Emergence of the virulence-associated PB2 E627K substitution in a fatal human case of highly pathogenic avian influenza virus A(H7N7) infection as determined by Illumina ultra-deep sequencing

Abstract

Avian influenza viruses are capable of crossing the species barrier and infecting humans. Although evidence of human-to-human transmission of avian influenza viruses to date is limited, evolution of variants toward more-efficient human-to-human transmission could result in a new influenza virus pandemic. In both the avian influenza A(H5N1) and the recently emerging avian influenza A(H7N9) viruses, the polymerase basic 2 protein (PB2) E627K mutation appears to be of key importance for human adaptation. During a large influenza A(H7N7) virus outbreak in the Netherlands in 2003, the A(H7N7) virus isolated from a fatal human case contained the PB2 E627K mutation as well as a hemagglutinin (HA) K416R mutation. In this study, we aimed to investigate whether these mutations occurred in the avian or the human host by Illumina Ultra-Deep sequencing of three previously uninvestigated clinical samples obtained from the fatal case. In addition, we investigated three chicken samples, two of which were obtained from the source farm. Results showed that the PB2 E627K mutation was not present in any of the chicken samples tested. Surprisingly, the avian samples were characterized by the presence of influenza virus defective RNA segments, suggestive for the synthesis of defective interfering viruses during infection in poultry. In the human samples, the PB2 E627K mutation was identified with increasing frequency during infection. Our results strongly suggest that human adaptation marker PB2 E627K has emerged during virus infection of a single human host, emphasizing the importance of reducing human exposure to avian influenza viruses to reduce the likelihood of viral adaptation to humans.

FIG 1

Steps of the data analysis pipeline and their effect on the influenza virus data set. (A) From the three control data sets, the mismatch frequency is expressed for each influenza virus nucleotide. This shows that the enhanced mapping algorithm was insufficient in removing the variation observed at specific nucleotides of the duplicate RT-PCR sets. (B) Focusing on 1% of the nucleotides that displayed the highest mismatch frequency, the sequence-specific errors are expressed by error bars. While plasmid nucleotides with the highest mismatch frequency were associated with sequence-specific errors, RT-PCR nucleotides with the highest mismatch frequency were not. (C) During the final step of the data analysis pipeline, all nucleotide positions displaying variation were assessed individually by mutational position graphs that demonstrate the location within Illumina sequence reads that code for variation. A mutational position graph of variation associated with the 5′-terminus of a PCR product is presented (top) together with a “hairy caterpillar”-shaped plot showing variation regardless of the position in the sequence fragment, thus considered a true variant (bottom). (D) Effect of each type of data filtering on the mismatch frequency expressed as mean (+ range) for the plasmid, RT-PCR 1, and RT-PCR 2 data sets. While the enhanced mapping algorithm (RHM) and removal of sequence-specific errors (SSE) successfully reduced the mismatch frequency of nucleotides in the plasmid data set, mutational position graphs (MPG) were required to reduce the maximum mismatch frequency to less than 2% of the RT-PCR data sets.

FIG 2

Timeline integrating the data on poultry and the fatal case and the corresponding virological data. It shows the increase of PB2 E627K during human infection in the absence of PB2 E627K detected at the avian source of infection. A quarter of the influenza A(H7N7) virus population in the lower respiratory tract of the veterinarian contained avian PB2 627E virus 11 days postexposure, while the postmortem lower lung tissue contained fully adapted (PB2 E627K) virus 16 days postexposure, suggesting that the PB2 E627K mutation emerged during infection of the human host.

FIG 3

Detection of influenza virus defective RNA by deep sequencing. Characterization of the start and endpoint of internal deletions in A(H7N7) virus gene segments by deep sequence analysis identified a multitude of influenza virus defective RNAs in chicken samples obtained from the source and control farm. Of the total reads that contained an internal deletion, the pie charts illustrate the contribution for each of the influenza A(H7N7) virus gene segments.