Genetic evolution of low pathogenecity H9N2 Avian influenza viruses in Tunisia: acquisition of new mutations

Background Since the end of 2009, H9N2 has emerged in Tunisia causing several epidemics in poultry industry resulting in major economic losses. To monitor variations of Influenza viruses during the outbreaks, Tunisian H9N2 virus isolates were identified and genetically characterized. Methods The genomic RNA segments of Tunisian H9N2 strains were subjected to RT-PCR amplifications followed by sequencing analysis. Results Phylogenetic analysis demonstrated that A/Ck/TUN/12/10 and A/Migratory Bird/TUN/51/10 viruses represent multiple reassortant lineages, with genes coming from Middle East strains, and share the common ancestor Qa/HK/G1/97 isolate which has contributed internal genes of H5N1 virus circulating in Asia. Some of the internal genes seemed to have undergone broad reassortments with other influenza subtypes. Deduced amino acid sequences of the hemagglutinin (HA) gene showed the presence of additional glycosylation site and Leu at position 234 indicating to binding preference to α (2, 6) sialic acid receptors, indicating their potential to directly infect humans. The Hemagglutinin cleavage site motif sequence is 333 PARSSR*GLF341 which indicates the low pathogenicity nature of the Tunisian H9N2 strains and the potential to acquire the basic amino acids required for the highly pathogenic strains. Their neuraminidase protein (NA) carried substitutions in the hemadsorption (HB) site, similar to those of other avian H9N2 viruses from Asia, Middle Eastern and human pandemic H2N2 and H3N2 that bind to α -2, 6 -linked receptors. Two avian virus-like aa at positions 661 (A) and 702 (K), similar to H5N1 strains, were identified in the polymerase (PB2) protein. Likewise, matrix (M) protein carried some substitutions which are linked with increasing replication in mammals. In addition, H9N2 strain recently circulating carried new polymorphism, "GSEV" PDZ ligand (PL) C-terminal motif in its non structural (NS) protein. Two new aa substitutions (I) and (V), that haven't been previously reported, were identified in the polymerase and matrix proteins, respectively. Nucleoprotein and non-structural protein carried some substitutions similar to H5N1 strains. Conclusion Considering these new mutations, the molecular basis of tropism, host responses and enhanced virulence will be defined and studied. Otherwise, Continuous monitoring of viral genetic changes throughout the year is warranted to monitor variations of Influenza viruses in the field.


Background
Avian Influenza virus (AIV) is a member of the family Orthomyxoviridae; containing negative sense single stranded RNA [1]. Based on their pathogenecity, Two types of AIV have been described, namely a highly pathogenic type (HPAIV) that causes severe disease with high mortality, and low pathogenic type (LPAIV) inducing asymptomatic or mild infection [2]. AIV subtypes, namely H5, H7 and H9, currently endemic in poultry in some regions of the world, have been shown to be capable of infecting humans [3,4].
By 1997, H9N2 viruses have been isolated in multiple avian species including chickens, ducks, turkeys, quail, geese and pigeons, throughout Asia, the Middle East, Europe and Africa and for the first time from humans in Hong Kong and China, in 1999 [5][6][7]. They could emerge as human pathogens through reassortment in intermediate hosts, such as pigs [6] and in avian species, or through direct adaptation in human host [8]. These viruses produce significant disease problems in poultry resulting in great economic losses due to reduced egg production or high mortality with co-infection with other opportunistic pathogens such as infectious bronchitis viruses [9], Staphylococcus aureus, Avibacterium paragallinarum, Escherichia coli, or immune suppression [10].
In addition, some of these currently circulating H9N2 strains have acquired human-like receptor specificity (α2, 6 sialic acids) [13] and there is concern about the potential risk for these viruses to cross the species barriers and affect human health as the consequences of genetic reassortments between mammalian and avian influenza virus.
Since the end of 2009, H9N2 has emerged in Tunisia causing several epidemics in poultry industry resulting in major economic losses. It should be noted that Tunisia was reportedly free from Avian Influenza infection [14].
In the present study, we report the isolation of two Tunisian H9N2 influenza viruses that were identified, and their genome sequences were analyzed. Phylogenetic analysis characterization was performed in comparison to H9N2 sequences available in the Genbank database.

