Here, using recombinant viruses generated with reverse genetics, the impact of three amino acid substitutions in PB2 on S-OIV replication, pathogenesis in animal models, and transmission was evaluated. The E627K and D701N substitutions were previously shown to be important for adaptation of avian influenza viruses to the human host, and the E627K and E677G substitutions were detected recently in patients with S-OIV infection. In vitro, the E627K and D701N amino acid substitutions in PB2 of S-OIV resulted in enhanced minigenome reporter expression compared to the wild-type PB2, but not in enhanced replication kinetics in MDCK cells. This apparent discrepancy between the minigenome reporter assay and the in vitro virus replication assay may be related to the different parameters under investigation; whereas mRNA synthesis and reporter protein expression is the primary readout in minigenome assays, for efficient virus replication, a proper balance between the syntheses of cRNA, vRNA, and mRNA is required.
In a mouse model, slightly elevated levels of virus replication were detected at 3 days p.i. in the lungs of animals inoculated with NL602/PB2-627K, and these animals showed a 2-day delay in regaining bodyweight. Mice inoculated with NL602/PB2-677G had slightly elevated levels of virus replication in the lungs at both 3 and 6 days p.i., which was not reflected by changes in bodyweight. In the ferret model, animals inoculated with the NL602/PB2-701N virus lost significantly more bodyweight than ferrets inoculated with the other viruses. Although this could suggest more severe disease, it should be noted that variation in bodyweight loss was relatively high between individual ferrets (Fig. ). The groups of ferrets were too small to detect statistically significant differences after 3 days p.i., when only 3 animals remained in the experiment. Major differences between the four groups of ferrets were not observed for other clinical parameters, such as lethargy, sneezing, ruffled fur, interest in food, and runny nose. This was in agreement with the virus titers in the respiratory tracts of inoculated animals, which were largely comparable for all groups. Minor variations in virus shedding, however, were also observed in ferrets; in throat swab samples, higher levels of virus shedding were detected for each of the PB2 mutants at 5 and 6 days p.i., and virus shedding from the throat was higher for animals inoculated with NL602/PB2-677G throughout the time course of this experiment.
It is unclear which of these observations has implications for infections of humans with S-OIVs. It is possible that S-OIVs with mutations in PB2 would also result in differences in virus replication in the human upper or lower respiratory tract and perhaps in differences in pathogenesis and transmission. However, from the accumulated data, we conclude that none of the tested mutations in PB2 had a major impact on the virulence of A/Netherlands/602/2009 in mice and ferrets. Certainly, for E627K and D701N, this was in contrast to our expectations, because a large impact of these mutations was observed in the context of other influenza viruses in mice and ferrets. In mice infected with HPAI A/Netherlands/219/2003 (H7N7) virus, PB2 E627K was the main determinant of pathogenicity, related to >1,000-fold differences in lung virus titers (21
). Adaptation of A/Equine/London/1416/73 (H7N7) to mice resulted, among others, in an E627K substitution in PB2, leading to 1,000-fold-increased virulence of the virus as measured by the dose lethal to 50% of infected mice (MLD50
). The PB2 E627K mutation was also the main determinant of differences in pathogenicity of 1997 HPAI H5N1 viruses, in which the MLD50
changed >1,000-fold upon introduction of the mutation (11
). In HPAI H5N1 viruses isolated from ducks in China, D701N was the main determinant of pathogenesis in the mouse model (15
). The pathogenicity of a variant of A/seal/Massachusetts/1/1980 (H7N7) that is highly pathogenic to mice, SC35, was also determined in part by D701N in PB2 (8
). Thus, the mouse model seems appropriate for testing the effects of both E627K and D701N in PB2, yet in the context of A/Netherlands/602/2009 (S-OIV), there was no such marked effect of these mutations on pathogenesis.
Ferrets are generally considered a more suitable animal model for influenza A virus infections in humans because they are susceptible to natural infection and develop respiratory disease and lung pathology similar to those of humans when suffering from seasonal, avian, or pandemic influenza virus infections (16
). Ferrets have also been used previously to map determinants of influenza virus pathogenicity and transmission to the PB2 gene (24
). However, there was no major effect on pathogenesis and transmission of A/Netherlands/602/2009 from any of the three PB2 mutations tested. Transmission was detected in 3 or 4 out of 4 animals that were placed adjacent to animals inoculated with viruses containing the PB2 mutations. For comparison, NL602 was transmitted to 4 out of 4 animals. The ferret transmission model was designed as a qualitative model for virus transmission via aerosols or respiratory droplets. With the limited number of animals in this experiment, quantitative information on virus transmission, with statistical support, could not be obtained. Therefore, we merely conclude from the experiment that the S-OIVs with and without mutations in PB2 were all transmitted via aerosols or respiratory droplets.
The polymerase complex of the S-OIV originated from triple-reassortant swine viruses, the PB2 gene of which is of avian origin and entered pigs around 1998 (9
). The fact that this avian-origin PB2 gene did not significantly benefit from the substitutions E627K and D701N for enhanced replication in mammals suggests that other mutations in the polymerase complex can compensate. It will be important to map which genetic changes in the polymerase complex of S-OIV facilitate efficient replication in mammalian cells in the absence of E627K and D701N in PB2, as this may reveal yet-unidentified virulence or host adaptation markers of influenza A virus. A recent study suggested a role for a PB2 SR polymorphism present in S-OIV, since mutation of this SR polymorphism to the consensus G590Q591 sequence resulted in reduced polymerase activity in human cells in vitro
). Animal experiments are needed to elucidate the role of SR polymorphisms in vivo
Although the PB2 mutations tested did not increase the pathogenicity of S-OIV in our experiments, it cannot be concluded that they will not affect virulence in the future. Because the mutations did not impair replication and transmission significantly, they could reemerge in the future in a different viral genetic background in which they could possibly give rise to more virulent strains. The present study implies that surveillance activities should not remain limited to known virulence factors but should include studies to search for new virulence markers. Surveillance studies should additionally focus on genetic signatures in S-OIVs that are associated with severe disease in humans and on the testing of mutations in several pathogenesis models, as described here.
More research is also needed to test the effects of additional known virulence markers in the polymerase genes or, for instance, in the HA, NA, NS1, or PB1-F2 protein. Furthermore, altered virulence of S-OIV may result not only from mutation, but also upon gene reassortment with contemporary seasonal H1N1 and H3N2 viruses, which could result in enhanced virulence and transmission. Further laboratory investigations and enhanced surveillance activities to detect the emergence of new S-OIV genotypes and phenotypes are needed to determine their importance for public health and to identify a potential need for intensification of surveillance.