We sequenced the whole genomes of 33 avian influenza virus (H5N1) isolates collected during poultry outbreaks in Vietnam from October 2005 through May 2007. All sequences that were generated in this study have been deposited in GenBank (CY029508–CY029771). The virus was primarily detected in aquatic poultry (ducks, muskovy ducks), but it was also isolated from 2 chickens in December 2006 and January 2007. The date and location of virus isolation are summarized in the Table and Figure 1, panel A.

To understand the developments of influenza virus (H5N1) in Vietnam, we characterized all 8 gene segments of these 33 viral isolates and phylogenetically analyzed them with all available influenza virus (H5N1) previously isolated from Vietnam, Thailand, Malaysia, Lao People’s Democratic Republic, and southern China and with reference viruses belonging to each of the designated clades of the WHO influenza (H5N1) nomenclature system. Sequence assembly, editing, multiple sequence alignment, neighbor-joining, and Bayesian phylogenetic analyses were conducted as previously described (3). Maximum-likelihood trees were constructed by using Garli version 9.04 (10).

The hemagglutinin (HA) genes of all 33 Vietnam isolates were derived from the Gs/GD-like lineage; however, they fell into 3 distinct sublineages (Figure 1, panel B). Seventeen of 33 isolates analyzed were clade 1; however, 15 isolates between March and May 2007 belonged to clade 2.3.4. A single virus isolated in February 2006 (Muscovy duck/Vietnam/1455/2006) clustered within clade 2.3.2.

Phylogenetic analyses also showed a geographic distinction among the isolates characterized in this study. Isolates from samples taken in the northern provinces of Vietnam belonged to clades 2.3.2 and 2.3.4, whereas all isolates in the southern provinces of Vietnam were clade 1 (Figure 1). The clade 1 viruses isolated in the southern provinces from October 2005 through January 2007 were most closely related to viral isolates from poultry in Cambodia in the same period. When one considers the shared land border between Cambodia and southern Vietnam, the close relatedness of these viruses is reasonable. Two of the human influenza (H5N1) cases that were detected from Cambodia were located near the Vietnam-Cambodia border region (11). The genetic similarity of strains of influenza virus (H5N1) in Cambodia and southern Vietnam have been observed since 2004 (4).

The clade 2.3.2 and clade 2.3.4 viruses isolated from northern Vietnam were most closely related to virus isolated from poultry in Guangxi, China (Figure 1, panel B). Because Guangxi shares a border with Vietnam and trade is extensive between those regions, these viruses were most likely introduced into northern Vietnam through poultry trade. However, the clade 2.3.4 viruses from Vietnam do not form a monophyletic group within clade 2.3.4. Therefore, unlike the single introduction of clade 1 viruses in 2003 (4), these results raise the possibility of multiple introductions of the clade 2.3.4 viruses into northern Vietnam.

Phylogenetic analyses of the neuraminidase gene and all internal gene segments (data not shown) show that while most of the isolates were genotype Z viruses, they also formed distinct groups that were broadly similar to the evolutionary relationships seen in the HA tree. Duck/Vietnam/37/2007, which belongs to clade 2.3.4, shared the internal gene constellation of clade 1 viruses (Figure 2), providing evidence for cocirculation of these virus groups and evidence of reassortment between different sublineages within Vietnam influenza (H5N1) isolates. A single influenza (H5N1) genotype G virus (Muscovy duck/Vietnam/1455/2006) was also identified in February 2006, which indicates that genotype G viruses may be persistent in poultry in Vietnam (Figure 2). These results confirm that the genetic diversity of strains of influenza virus (H5N1) in Vietnam is similar to that in southern China during the same period (3).

The motif of multiple basic amino acids at the HA cleavage site that is characteristic of HPAI was maintained in all viruses characterized; however, clade-specific mutations were observed in different sublineages, consistent with previous reports (3,4,8,12). The receptor-binding pocket of HA1 retains Gln 222 and Gly 224 (H5 numbering), which preferentially binds avian-like α2,3-NeuAcGal linkages. However, all 12 clade 2.3.4 viruses and the single clade 2.3.2 virus have an Arg-212-Lys mutation in the HA, whereas 12 clade 1 viruses (marked on Figure 1) have a Ser-123-Pro mutation, previously reported from a Vietnam influenza (H5N1) human isolate (8), which has been associated with receptor binding. The importance of this change is not clear (13,14).

Mutations in the Matrix protein 2 ion channel associated with amantadine resistance were detected in all clade 1 virus isolates tested. These viral strains retained the dual mutations Leu26Ile and Ser30Asn in the M2 protein similar to previous clade 1 strains (15). No mutation associated with amantadine resistance was recognized in those clade 2.3.2 and clade 2.3.4 strains except the reassortant Dk/VNM/37/07, which had an additional Val-27-Ala mutation in the M2 protein. Thus we recorded an HPAI (H5N1) virus strain with a triple mutation associated with amantadine resistance. All viruses characterized do not have mutations in the NA gene that confer resistance to oseltamivir. Other known virulence mutations, including at polymerase basic protein 2 position 627, were not present in any of the viruses characterized.

This study confirms that clade 2.3.4 virus sublineages that are dominant in southern China have now spread to northern Vietnam (3). These viruses appear to have replaced the clade 1 viruses in northern Vietnam just as previous influenza (H5N1) sublineages were replaced in southern China (3); however, clade 1 viruses are still detected in the southern provinces of Vietnam. It is, therefore, possible that the clade 1 viruses in southern Vietnam may eventually be replaced by clade 2.3.4. The availability of extensive genetic data from southern China enables us to recognize the development of influenza virus (H5N1) in Vietnam and indicates that clade 2.3.4 viruses may have been introduced into Vietnam on multiple occasions. However, because systematic surveillance data are lacking, determining the interaction of viruses between the northern and southern provinces of Vietnam, and also between different countries in Southeast Asia, is not possible. When one considers that multiple sublineages of influenza virus (H5N1) are simultaneously endemic to Southeast Asia, systematic surveillance in poultry remains essential to understand the further evolution of this subtype in this region and the potential for pandemic emergence, as well as to monitor the efficacy and cross-protection of poultry vaccines.