Materials and methods
Fecal and tissue samples were collected from affected commercial poultry and from migratory bird during the initial outbreak. H9N2 avian influenza viruses were isolated and identified using classical laboratory methods [15]. The isolates were propagated in specific-pathogen free (SPF) chicken embryonated eggs via the allontoic route, and the eggs were incubated for 72-96 h at 37°C. The collected allontoic fluids were subjected to hemagglutination and hemagglutination inhibition tests using H9 monospecific polyclonal antisera obtained from F.A. O. The genomic RNA segments of viruses were extracted using viral Trizol LS reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's instruction. The first-strand cDNA was synthesized using SuperScript™ III Reverse transcriptase (Invitrogen, Carlsbad, CA) and appropriate upstream primers. cDNA was amplified by PCR using described IA specific primers listed in Table 1. 5 μl of cDNA was added to the PCR reaction mixture (25 μl) containing 2, 5 μl of 10× PCR buffer, 2, 5 μl of 2, 5 mM dNTPs, 0, 5 μl Taq   Table 1 Primer sequences as used in the RT-PCR in our study  [19,20], 72°C for 1 min followed by 72°C for 10 min. PCR products were purified with gel DNA purification kit (GenClean ® II kit, North America, MP). ABI Big Dye Terminator version 1.1 sequencing Kit run on 3730 XL DNA Analyzed (ABI prism 377, DNA sequencer, Applied Biosystem Inc., CA, USA) sequencer was used for sequencing PCR products.
The Bioedit program 5.0.6 software and ClustalW alignement algorithm (version 1.83) were initially used to compare and align nucleotide sequences. Phylogenetic trees were constructed using MEGA5.01 program version 3.65 with neighbor-joining method. The robustness of the groupings in the neighbor-joining analysis was assessed with 1000 bootstrap resamplings. The Blast software and Bioedit programs were used to determine the sequence similarity of the Tunisian strains Nucleotide sequence accession numbers: the nucleotide sequences obtained in this study were deposited in GenBank data library under accession numbers: JF323006 to JF323016.
Number of amino acid residues: amino acid residues were numbered according to the HA sequences of Qu/ HK/G1/97 (H9) with GenBank accession number AF156378.
Phylogenetic analyses based on the HA ( Figure 1a) and NA (Figure 1b) genes (nucleotides 125 to 1272 and 961 to 1370, respectively), revealed that current Tunisian (H9N2) viruses are closely related to each other and shared closed relationship with the Middle Eastern strains, especially A/Av/SA/91013406/06, A/Ck/Em/ R666/02, A/Ck/Du/338/01 and Ck/Pak/UL01/08. Percentage similarity scores, calculated from pairwaise alignments, supported the phylogenetic results and revealed a similarity score of more than 96% (Table 3). Tree analyses showed that Middle Eastern strains are grouped in G1 lineage and shared a common ancestor with A/Quail/Hong Kong/G1/1997 isolates.

Phylogenetic analyses of internal genes
Four internal genes (PB2, NP, M and NS) of Tunisian H9N2 strains showed more than 96% nucleotide identity with those of the Middle Eastern strains isolated from 2001 and 2008 (Table 3).
The phylogenetic studies of the PB2 genes were analysed using the 540-base-long nucleotide sequence   (nucleotides 1591 to 2130) that codes the region of PB2 protein located between positions 530 and 710. Tree analysis of the PB2 gene showed that Ck/TUN/12/10 and Migratory bird/TUN/51/10 isolates are related to the Middle Eastern isolates (more than 97% similarity) and fall into a single group within the G1-like lineage ( Figure 1c). In the NP gene tree (Figure 1d), representative H9N2 viruses clustered into two distinct lineages, the G1-lineage including Middle Eastern, Qa/HK/G1/97, Asian strains, and Dk/HK/Y283/97 H5N1 like lineage. Besides, the H9N2 Tunisian isolates shared 97% to 98% similarity with other isolates such as A/Av/SA/91013406/06, A/ Ck/Em/R666/02, A/Ck/Du/338/01 and showed a "sister group" relationship with H6N1 viruses of more than 97%.

