Surveillance for influenza viruses in poultry began in Bangladesh in November 2008. Each month, trained personnel collected 200–600 samples from apparently healthy domestic live birds (chickens, quail, pigeons, ducks, and turkeys) at retail markets, a pet market, chicken layer and duck farms, and wild birds (50–200 samples from each location). Samples consisted of oropharyngeal, cloacal, and environmental (fecal matter and water samples from cages and fecal digesters) swab specimens.

During November 2008–August 2011, a total of 17,438 samples (3,078 oropharyngeal, 3,377 cloacal, 10,983 environmental) were collected (Table 1). Screening for AIVs was performed as described (26). Briefly, all samples were subjected to real-time reverse transcription PCR by using influenza A–specific primer and probes; samples with positive results were injected into egg to confirm the presence or absence of virus. A total of 734 AIV (H9N2) isolates were extracted from the 17,438 samples collected (Table 1).

To determine the genetic and evolutionary diversity of the AIV (H9N2) viruses circulating in Bangladesh, we selected 44 of the 734 isolates for further analysis (Table 2); these isolates were representative of the location, time, species, and sample types from which they were isolated. With the exception of 1 isolate from a chicken on a farm, all isolates examined were from chicken, duck, and quail samples collected at retail markets. The viruses were propagated in 10-day-old embryonated chicken eggs, and initial subtyping was done by sequencing the hemagglutinin (HA) and neuraminidase (NA) genes, as described (26).

Polyclonal serum samples were obtained from ferrets that had been inoculated with 1 of the following: prototype viruses from different influenza (H9N2) lineages; influenza (H9N2) isolates from chickens or quail collected in Bangladesh; or a prototype human isolate provided by the US Centers for Disease Control and Prevention. We then conducted hemagglutination inhibition (HI) assays for these serum samples by using 0.5% chicken erythrocytes, as described (27).

Viral RNA extraction and reverse transcription PCR were performed as described (28,29). DNA sequencing using specific primers was completed in the Hartwell Center at St. Jude Children’s Research Hospital (Memphis, TN, USA). Samples were analyzed by using the BigDye Terminator v3.1 Cycle Sequencing Kit (Invitrogen, Carlsbad, CA, USA) on 3730XL DNA analyzers (Applied Biosystems, Foster City, CA, USA), according to the manufacturers’ recommendations. DNA sequences for all genes were edited, compiled, assembled, and analyzed by using SeqMan in Lasergene 8 (DNASTAR, Madison, WI, USA). Nucleotide sequences were compared with sequences available in GenBank. Multiple nucleotide sequence alignment and the alignment of the deduced amino acids of all of the gene segments were conducted by using ClustalW in BioEdit 7.09.0 (www.mbio.ncsu.edu/bioedit/bioedit.html). The phylogeny of each gene was investigated within the maximum-likelihood framework by using PhyML under the general time reversible substitution model with gamma rate heterogeneity (30). The robustness of the grouping was assessed by using 100 bootstrapping replicates (30). MegAlign in Lasergene 8 (DNASTAR) was used to determine the percentage of nucleotide sequence similarities, and the NetNGlyc 1.0 Server (31) was used to predict the glycosylation sites in the HA and NA genes. The sequences we obtained (Technical Appendix, Table 1) were deposited into GenBank under accession nos. KC757782–KC758115.

During November 2008–August 2011, a total of 17,438 samples were collected from live-bird markets, farms, and wild and environmental sources (26). In the markets and farms, samples were predominantly collected from chicken, quail, and ducks. The environmental samples were collected from cages in the markets in which only 1 species was housed (e.g., only chickens or only ducks). The wild bird samples were from a lake that was a feeding station for wild migratory ducks.

The predominant subtypes identified in these samples were low pathogenicity H9N2 (n = 734) and highly pathogenic H5N1 (n = 92) (Table 1). Other influenza subtypes identified, at low percentages and only during 2008–2009, were H1N2, H1N3, H3N6, H3N8, H4N2, and H10N7 (Table 1). None of the wild-bird samples tested and only 35 (0.87%) of 4,045 farm samples had positive test results for avian influenza (Table 1). The subtype H9N2 and H5N1 viruses were isolated from domestic poultry, mainly chickens and quail that were sampled in live-bird or pet markets (Table 1).

The number of subtype H5N1 viruses isolated from field samples increased during the surveillance period (Figure 1, panel B). The numbers of other influenza subtypes identified were low during 2009–2010; during 2010–2011, subtypes H9N2 and H5N1 were most commonly isolated.

