Enhanced Replication of Highly Pathogenic Influenza A(H7N9) Virus in Humans

To clarify the threat posed by emergence of highly pathogenic influenza A(H7N9) virus infection among humans, we characterized the viral polymerase complex. Polymerase basic 2–482R, polymerase basic 2–588V, and polymerase acidic–497R individually or additively enhanced virus polymerase activity, indicating that multiple replication-enhancing mutations in 1 isolate may contribute to virulence.

We compared the viral polymerase activity of wildtype GD with that of A/Anhui/1/2013(H7N9) virus (AN) in human A549 cells at 33°C or 37°C (temperatures of the human upper and lower respiratory tract) and in chicken DF-1 cells at 39°C (body temperature of birds). Although both viruses exhibited comparable activity in DF-1 cells, AN activity was higher than GD activity in A549 cells at both temperatures because wild-type AN/PB2 acquired polymerase activity-enhancing K at position 627 of PB2 during replication in the infected human (8). We therefore tested AN/PB2-627E, which possesses an avian ancestral amino acid in PB2-627, and AN/PB2-627E-701N, which possesses polymerase-enhancing PB2-701N (8). In human A549 cells, wild-type GD showed viral polymerase activity comparable to that of AN/PB2-627E-701N (online Technical Appendix Figure 1, panel A, https://wwwnc.cdc.gov/ EID/article/24/4/17-1509-Techapp1.pdf). These results indicate that the viral polymerase activity of wild-type GD in mammalian cells has increased more than that of virus bearing avian-like ancestral AN/PB2-627E.
To determine which component of the viral replication complex (PB2, PB1, polymerase acidic [PA], or nucleoprotein) contributes to the activity of the GD polymerase complex, we tested the polymerase activity of GD replication complexes in which we had replaced each viral protein with its AN/PB2-627E counterpart. We found that the viral polymerase activity in A549 cells was remarkably decreased by AN/PB2-627E and moderately decreased by AN-PA (online Technical Appendix Figure 1, panel B). These results suggest that the PB2 and the PA of GD are involved in the relatively high polymerase activity of the GD replication complex.
When we compared the amino acid sequences of GD-PB2 and GD-PA with those of AN/PB2-627E and AN-PA, we found 8 and 6 differences, respectively (Table 1). To identify which substitutions contributed to the enhanced polymerase activity, we constructed a series of plasmids encoding GD-PB2 or GD-PA harboring single substitutions and examined polymerase activity. Of the 8 PB2 mutants, GD/PB2-482K and GD/PB2-588A drastically reduced viral polymerase activity in A549 cells, although this activity was slightly higher than that of AN/PB2-627E ( Figure 1, panel A). Therefore, we tested the viral polymerase activity of GD-PB2 possessing both mutations (GD/PB2-482K-588A) and found a further decrease in the double mutant. Of the 6 PA mutants, GD/PA-497K showed reduced

Enhanced Replication of Highly Pathogenic Influenza A(H7N9) Virus in Humans
viral polymerase activity in A549 cells ( Figure 1, panel B). Compared with GD/PB2-482K-588A or GD/PA-497K alone, the polymerase activity of GD/PB2-482K-588A plus GD/PA-497K was further reduced ( Figure 1, panel C). Collectively, these data demonstrate that PB-482R, PB2-588V, and PA-497R play crucial roles in the enhanced activity of the GD polymerase complex.
To examine the role of these amino acids on virus growth in human cells, we prepared wild-type and mutant viruses in the GD background by using reverse genetics. GD/PB2-482K and GD/PB2-588A viruses replicated less efficiently than wild-type GD virus in A549 cells (online Technical Appendix Figure 2, panels A, B). Replication of GD/PB2-482K-588A and GD/PB2-482K-588A+PA-497K  PB2  PA  191  340  482  559  560  584  588  702  100  262  387  394  465  497  GD wild-type  E  K  R  T  I  I    . Three animals per group were euthanized on postinfection days 3 and 6. Statistically significant differences compared with GD wild-type-infected mice were determined by use of a 1-way analysis of variance followed by a Dunnett test. GD, A/Guangdong/17SF003/2016; ND, virus not detected (detection limit 2 log10 PFU/g); NT, nasal turbinates; PA, polymerase acidic; PB, polymerase basic.
viruses was less efficient than that of GD/PB2-482K and GD/PB2-588A viruses. GD/PA-497K virus showed growth comparable to that of the wild-type GD virus. In DF-1 cells, all tested viruses produced similar growth curves (online Technical Appendix Figure 2, panel C). These results indicate that PB2-482R and PB2-588V play a central role in enhancing virus replication in mammalian cells.
To assess the role of these amino acids in vivo, we compared virus titers in the lungs, nasal turbinates, and brains of mice infected intranasally with 10 2 PFU of each virus. On day 3 after infection, we did not detect GD/PB2-482K, GD/PB2-482K-588A, or GD/PB2-482K-588A+PA-497K viruses in the lungs or turbinates (  decreased in the lungs and slightly decreased in the turbinates. On day 6 after infection, we found similar trends to those observed on day 3. Wild-type GD virus was detected in the brain of 1 of 3 mice. Next, we evaluated virus pathogenicity in mice infected with 10 2 -10 5 PFU of each virus by monitoring changes in body weight. When inoculated with wild-type GD virus, almost all mice had to be euthanized, resulting in a 50% mouse lethal dose (MLD 50 ) of 10 1.9 PFU (Figure 2, panels A and B). Transient or severe weight loss was caused by GD/PB2-482K, GD/PB2-588A, and GD/ PB2-482K-588A in 1-3 of 20 euthanized mice and by GD/PA-497K viruses in 9 of 20 euthanized mice; MLD 50 values were higher than those for wild-type GD virus. The GD/PB2-482K-588A+PA-497K virus did not affect body weight (MLD 50 >10 5 PFU). Virulence of GD/PB2-588A in mice was comparable to that of GD/PB2-482K and GD/PB2-482K-588A, although GD/PB2-588A replicated better in the lungs and turbinates than GD/PB2-482K and GD/PB2-482K-588A (Table 2); however, the levels of GD/PB2-588A replication in mice were lower than those of wild-type GD, resulting in reduced pathogenicity in mice. These results demonstrate that PB2-482R and PB2-588V contribute to high virulence in mice and that PA-497R is also involved.
PB2-482R was located in 1 of 2 nuclear localization signals spanning amino acids 449 to 495 (13) within the cap-binding domain of the influenza A virus polymerase complex (online Technical Appendix Figure 3, panel B).
PB2-588V was located near PB2-627K in the 627 domain. PA-497R was located at 1 of 2 binding sites for transcriptional activator hCLE (14) but was not exposed on the protein surface. PB2-588V is probably involved in ANP32A-dependent high polymerase activity in mammalian hosts (15), and the role of PB2-482R might differ from that of other polymerase-enhancing amino acids in PB2.