The effect of the PB2 mutation 627K on highly pathogenic H5N1 avian influenza virus is dependent on the virus lineage

Abstract

Clade 2.2 Eurasian-lineage H5N1 highly pathogenic avian influenza viruses (HPAIVs) were first detected in Qinghai Lake, China, in 2005 and subsequently spread through Asia, Europe, and Africa. Importantly, these viruses carried a lysine at amino acid position 627 of the PB2 protein (PB2 627K), a known mammalian adaptation motif. Previous avian influenza virus isolates have carried glutamic acid in this position (PB2 627E), commonly described to restrict virus polymerase function in the mammalian host. We sought to examine the effect of PB2 627K on viral maintenance in the avian reservoir. Viruses constructed by reverse genetics were engineered to contain converse PB2 627K/E mutations in a Eurasian H5N1 virus (A/turkey/Turkey/5/2005 [Ty/05]) and, for comparison, a historical pre-Asian H5N1 HPAIV that naturally bears PB2 627E (A/turkey/England/50-92/1991 [50-92]). The 50-92 PB2 627K was genetically unstable during virus propagation, resulting in reversion to PB2 627E or the accumulation of the additional mutation PB2 628R and/or a synonymous mutation from an A to a G nucleotide at nucleotide position 1869 (PB2 A1869G). Intriguingly, PB2 628R and/or A1869G appeared to improve the genetic stability of 50-92 PB2 627K. However, the replication of 50-92 PB2 627K in conjunction with these stabilizing mutations was significantly restricted in experimentally infected chickens, where reversion to PB2 627E occurred. In contrast, no significant effects on viral fitness were observed for Ty/05 PB2 627E or 627K in in vitro or in vivo experiments. Our observations suggest that PB2 627K is supported in Eurasian-lineage viruses; in contrast, PB2 627K carries a significant fitness cost in the historical pre-Asian 50-92 virus.

Fig 1

Rescue of recombinant viruses altered in PB2 at residue 627 and spontaneous selection of additional mutations upon passage of pre-Asian H5N1 50-92 PB2 627K virus. Viruses generated by reverse genetics (RG) were generated by one passage in MDCK cells. Viral RNA of virus generated by reverse genetics was extracted, amplified by RT-PCR, and sequenced in the PB2 627 region. (A) DNA sequencing chromatograms of the sequence from nucleotides 1867 to 1884 of the PB2 gene of viruses generated by reverse genetics (Ty/05/E, Ty/05/K, 50-92/aEQ, 50-92/aKQ) and/or passage mutants generated by reverse genetics (50-92/gKQ, 50-92/gKR). Nucleotide changes are shaded in gray. Mutations introduced by site-directed mutagenesis are underlined by a solid line. Mutations detected upon virus passage are underlined by a dashed line. (B to F) Viruses generated by reverse genetics were blind passaged in embryonated chicken eggs up to egg passage 3 (EP3), and the PB2 627 region was identified by Sanger sequencing (n = 7). Eggs were inoculated with <4 hemagglutination units determined by hemagglutination assay using turkey red blood cells and incubated at 37°C. (E and F) 50-92/aKQ and 50-92/gKQ were twice retested using separate viruses generated by reverse genetics displayed on the left and right. (F) 50-92/gKQ input has an unknown mix of 50-92/gKQ and 50-92/gKR viruses after generation through one MDCK passage.

Fig 2

H5N1 50-92/gKR virus is attenuated in vivo. Seven 10-week-old chickens were oculonasally inoculated with 10⁶ EID₅₀ units of 50-92/aEQ or 50-92/gKR. Viral RNA was extracted from oropharyngeal swab, cloacal swab, or feather samples, and the virus titer was measured by qRT-PCR of the matrix gene, displayed as the mean log₁₀ number of REU. (A) Oropharyngeal virus shedding; (B) cloacal virus shedding; (C) area-under-the-curve (AUC) analysis of shed virus up to 36 h.p.i.; (D) virus titer of the basal calamus of feather samples taken from the brow of chickens at 36 h.p.i.; (E) percent survival of chickens; (F) average number of clinical signs observed per bird at chosen time points. Log values (plus an arbitrary 0.01 to allow log transformation of 0 values) were used for suitable statistical analysis by two-way analysis of variance corrected by the Bonferroni posttest (A, B, C, F) or by two-tailed paired t test (D). Error bars display the SEMs. *, P < 0.05; **, P < 0.01.

Fig 3

Reversion of PB2 K627E and R628Q in mutant non-Asian-lineage H5N1 50-92/gKR upon replication in chickens. At chosen time points, viral RNA was extracted from virus-positive oropharyngeal and cloacal swab samples from chickens infected with 50-92/gKR (Fig. 2), amplified by RT-PCR, and sequenced in the PB2 627 region. Graphs show where 627E was detected; calculated percentages include data for undetectable shedding and dead birds. (A) Oropharyngeal swab samples; (B) cloacal swab samples.