AIVs and Newcastle disease virus (NDV) are the 2 most common viruses that infect poultry in Bangladesh; therefore, all of the influenza-positive samples that were subtyped and used in this study were screened again for H9N2, H5N1, and NDV to determine whether a single host species was co-infected with AIV and NDV. In a substantial proportion of the isolates screened, we identified mixtures of either subtype H9N2 and NDV or subtypes H9N2 and H5N1 (Table 2). All of the samples containing a mixture of viruses were obtained from apparently healthy birds. The highest incidence of influenza (H5N1) virus isolation was during the winter months (January–March), whereas subtype H9N2 was isolated year-round (Figure 1, panel A)

Antigenic analysis showed that most chicken H9N2 viruses circulating in Bangladesh were antigenically homogenous and similar to the human H9N2 virus (A/BD/0994/11) isolated from Bangladesh (Table 3, Appendix). Antigenic variation was seen in subtype H9N2 viruses isolated from quail; specifically, A/Env/BD/907/2009 (H9N2) (quail) was distinguishable from other contemporary quail isolates when postinfection ferret serum of A/Env/BD/907/2009 (H9N2) (quail) and chicken H9N2 viruses was used (Table 3, Appendix). The chicken H9N2 viruses had moderate cross-reaction to antiserum of A/Env/BD/907/2009 (H9N2) (quail), indicating slight antigenic variation from subtype H9N2 viruses isolated from quail (Table 3, Appendix). The low level of cross-reactivity seen across the quail viruses and between the quail and chicken viruses indicates that the HAs of the subtype H9N2 viruses maintained in quail were antigenically distinct from those of subtype H9N2 viruses maintained in chickens. In addition to the quail isolate, A/Env/BD/12068/2011 (H9N2) (pigeon) reacted with 2 serum samples, the polyclonal serum against A/Env/BD/907/2009 (H9N2) (quail) and that against A/Ck/Pk/(NARC-2434)/06 (H9N2). The HI titers of those reactions were 320 and 160, respectively.

These antigenic analyses show that antigenic variation occurs among influenza (H9N2) viruses isolated from different host species. The AIV (H9N2) viruses from Bangladesh cross-reacted well to antiserum against G1 clade H9N2 viruses that were isolated in Bangladesh, Dubai, or Pakistan; however, they showed low reactivity to antiserum against A/Qa/HK/G1/97 (prototype G1), indicating that the viruses from Bangladesh were related to but very distinct from the G1 clade isolates. These results indicate antigenic diversity within the G1 clade.

The phylogenetic relationships of all 8 genes of the 44 representative AIV (H9N2) viruses were analyzed on the basis of their nucleotide sequences and by comparing them with nucleotide sequences of subtype H9N2 viruses belonging to the main lineages from Asia (14). All sequences from Bangladesh except 1 formed a monophyletic cluster. Phylogenetic analyses showed that all the subtype H9N2 viruses isolated from Bangladesh are part of the lineage represented by A/Qa/HK/G1/97 (G1), which is the most dominant lineage worldwide. The HA, NA, nucleoprotein (NP), matrix (M), and polymerase PB2 genes of the viruses from Bangladesh originated from the prototype G1 virus A/Qa/HK/G1/97 (H9N2). The nonstructural (NS) and polymerase PA and PB1 genes were closely related to those in the subtype H7N3 isolate from Pakistan (14).

Phylogenetically, the HA and NA genes of most of the subtype H9N2 viruses from Bangladesh tightly clustered with those of viruses from India and Pakistan (Figure 2; Technical Appendix Figure). These viruses had >95% sequence homology, grouped within the G1 lineage, and shared a common ancestor with A/Qa/Hk/G1/97. One duck isolate, A/Env/BD/1041/09 (H9N2) (duck), clustered with subtype H9N2 viruses from Korea and shared a common ancestor with A/Dk/HK/Y439/97 (Figure 2). This isolate also showed only 85% nt similarity with the other viruses from Bangladesh.

Three internal genes (NP, M, and PB2) of the subtype H9N2 viruses from Bangladesh showed a high level of nucleotide homology with the G1-like lineage. All 3 genes were phylogenetically closely related to viruses from India, Pakistan, or the Middle East (Technical Appendix Figure). The duck isolate A/Env/BD/1041/09 (H9N2) (duck) showed similar characteristics in the HA and NA trees, where it shared a common ancestor with a lineage from Korea. The other 3 internal genes (NS, PB1, and PA) of all the subtype H9N2 viruses from Bangladesh grouped into a single cluster, sharing high sequence homology with A/Ck/Karachi/NARC-100/04 (H7N3) (Figure 3; Technical Appendix Figure). These genes formed a distinct clade adjacent to the G1 and Ck/Bei lineages. Although subtype H5N1 viruses were isolated separately and in a mixture with subtype H9N2 viruses, phylogenetic analysis showed no reassortment between the isolates characterized in this study and the subtype H5N1 strains.

