Highly pathogenic avian H5N1 viruses have been evolving rapidly, extending both their geographic distribution and host range since 1996 (45
). H5N1 continues to infect humans, often at the same time that H1N1 and H3N2 influenza virus epidemics occur within a community (reviewed in reference 26
). Reassortment of H5N1 gene segments with those of circulating human influenza viruses is a potential mechanism by which H5N1 viruses could become transmissible and then spread among humans. However, it is not known whether coinfection of a host could result in reassortment between an avian H5N1 virus and a human seasonal influenza virus strain. This risk assessment study has shown that reassortant viruses generated from recent avian H5N1 and human H3N2 viruses can be recovered from the upper respiratory tract of a mammalian host species. After coinfection, 9% (32 reassortants viruses per 360 viruses isolated) of viruses recovered from ferret nasal secretions were reassortants. Importantly, over 34% (11 H5 subtype reassortant viruses per 32 reassortant viruses) of reassortant viruses recovered possessed the avian H5 HA, comprising five unique genotypes. Taken together, these data indicate that recovery of H3 and H5 subtype reassortant viruses from nasal secretions is a likely event in ferrets that are exposed simultaneously to both viruses.
All H5 subtype reassortant viruses examined replicated less efficiently than the Thai16 parental virus in human alveolar lung epithelial (A549) and canine kidney (MDCK) cell cultures. Nevertheless, titers of 106
PFU/ml or greater were achieved in either cell type within 48 h, indicating their ability to replicate in relevant cells of animal and human origin. The reduced viral replication compared to the parental Thai16 virus seen both in vitro and in vivo may be a result of decreased virus fitness caused by functional incompatibilities between reassorted genes. H5N2 subtype reassortants, with the exception of Thai16:Wyom03-NA-NP-M-NS, replicated relatively poorly in the two cell lines compared to parental Thai16 virus, a finding that is consistent with previous studies showing that optimal replication in vitro requires balanced receptor attachment (HA) and release (NA) activities for efficient multicycle replication (11
). Nevertheless, these same H5N2 reassortants replicated with considerably high efficiency in ferrets, suggesting that the requirements for virion surface protein compatibility are host dependent. On the other hand, H5 subtype reassortants that possess the Wyom03 M protein replicated efficiently in vitro but not in vivo. Impaired in vivo replication of HA and M gene reassortant viruses has been attributed to disruption of assembly and budding pathways or inadequate pH regulation by M2 in the trans
Golgi pathway (4
; reviewed in references 21
) although other genes could also play a significant role (1
). Overall, the poor correlation between the in vivo and in vitro replication of reassortants suggests that different factors may govern fitness of reassortant viruses in vitro and in vivo and emphasizes the need for verification of in vitro viral fitness assays using animal models.
All subtype H5 reassortants analyzed in this study failed to transmit efficiently between infected and naïve contact ferrets. None of the recovered hybrid genotypes with an H5 HA possessed the virulence of parental Thai16 virus and/or the transmissibility of parental Wyom03 virus. Previous studies in ferrets have shown that decreased in vivo replication levels are correlated with reduced transmission efficiency (10
). However, it has also recently been shown that even a reverse genetics-generated reassortant virus with the HA and NA from a transmissible H3N2 human virus and the remaining genes from an H5N1 virus did not transmit efficiently in a ferret model (17
). In addition, a recent study has established a correlation between influenza virus transmissibility and the preferential binding of HA to structurally distinct α(2,6)-linked sialic acid receptors (37
). Taken together, the evidence suggests that transmissibility of H5N1 viruses is multifactorial.
