Molecular Basis for High Virulence of Hong Kong H5N1 Influenza A Viruses

See allHide authors and affiliations

Science  07 Sep 2001:
Vol. 293, Issue 5536, pp. 1840-1842
DOI: 10.1126/science.1062882


In 1997, an H5N1 influenza A virus was transmitted from birds to humans in Hong Kong, killing 6 of the 18 people infected. When mice were infected with the human isolates, two virulence groups became apparent. Using reverse genetics, we showed that a mutation at position 627 in the PB2 protein influenced the outcome of infection in mice. Moreover, high cleavability of the hemagglutinin glycoprotein was an essential requirement for lethal infection.

An outbreak of H5N1 influenza A virus in Hong Kong in 1997 resulted in 6 deaths among 18 people infected (1–3). The virulence seen in mice when infected with these human isolates showed some correspondence with the severity of disease in adult patients, although there are exceptions (3–7). To determine the molecular basis for the difference in virulence among Hong Kong H5N1 viruses, we generated reassortants between virulent and avirulent viruses (Fig. 1) using a plasmid-based reverse genetics system (8). First, we constructed plasmids in which a human RNA polymerase I promoter and a mouse polymerase I terminator flanked cDNAs for the full-length RNAs of A/Hong Kong/483/97 (HK483) (which caused lethal systemic infection in mice) or A/Hong Kong/486/97 (HK486) (which produced nonlethal respiratory infection) viruses (9). Two forms of the HA gene were identified in our HK486 virus stock: one possessing Ser at position 227 (H3 numbering) (486HA227S) and another with Ile at position 227 (486HA227I). Using these plasmids, we produced HK483 and HK486 transfectant viruses (10)—designated HK483RG, HK486RG(HA-227S), and HK486RG(HA-227I)—and inoculated them into Madin-Derby canine kidney (MDCK) cells to produce virus stocks.

Figure 1

Pathogenicity of transfectant viruses. Bars indicate the origin of the viral gene: A/Hong Kong/483/97 (HK483, red), A/Hong Kong/486/97 (HK486, blue). I and S in blue bars indicate the HK486 HA possessing either Ile or Ser at position 227, respectively. Red hatched bar, amino acid sequence at the HA cleavage site of HK483 changed from PQRERRRKKR/G to PQ----RETR/G, as described in the text. The virus dose lethal to 50% of mice (MLD50) was determined as described (5).

Figure 1 and Table 1 compare the pathogenicity of the transfectants, based on the virus dose lethal to 50% of infected mice (50% mouse lethal dose, MLD50) and virus growth in the organs of mice infected intranasally with 100 plaque-forming units (PFU) of virus. HK483RG, like the original HK483 virus, produced lethal systemic infection in mice (MLD50, 1.7) (Table 1). By contrast, the HK486 transfectants [HK486RG(HA-227S) and HK486RG(HA-227I)] did not kill mice even at a dose of 103PFU, and both were recovered only from respiratory organs (Table 1). However, the MLD50 for HK486RG(HA-227S) was 4.6 × 104, whereas HK486RG(HA-227I) failed to kill mice at the highest dose tested, 1.3 × 106 PFU.

Table 1

Tissue tropism of H5N1 transfectant viruses. BALB/c mice, anesthetized with methoxyflurane, were infected intranasally with 50 μl of virus (100 PFU). Three mice from each infected group were killed on day 3 or day 6 after infection (or both) for virus titration. When virus was not recovered from all three mice, individual titers were recorded. (–) Virus not isolated.

View this table:

To elucidate the molecular basis of the virulence discrepancy between the HK483 and HK486 viruses, we generated a spectrum of single-gene reassortants (Fig. 1) and tested their pathogenicity in mice. Among nine single-gene reassortant viruses, each containing one gene segment from HK486 virus and the remaining segments from HK483 virus, only those possessing the PB2 (HK3/6PB2) or HA-227I (HK3/6HA227I) gene from the HK486 virus were appreciably attenuated (MLD50, 1.0 × 104 or 1.1 × 102PFU, respectively, compared with <5 PFU for all other transfectants) (Fig. 1). Moreover, although more than 106 PFU/g of virus was detected at 6 days after infection in the lungs of mice infected with either HK3/6PB2 or HK3/6HA227I virus (Table 1), virus was not recovered from other nonrespiratory organs (with the exception of the heart in mice infected with HK3/6PB2 virus). By contrast, the single-gene reassortants containing the HK486HA227S (HK3/6HA227S) gene or the HK486 NA (HK3/6NA) gene (both tested as representatives of virulent single-gene reassortants) were not attenuated and caused systemic infection (Table 1).