Molecular characteristics
To analyze the receptor binding site, the connecting peptide and the Hemadsorption site of neuraminidase specificity of the Tunisian isolates, we have compared the HA amino acid sequences (deduced from the nucleotide sequences) with those of representative viruses from GenBank.

Hemagglutinin
The molecular determinants of pathogenecity and virulence of the HA protein are the HA1/HA2 connecting peptide sequence, specific amino acids (aa) residues at the receptor binding site (RBS), and the presence or absence of glycosylation sites around the receptor binding site.
HA cleavage site motif sequence of the Tunisian isolates was 333 PARSSR* GLF 341 which is similar to other Asian H9N2 viruses including those from Hong Kong, Japan, Iran, Israel, Pakistan, and Saudi Arabia, Dubai. Residues at positions 110, 161,163,191,198,234,235 and 236 are major components of the receptor binding site of the HA molecule. Tunisian viruses showed conservation of residues P 110 , W 161 , T 163 , H 191 , A 198 and I 235 in the receptor binding pocket. The left (aa residues at position 234-239) and right (aa residues at position 148-152) edge of the binding pocket motif were: NGL/ QIGR and GTSKS, respectively.
The A/CK/TUN/12/10 H9N2 strain carried the amino acid substitution (Q234L), a mutation that correlates with a shift in affinity of the HA from avian type sialic receptors to human type and from a preference for a 2 .. 3 . link to a 2 .. 6 . link between the sialic acid residues and galactose, whereas A/Migratory Bird/TUN/51/10 having Glu Q at this position. But, both strains had Gly (G) at position 228 at the RBS.
Analysis of HA protein sequences showed that Tunisian H9N2 isolates have many potential glycosylation   Table 4).

HB site of Neuraminidase
Analysis of the neuraminidase active and framework sites of the NA protein revealed mutations in aa residues on 3 loops that interact directly with sialic acid; on loops carring amino acids 367-370-372, the three Ser (S) were substituted by KLA respectively, also D and Y substitution were found at residues 401, 406 respectively. Otherwise, the framework site contains R 371 , A 372 , N 402 , and E 425 ( Table 5).

Molecular characteristics of internal proteins
The current Tunisian strains displayed Glu (E) at position 627 of the PB2 protein; therefore, a change to Lys could serve as a marker of the virulence phenotype. In addition, two avian virus-like aa at positions 661 (A) and 702 (K), similar to Ck/HK/8733/01 H5N1 strain isolated in 2001, were found ( Table 5).
The nucleoprotein (NP) of Tunisian H9N2 strains retained conserved G1-lineage defining residues: R422, K430, T442, D455 and D480. However, Ser to Phe substitution at residue 402, found exclusively in Tunisian strains, similar to H5N1 strain circulated in 1997. Other aa specific for the nucleoprotein (NP) gene associated with host range, were not studied ( Table 5).
Alignment of the M2 protein of current Tunisian H9N2 strains showed conserved G1 lineage defining Leu residue at position 10. None of the Tunisian isolates contained substitutions at amino acid positions 26, 27, 30, 31 or 34, suggesting the absence of resistance to M2 blockers antiviral drugs such as amantadine. However, we have identified some substitutions at position 16 (Gly, G), 28 (Ile, I), related to the ion channel domain and associated with host range. In addition, Ile, Val and Pro substitutions were present in Tunisian H9N2 viruses at residue 15 (V15I), 46 (L46V) and 54 (S54P) of M1 protein, respectively. Avian virus like aa such as: V 115 , T 121 and T 137 were also observed ( Table 6).
Based on the analysis of the non structural protein gene (NS), our recent isolates did not show the five amino acid (80TIAS84) deletions, resulting in an NS1 protein of 230 amino acid in length and containing the "GSEV" PDZ ligand (PL) C-terminal motif, and did not harbor the mutation of Asp (D) to Glu (E) at position 92, required for high virulence. Besides, The RNA-binding domain exhibited amino acid residues P31, D34, R35, R38, K41, G45, R46 and T49. However, the effector domain carried the F103L, R184G and K217N   substitutions. Moreover, E227G mutation in C-terminal domain of the NS1 protein, introducing an S70I mutation into nuclear export protein (NEP), was also observed ( Table 6).