To identify the possible determinants of transmission of AIV (H9N2) from birds to humans, we aligned the amino acid sequences of all genes of the subtype H9N2 viruses from Bangladesh and compared them with those of representative subtype H9N2 viruses from different clades (Technical Appendix Tables 2, 3). When compared with the prototype G1 virus, the viruses from Bangladesh showed that they have evolved to acquire mammalian host–specific mutations throughout the genome (Technical Appendix Tables 2, 3). Comparing the amino acid sequences of all genes of the subtype H9N2 viruses from Bangladesh with the prototype subtype H9N2 viruses showed that certain amino acid substitutions throughout the viral genome have become fixed as the viruses have evolved (Figure 4). The receptor-binding site (RBS) of the virus HA influences the generation of human viruses from avian precursors (1). Within the RBS, 42 (95.5%) of 44 isolates had leucine (L) at position 226 (H3 numbering), whereas the other 2 viruses had glutamine (Q) at the same position. Of the 2 isolates with 226Q, 1 was a duck virus isolated in early 2009 and 1 was a quail virus isolated in 2011. All viruses with 226L were isolated during 2009–2011 and were found in all of the poultry species sampled. Amino acids at positions 183, 189, 190, and 226 (H3 numbering) are located within the RBS of the HA protein, and the combination of these 4 residues is essential for respiratory droplet transmission of a subtype H9N2 or H3N2 reassortants in ferrets (6). Of the 44 viruses analyzed, 42 had 2 (183H and 226L) of the 4 (183H, 189H, 190E, 226L) substitutions. Variations in the number of potential glycosylation sites within the HA are thought to be associated with the adaptation of duck viruses to poultry (7). Seven potential glycosylation sites (29, 105,141, 298, 305, 492, and 551) were found in the HAs of 43 (97.7%) of the 44 isolates; the N-X-T/S motif (in which X may be any amino acid except proline) and 2 potential glycosylation sites were lost at positions 208 and 218 (Technical Appendix Tables 2, 3). In the NA proteins of all of the analyzed viruses, we found no R292K substitution, which is associated with resistance to the neuraminidase inhibitor oseltamivir.

Sequence analysis of the internal gene products also identified mammalian host–specific markers (11,14) that were fixed in the M protein (M1 V15I, 43 isolates; M2 L55F, 44 isolates) and NP (11,14) (R214K, 37 isolates; E372D, 40 isolates; PB1-L13P, 40 isolates) (Figure 4). In the NS protein, all isolates were full length and possessed the PDZ-(PL) C-terminal motif (11,14). The NS1 protein also harbored the mammalian-specific glutamate (E) to lysine (K) substitution at position 227 in 43 of the 44 isolates tested (Figure 4).

Our comparison of the deduced amino acid sequences of the subtype H9N2 viruses from Bangladesh revealed that, in positions that were previously identified as important for host specificity, unique substitutions whose functions are unknown have become fixed. In the RBS of the HA protein, the Q227I substitution was in 35 of 44 isolates. This substitution was previously identified in subtype H9N2 viruses from Pakistan and the Middle East but was not found in any of the reference strains. The other unique substitutions were in the M2 protein (E16V and I28F, 43 isolates) and the PB1-F2 protein (R79L, 40 isolates) (Figure 4). Alignment of the M2 proteins showed that 41 of the 44 isolates had a substitution at position 31 (S31N), suggesting resistance to M2-blocker antiviral drugs such as amantadine.

The amino acid sequence of the HA1/HA2-connecting peptide of HA is a major determinant of pathogenicity in terrestrial poultry, and the pathogenicity of highly pathogenic subtype H5N1 and H7N3 viruses is influenced by the presence of a polybasic cleavage site in the connecting peptide (32). The amino acid sequences of the cleavage site of the HA1/HA2 junction in the subtype H9N2 viruses from Bangladesh possessed 4 cleavage motifs (Figure 4). Of the 44 isolates we tested, 28 had the KSKR/GLF motif (identified in all recent poultry isolates and a 2011 quail isolate); 14 isolates had the KSSR/GLF motif (identified in chickens, ducks, and older quail isolates); and 1 duck and 1 pigeon isolate had the ASDR/GLF and KASR/GLF motifs, respectively. Thus, the altered amino acid sequences in the connecting peptide of viruses we analyzed were distinct from the RSSR/GLF motif found in the connecting peptides of prototype subtype H9N2 viruses. This finding is clear evidence of genetic variation within isolates from different poultry species.