We next tested which gene from HK483 virus converts the HK486 virus to high virulence, using an approach similar to the one described above. The reassortant possessing the PB2 gene from HK483 virus, 486HA227S HA gene, and the remaining segments from the HK486 virus (HK6HA227S/3PB2) was the only construct with noteworthy virulence (MLD50, 0.4 PFU), killing all mice by day 5 after infection and being recovered from all organs tested (Fig. 1 and Table 1); however, the reassortant possessing the PB2 gene from HK483, 486HA227I, and the remaining genes from HK486 virus (HK6HA227I/3PB2) was attenuated with an MLD50 of 3.4 × 103, causing only respiratory infection. None of the other single-gene reassortants, possessing only one gene from HK483 virus and the remaining genes from HK486 virus, was as virulent as the HK483 virus (MLD50, >102 PFU). The single-gene reassortants possessing 486HA227I were more attenuated than those possessing 486HA227S. The HK6HA227S/3NA and HK6HA227I/3NA reassortants, possessing the HK483 NA gene and all remaining genes from HK486 (tested as representatives of avirulent single-gene reassortant viruses), were recovered only from respiratory organs (Table 1). These results indicate that the PB2 viral protein is responsible for the difference in virulence between the two Hong Kong H5N1 viruses, and that a Ser-to-Ile substitution at position 227 of the HK486 HA can reduce the virulence potential of the virus.

There are eight amino acid differences between the PB2 proteins of HK483 and HK486 viruses (Fig. 2). To identify the specific changes that give rise to virulence, we generated mutant HK483 viruses possessing a chimeric PB2 protein (Fig. 2) and determined their MLD50 values. The mutant viruses possessing chimera 2, 4, or 5 were attenuated (MLD50, >103 PFU), whereas those with chimera 1, 3, or 6 were virulent (MLD50, <3.1 PFU). These results implicated amino acid residue 627 or 675 (or both) in contributing to the pathogenicity of HK483 virus. A mutant HK483 virus containing a single Leu-to-Ile substitution at position 675 of PB2 (HK3PB2-675I) and another mutant virus containing an Ile-to-Leu substitution at position 675 of HK486 PB2, with all remaining genes from the HK483 virus (HK6PB2-675L), were constructed. These mutations failed to alter the ability of PB2 to contribute to viral virulence (Fig. 2). However, a Lys-to-Glu substitution at position 627 in PB2 (HK3PB2-627E) resulted in marked attenuation of HK483 virus (MLD50, 2.3 × 103). Finally, we tested a mutant of HK486RG(HA-227S) characterized by a single Glu-to-Lys substitution at position 627 in PB2 (HK6PB2-627K). It killed all mice by day 5 after infection (MLD50, 5.8 PFU) and was recovered in high titers from all organs examined (Fig. 1 and Table 1).

Figure 2

Schematic diagram of chimeric and single–amino acid PB2 mutants, with their virulence in mice (MLD50). Differences in PB2 amino acid residues between A/Hong Kong/483/97 and A/Hong Kong/486/97 are shown as single-letter amino acid codes with their positions indicated at the top of the diagram. The red and blue bars indicate whether the amino acid regions originated from HK483 or HK486, respectively. MLD50 values are reported at the far right. Single-letter abbreviations for the amino acid residues are as follows: A, Ala; E, Glu; I, Ile; K, Lys; L, Leu; Q, Gln; R, Arg; S, Ser; T, Thr; and V, Val.

For avian influenza A viruses, cleavability of the HA molecule plays a major role in virulence in birds, although other genes also contribute to this property (11–13). Thus, we generated a mutant HK483 virus in which the amino acid sequence at the HA cleavage site, PQRERRRKKR/G, was converted to the sequence of typical avirulent avian viruses, PQ----RETR/G (where a dash indicates a deletion). When tested in mice, this HA mutant (HK3HAavir) was highly attenuated (MLD50, >105 PFU) (Fig. 1), and none of the infected mice showed signs of disease. On intranasal inoculation into mice (100 PFU), this mutant was recovered from lung and nasal turbinates, but not from any other organs 3 days after infection. By 6 days after infection, the virus could not be recovered from any organ (Table 1). The virulence of HK483 virus in mice appears to involve HA cleavability, as well as an amino acid substitution in the PB2 viral protein. We suggest that this mutation enables HK483 virus to adapt to efficient growth in mice and possibly humans. Indeed, a mutation at position 627 of PB2 has been linked to the host range of influenza A viruses (14). The virulent HK483 and avirulent HK486 viruses also differed in their ability to invade systemic organs, suggesting that the PB2 protein may be a determinant of cell tropism, in accordance with previous findings (14–17). Whether the mechanisms for these effects are discrete or overlapping and rely on known properties of PB2 [e.g., cap-binding (18, 19)] remains to be determined.

The HK486 virus consists of two types of HA molecules that differ by a single amino acid, resulting in different abilities to support virulence in mice. The Ile at position 227 of the HA, which reduced the ability to support virulence, was found only in the HK486 virus. This amino acid is located in the receptor-binding pocket on the distal tip of the HA (20) and therefore may affect the ability of the virus to recognize and bind to target cells. It will be important in future experiments to compare the receptor binding of the two types of HK486 HAs.

Here we have demonstrated that single amino acid substitutions in PB2 and HA are principal determinants of the difference in virulence between the two viruses tested. However, genes other than PB2 and HA may also contribute to this difference to a lesser extent, as suggested by the limited, but appreciable, increase in virulence upon replacement of the HK486 NA with that of HK483 virus (HK6HA227S/3NA) (Fig. 1), consistent with the concept that influenza virus pathogenicity is multigenic (11–13).

  • * To whom correspondence should be addressed. E-mail: kawaokay{at}


View Abstract

Navigate This Article