The tissue tropism of avian and human influenza viruses in humans has been correlated with the expression of surface receptors that may determine the primary site of coinfection and subsequent reassortment (24
). Studies have shown that HPAI H5N1 viruses may target cells throughout the human respiratory tract (i.e., nasopharynx, adenoid, tonsil, trachea, bronchi, and lung) (5
), where α(2,3)-linked sialic acids are available (24
), while human viruses may target cells primarily found toward the human upper respiratory airways (24
), where there may be a higher abundance of α(2,6)-linked sialic acids (6
). The available evidence suggests that the distribution of glycans with α(2,6)- and α(2,3)-linked sialic acids in human respiratory tract epithelia is similar to that of ferrets. If this assumption is correct, avian H5N1 and human seasonal influenza viruses are predicted to coinfect cells within the upper airways of ferrets. The anatomical sites from which H5 subtype and H3 subtype viruses were recovered in our study are consistent with this hypothesis (42
). H3 reassortant viruses were recovered twice as frequently (66%; 21 H3 subtype reassortant viruses per 32 reassortant viruses) as H5 reassortant viruses (34%; 11 H5 subtype reassortant viruses per 32 reassortant viruses) from ferret nasal secretions. The preponderance of H3 subtype reassortants may be explained by the abundance of α(2,6) receptors for these viruses in the upper respiratory tract of ferrets. Conversely, the delayed nasal shedding of H5 subtype viruses compared to that of H3 viruses may reflect the lower α(2,3) glycan receptor availability as well as other factors such as optimal temperature for replication (7
The results of this study suggest that airways in and toward the upper respiratory tract of ferrets are permissive for coinfection and subsequent reassortment between Wyom03 and Thai16 viruses. The undetectable levels of H3 HA RNA and viruses in lung tissue are consistent with the lack of reassortment in this tissue and with previous study reports (17
) although strain-to-strain variability among human H3N2 viruses with respect to infection of the lower respiratory tract of ferrets has been demonstrated (27
). On the basis of these results and the parallels between influenza virus infections in the ferret model and in humans, we postulate that the site with the highest propensity for reassortment between H5N1 and H3N2 viruses in coinfected humans is located in areas toward the upper airway. However, additional studies will be required to more fully describe the strain-specific tissue tropism of both avian and human viruses in the ferret and to definitively identify the coinfected tissues and cells that serve as the source of reassortant viruses.
In summary, we analyzed the potential generation of an H5 influenza virus with pandemic properties by in vivo reassortment of an H5N1 HPAI with a human influenza virus. Although such studies have the acknowledged potential to generate a highly pathogenic and transmissible H5 subtype virus, these experiments address an important question of immediate public health concern as H5N1 subtype viruses continue to infect humans. Furthermore, all experiments described in this study were performed under controlled biosafety level 3+ laboratory conditions, with specific additional safety precautions taken when reassortant viruses and animals infected with such highly pathogenic viruses were handled. Using the ferret model, we have found a relatively high incidence of HPAI H5N1 and human H3N2 reassortant viruses after coinfection, indicating that susceptible mammalian hosts can support the formation of new reassortant viruses generated from these subtypes. Further study of the anatomical distribution of parental and reassortant viruses from various tissues of the respiratory tract suggests that generation of reassortant viruses most likely occurred in airway epithelia toward and within the upper respiratory tract of ferrets. When characterized further, H5 subtype reassortant viruses recovered from coinfected ferrets showed decreased fitness and lack of transmissibility in this model, which reduces their current potential threat to human health. Nevertheless, since H5N1 viruses continue to circulate and evolve in poultry, it remains possible for an H5 reassortant virus, which has acquired key, as yet ill-defined determinants of transmissibility, to arise among humans and spread via shedding from the upper respiratory tract. Controlled animal studies such as those described here with newly emergent H5N1 and H3N2 viruses as well as those of additional subtypes are required to continually analyze the risk for genetically drifted strains to generate highly pathogenic and transmissible reassortant viruses. These studies are warranted prior to the uncontrolled emergence of such reassortant viruses in nature, where they may pose an imminent pandemic threat to both animal and human populations. Vaccination and surveillance for seasonal influenza virus in areas where H5N1 is endemic remain important public health measures to minimize the risk of the emergence of such a reassortant pandemic strain.