Fig 4

PB2 627K does not carry a fitness cost for Eurasian-lineage H5N1 Ty/05. Twelve 12-week-old Pekin ducks were oculonasally inoculated with 10⁶ EID₅₀ units of Ty/05/K or Ty/05/E. Five 4-week-old naive contact Pekin ducks were cohoused at 1 d.p.i. Five 12-week-old ducks were euthanized for postmortem analysis at 2 d.p.i. Virus RNA was extracted from oropharyngeal swab, cloacal swab, or feather samples, and the virus titer was measured by qRT-PCR of the HA gene, displayed as the mean log₁₀ number of REU. (A) Daily oropharyngeal virus shedding of 12-week-old ducks; (B) daily oropharyngeal and cloacal virus shedding of 4-week-old contact ducks; (C) virus titer of the basal calamus of feather samples taken from the breast of dead or euthanized 4-week-old ducks; (D) percent survival of 4-week-old contact ducks. Error bars display the SEMs.

Fig 5

H5N1 Ty/05/K and Ty/05/E virus antigen distribution in experimentally infected ducks. Immunohistochemical demonstration of influenza A virus nucleoprotein (brown) in tissues of ducks challenged with either Ty/05/K (left) or Ty/05/E (right) at 2 d.p.i. (A) Nasal turbinate. Lymphohistiocytic and endothelial cells in the lamina propria are labeled. (B) Air sac. Note the widespread antigen detection in the epithelium. (C) Spleen. Several positive cells in red pulp are shown. (D) Bursa of Fabricius. Immunolabeling can be seen in lymphocytes, macrophages, and reticular cells in the medulla of bursal follicles. (E) Ovary. Note detection of virus antigen in stromal cells. Bars, 50 μm.

Fig 6

Growth kinetics in avian cells. Primary CEF cells were inoculated with 50-92/aEQ, 50-92/gKR, 50-92/gKQ, or 50-92/aKQ at an MOI of 10⁻³. Cells were incubated at 37°C (A) or 41°C (B). (A and B) At 28 h postinfection, viral RNA was extracted, amplified by RT-PCR, and sequenced in the PB2 627 region. Reversion from 50-92/aKQ to 50-92/gKQ was detected at 28 h.p.i. (black arrow). In addition, CEF cells were inoculated with Ty/05/K and Ty/05/E at an MOI of 10⁻⁴. Cells were incubated at 37°C (C) or 41°C (D). Virus RNA was extracted from the supernatant, and virus titer was measured by qRT-PCR of the matrix (50-92) or HA (Ty/05) gene, displayed as the mean log₁₀ number of REU. Values were determined in triplicate. These trends were reproducible. Error bars display the SEMs. Statistical analysis was by either two-way analysis of variance corrected by the Bonferroni posttest (50-92) or multiple two-tailed t test corrected by the Holm-Sidak posttest (Ty/05). *, P < 0.05; **, P < 0.01; ****, P < 0.00001.

Fig 7

Differences in polymerase activity. Polymerase complex and NP expression plasmids, together with the minigenome firefly luciferase reporter plasmid pChicken-PolI-Firefly or pHuman-PolI-Firefly, were transfected into avian DF-1 or human 293T cells, respectively. Cells were lysed after 20 h, and the level of firefly luciferase activity was measured. (A) Log₁₀ firefly luciferase activity of Ty/05/K, TY/05/E, 50-92/EQ, and 50-92/KQ with cognate polymerase complex and NP conducted at 37, 39, and 41°C in DF-1 cells; (B) log₁₀ firefly luciferase activity of 50-92/EQ or 50-92/KQ together with the 628R mutation, conducted at 37 and 41°C in DF-1 cells; (C) log₁₀ firefly luciferase activity of Ty/05 and 50-92 PB2 E/K constructs together with pH1N1 human control in 293T cells at 37°C. Mock, transfection of firefly luciferase reporter only. Values were determined in triplicate. These trends were reproducible. Error bars display the SEMs. Statistical analysis was by two-way analysis of variance corrected by the Bonferroni posttest. **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Fig 8

Altering the genetic constellation of 50-92 prevented reversion of the 627K genotype. Viruses generated by reverse genetics were blind passaged by MDCK cells in T25 flasks (n = 1). Cell supernatant was harvested at each passage, viral RNA was extracted, and the PB2 627 region was amplified by RT-PCR and sequenced by Sanger sequencing. Chromatogram trace files for each virus and passage of PB2 from nucleotides 1867 to 1869 and 1879 to 1984 are shown. Solid arrows, nucleotide change.