three restore wild-type-like polymerase activity in a minigenome assay while decreasing the level of polymerase dimerization. Overall, our data show that the polymerase subunits co-evolve to ensure not only optimal inter-subunit interactions within the heterotrimer, but also proper levels of dimerization of the heterotrimer which appears to be essential for efficient viral RNA replication. Our findings point to influenza polymerase dimerization as a feature that is controlled by a complex interplay of genetic determinants, can restrict genetic reassortment, and could become a target for antiviral drug development.