Replacement of Sublineages of Avian Influenza (H5N1) by Reassortments, Sub-Saharan Africa

The 3 sublineages initially introduced into Nigeria in 2006 were gradually replaced by different reassortants.

H ighly pathogenic avian infl uenza (HPAI) virus subtype H5N1 in Africa was fi rst reported from northern Nigeria in February 2006. Phylogenetic analysis of the complete genome showed that these viruses were clearly distinct from the 2 lineages that were found during the same period in southwestern Nigeria (1,2). The 3 sublineages (referred to as A, B, and C), 2 of which emerged from a common node, had evolved from subtype H5N1 strains that were originally found around Qinghai Lake in 2005. These strains clustered with viruses isolated from 2006 from southern Russia, Europe, and the Middle East (clade 2.2, www.who.int/csr/disease/infl uenza/tree_large.pdf) but not with the strains prevalent in southeast Asia (3). The timeline, the observed infl uenza A (H5N1) substitution rates in Africa, and the phylogenetic relationship suggested that the sublineages were independently introduced into the country (1,2). These sublineages were later found throughout Africa with a distinct geographic distribution (2,4). Sublineage A was also found in Niger and Togo (hemagglutinin [HA] sequence); sublineage B was detected in Egypt and in a human patient in Djibouti (partial HA sequence), and sublineage C was found in Burkina Faso, Sudan, Côte d'Ivoire, Ghana (HA and neuraminidase [NA] sequences) (5) and Cameroon (NA sequence) (6). Sublineage A strains were also referred to as EMA 2, and both sublineages B and C belong to EMA 1 (3). In 2006, one strain with reassorted genes was reported among 35 full-length sequences of the European-Middle Eastern-African lineage (1)(2)(3)(4). We describe new HPAI (H5N1) strains collected in southwestern Nigeria during the second half of 2007, most of which were different reassortants of sublineages A and C.

Materials and Methods
Cloacal swabs were obtained from 8 chicken farms in Lagos (1), Ogun (5), Oyo (1) and Ekiti (1) States from June through November 2007. RNA extraction from cloacal swabs, reverse transcription-PCR amplifi cation, and gene sequencing were conducted as described (1). For most viruses, complete sequences were obtained for all gene segments. Kimura distances were calculated on the basis of complete or partial gene sequences by including the maximum sequence length available from all strains included in the comparison. Phylogenetic trees were calculated by using PAUP version 4.0 beta 10 (7) with the maximum-like-lihood method. The best model was determined by using MODELTEST (8). The sequences have been submitted to GenBank with the accession nos. FM160635-FM160642 and FM164800-FM164855.

Reassortants
All genes of A/chicken/NIE/EKI15/2007 and A/chicken/NIE/OYO14/2007 clustered phylogenetically with sublineage A strains (Figures 1, 2). The Kimura distances between the genes of these viruses were 0.4%-1.4%. Among all subtype H5N1 virus sequences published in the Infl uenza Sequence Database

Mutations
The amino acid sequences of the HA cleavage site (PQGERRRKKRG) of the strains described here are identical to those of all HPAI (H5N1) strains reported from West Africa. All viruses had identical amino acids in all positions of the HA protein that are associated with preferential binding to α2,3-linked sialic acid (9,10) as described (2).
As for all HPAI (H5N1) strains from Africa, the above viruses had the virulence marker lysine (K) in position 627 of PB2 associated with accelerated viral replication, reduced host defense, higher mortality rate in mice (11), and a wider host range of subtype H5N1 strains (12). None of the known markers in the matrix 2 gene associated with resistance to amantadine (13) and in the NA gene associated with resistance to oseltamivir (H274Y) (14) were detected.

Discussion
Gene sequences of all 8 HPAI viruses (H5N1) described here were more closely related to sublineages A or C strains found in Nigeria than to any other published H5N1 virus subtypes. In particular, they were more closely related to the fi rst strains found in Nigeria in the beginning of 2006 than to any strains found outside the country. Thus sortants have largely replaced the initial sublineages from which they were derived and that reassortments are pervasive. This fi nding confi rms that reassortments between subtype H5N1 viruses occur frequently when different strains cocirculate in the same region (16) and is of particular concern if the increasing prevalence is the result of adaptation to the African environment. Although segments of the replication complex (PB1, PB2, PA, and NP) may reassort individually without affecting viral fi tness (16), there seems to be a coordinated evolution of the HA and NA genes (17). In all but 1 of the Nigerian reassortants, HA and NA genes originated from different sublineages (C and A), suggesting compatibility between phenotypes of both sublineages. All reassortants from Nigeria included sublineage C-derived NS genes, which may suggest a higher fi tness of these viruses. Sublineage C-derived NS1 and NS2 sequences from all Nigerian reassortants and 11 unpublished sequences from AC HA/NS reassortants identifi ed in other sub-Saharan regions showed 2 amino acids (NS1 V194 and NS2 R34), which were never identifi ed in sublineage A viruses. It has been shown that modifi cations in the NS proteins, including amino acids adjacent to V194, may modulate the virulence of HPAI (H5N1) (18,19). Alternatively, the observation that all reassortants in West Africa have sublineage C-derived NS genes may suggest a better adaptation to the African environment of viruses that came from the cold temperatures of central Asia. Thus, the infl uence of differences in ecology between Africa and Eurasia on viral selection and dynamics deserves further attention.
Although no reassortments have been reported among clade 2.2 viruses (www.who.int/csr/disease/infl uenza/ tree_large.pdf) in Central Asia, Europe, and the Middle East since their emergence from Qinghai Lake region in 2005, reassortments of these viruses seem to be rampant in sub-Saharan Africa, where they have become the critical determinant of genetic diversity of HPAI (H5N1). Because of low prevalence, mainly in wild birds, clade 2.2 viruses have few opportunities to reassort in Eurasia. In contrast, opportunities to reassort seem to be frequent in sub-Saharan Africa because of great diffi culties in setting up a sensitive surveillance system in a complex socioeconomic environment, where backyard farms and large commercial farms with variable biosafety levels coexist, and where culling may threaten the livelihood and survival of the farm.
If the high prevalence of reassortants was typical for West Africa in 2007, the absence of such reassortants anywhere else suggests that reintroductions of subtype H5N1 from Western Africa into Eurasia must be rare. Moreover, all HPAI (H5N1) strains from Nigeria in 2007 were more similar to those found in Nigeria in 2006 than to even the closest relative from Europe in 2007 (Hungary). Although subtype H5N1 has been found in wild birds from Africa, such as vultures (4), HPAI (H5N1) has so far not been reported in long-distance migrating birds in West Africa. Thus, the exchange of subtype H5N1 between Eurasia and Africa seems to be a rare event, which in 2006 may have been triggered by unusual bird migration as a result of the central Asian cold spell.
The biological signifi cance of reassortments between genetically similar viruses may be arguable, but the frequency of reassortment events is an important marker of virus endemicity in a region. Moreover, endemicity of HPAI (H5N1) and a high propensity of reassorting in a region where seasonal infl uenza is unchecked are essential ingredients of the anticipated pandemic.