Clues to the Virulence of H5N1 Viruses in Humans

Science  17 Mar 2006:
Vol. 311, Issue 5767, pp. 1562-1563
DOI: 10.1126/science.1125998

H5N1 strains of avian influenza A virus have already caused the deaths of more than 90 people since the outbreak of infection in Southeast Asia in 1997, corresponding to a death rate of ∼50% for known infections. These viruses, which have now spread from Asia to Europe and Africa, are strong candidates for causing the next flu pandemic if they acquire the ability for efficient human-to-human transmission. A major research goal has been to identify the molecular basis of the virulence of H5N1 viruses in humans (1, 2). Several virus-encoded proteins will likely contribute to virulence in humans, because previous studies have shown that the virulence of influenza A virus of different organisms is caused by multiple genes (2). The study by Obenauer et al. (3) on page 1576 of this issue, presents evidence suggesting that the virulence of H5N1 viruses may be caused at least in part by the function of a previously unnoticed amino acid sequence motif in the virus-encoded nonstructural protein called NS1 (see the figure).

The NS1 protein is synthesized in infected cells but not incorporated into virus particles. Rather, this small, multifunctional protein participates in both protein-RNA and protein-protein interactions during infection. Its amino-terminal RNA-binding domain binds to double-stranded RNA (dsRNA) with low affinity (4), but the significance of this activity during viral infection is controversial (5, 6). The NS1 protein also binds and inhibits the function of two cellular proteins that are required for the modification of the 3′ ends of cellular messenger RNAs (mRNAs) (6). Consequently, the production of a key component of the host antiviral response, interferon-β mRNA, is substantially reduced, although not eliminated (7). The NS1 sequences that participate in binding to either these two cellular proteins or dsRNA are not part of the putative new virulence determinant in NS1.

Variation locations.

Strains of the H5N1 influenza A virus that are virulent in mammals, including mice and humans, have alterations in the sequences of any of three viral proteins hemagglutinin (HA), the viral polymerase protein PB2, and the nonstructural protein NS1. Influenza A virus has 8 genomic RNA strands and 10 proteins, as shown.


Ten other proteins are encoded by influenza A virus, whose genome consists of eight single-stranded RNAs (8). Three proteins (PB1, PB2, and PA) comprise the polymerase that is associated with each of the viral genomic RNAs in the virus particle. The polymerase copies these genomic RNAs into viral mRNAs and also catalyzes the replication of the genomic RNAs in infected cells. Investigators have identified the amino acid sequences of the PB1, PB2, and PA proteins that function in specific steps of virus-specific RNA synthesis or in mediating interactions between the three proteins (2). The amino acid at position 627 in PB2, which has been implicated in human virulence of H5N1 viruses, does not participate in these known functions. H5N1 viruses that are virulent in mice encode lysine at this position in PB2, whereas H5N1 viruses that are not virulent in mice, as well as other avian influenza A virus strains, encode glutamic acid at this position (9). It is thought that this change from glutamic acid to lysine represents an adaptation of H5N1 viruses for efficient replication in mammalian cells (10).

Another virulence determinant for the H5N1 virus in mammals has previously been identified in the hemagglutinin, the major surface protein of the virus (8). Hemagglutinin, which binds to sialic acid-containing receptors on host cells, is the protein against which neutralizing antibodies are produced. Because the H5 type of hemagglutinin in avian influenza A viruses has not been found in previously circulating human influenza A virus strains, humans are potentially susceptible to infection by these viruses. Cleavage of hemagglutinin into two disulfide-linked sub-units is a prerequisite for initiating infection (8). H5N1 viruses that are highly pathogenic in mice contain a stretch of basic residues adjacent to the hemagglutinin cleavage site, enabling these hemagglutinins to be cleaved by ubiquitous intracellular proteases, including furin. Recombinant H5N1 viruses lacking these basic amino acids are no longer virulent in mice (9), demonstrating that the presence of these amino acids, and the consequent cleavage by intracellular proteases, are required for the virulence of these viruses.

To further understand the molecular basis of virulence, Obenauer et al. first sequenced the genes of a large number of H5N1 viruses isolated from wild birds and poultry, providing an invaluable resource for many investigators. This analysis revealed not only the expected variability in the sequences of the two major surface proteins of the virus, hemagglutinin and neuraminidase, but also variability in the sequence of the NS1 protein. Despite variability in the latter, it was noted that the carboxyl terminus of the NS1 proteins of the vast majority of avian H5N1 viruses contains a sequence motif, Glu-Ser-Glu-Val (ESEV). These residues are predicted to mediate binding to proteins bearing a region called a PDZ domain. The multitude of human proteins that contain a PDZ domain function in diverse cellular signaling pathways including those that regulate protein traffic within the cell and those that maintain cell morphology and organization. Another PDZ-binding sequence, Glu-Pro-Glu-Val (EPEV), was identified at the carboxyl terminus of the NS1 proteins of all the virulent H5N1 viruses isolated from humans. In contrast, the carboxyl terminus of the NS1 proteins of low-virulence human influenza A usually contains a different sequence, Arg-Ser-Lys-Val (RSKV), which is not a PDZbinding motif. Further, Obenauer et al. verified that the carboxyl-terminal ESEV and EPEV sequences indeed bind to PDZ domains. Consequently, the presence of a functional carboxyl-terminal PDZ-binding domain in the NS1 protein of H5N1 viruses correlates with human virulence. This supports the authors' hypothesis that the carboxyl-terminal domain of the NS1 proteins of avian H5N1 viruses acts as a virulence factor by binding cellular PDZ-containing proteins and disrupting their participation in important cellular processes.

This is an intriguing hypothesis that, however, needs to be evaluated in animal experiments with H5N1 viruses that have been altered to express a NS1 protein lacking the carboxyl-terminal ESEV/EPEV sequence. Such experiments are critical because it has already been established that this carboxyl-terminal sequence is not required for the virulence of previously isolated H5N1 viruses in ferrets (11). An analysis of the virulence of H5N1 viruses isolated in 2004 identified the human isolate A/Vietnam/1203/04 as the most pathogenic isolate. The NS1 protein encoded by this virus is truncated and consequently lacks the suspect carboxyl-terminal ESEV/EPEV motif. Future experiments will establish whether eliminating the carboxyl-terminal ESEV/EPEV sequence of the NS1 protein of other H5N1 viruses has any effect on their virulence in animal models. In addition, the search for other molecular determinants of the virulence of H5N1 viruses in humans will undoubtedly continue.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.


Navigate This Article