Infections of domestic poultry, especially chickens with H9N2 subtype of avian influenza virus (AIV) have been frequently reported in China, other Asian and North
American countries, since the late 1990s. They have been grouped in different sublineages on the basis of antigenic and genetic properties. Although, H9N2 viruses do not satisfy the criteria for highly pathogenic avian influenza, they are unique among this category, infecting a wide variety of species, including chickens [21,22], quails [23], pigeons [24], turkeys [25], ducks [25][26][27], geese, pigs [28], and humans. Several studies have reported that G1-like H9N2 viruses can infect humans and can replicate in human alveolar epithelial cells and mouse respiratory system. Interestingly, they could emerge as human pathogens though reassortement in intermediate hosts, such as pigs and avian species, or through direct adaptation to human. Since late 2009, H9N2 influenza A viruses have caused many outbreaks in Tunisian flocks. A second wave of AI was reported during July-October 2010. Here, we have provided the first comprehensive genetic data for the H9N2 subtype viruses circulating in Tunisian poultry flocks. Strains were collected from two different governorates situated in the northeastern part of Tunisia. Our findings revealed that H9N2 virus infection is well established in some Tunisian endemic areas. Therefore, understanding the genetic and the biological characteristics of H9N2 virus isolated from different species and regions can provide a comprehensive insight into the biology of H9N2, the ecology of AI virus, and the ability of migratory birds to disseminate influenza viruses.  The Blast analysis (NCBI) of the nucleotide sequences of HA and NA genes showed that A/Ck/TUN/12/10 and A/Migratory Bird/TUN/51/10 were the most closely related to the Middle Eastern isolates belonging to the G1-like lineage in the H9N2 subtype (more than 96.5% similarity).
The results of the phylogenetic analyses were basically in agreement with the blast data, and confirmed that our isolates fall, together with the Middle Eastern strains, into a distinct cluster, related to the G1 lineage; a finding that may indicate a common origin.
The internal PB2, NP, M, NS genes were also similar to those of other Middle Eastern strains, which can be traced back to the same G1-like lineage. In addition, NP and NS genes seemed to have undergone broad reassortments with H5N1, H7N3, H7N1, H6N1 influenza virus subtypes as described for previous isolated strains [18,29,30]. This diversity reflects an increased spread of such viruses through avian species' migration. It is difficult to explain whether and how these reassortments have occurred because of the lack of local and regional epidemiological information.
Based on the deduced amino acid sequences, the HA1-HA2 connecting peptides of the Tunisian avian H9N2 isolates did not harbor multiple basic amino acids: PARSSR/GL as found for recently isolated H9 viruses in Middle East [31] and Asia [32,33]. This might indicate the LPAI nature of H9N2 strains, although the motif for these viruses is similar to the RX-RYK-R required for the highly pathogenic H5 and H7 subtypes [34,35]. These genetic findings suggested that our H9N2 viruses may have the potential to acquire basic amino acids in the HA connecting peptide sequence needed to become highly pathogenic through the addition of single basic amino acid at the -4 position.
Moreover, A/Migratory bird/TUN/51/10 possessed a Glu (Q) at position 234 in the HA1 portion (H9 numbering; 226 in H3 numbering), whereas A/Ck/TUN/12/ 10 showed a leu (L), a receptor binding site residue, typical for human influenza virus displaying human virus-like cell tropism through an association with a preferential binding of sialic acid (SA) to galactose in α 2, 6 linkage [9,36]. A leu (L) residue at position 234 in the HA receptor binding site (RBS) was found to be  important for the transmission of the H9N2 viruses in ferrets [37]. In previous study, Wan suggested that Q234L substitution, found in G1, Y280 and G9 lineages, isolated in Hong Kong, allows H9N2 viruses to preferentially infect non ciliated cells and grow more efficiently in human airway epithelial cell cultures; thus, increasing the infection severity in humans [12]. Our findings suggested that A/Ck/TUN/12/10 may have segregated from the migratory A/Migratory bird/TUN/51/ 10 strain acquiring an affinity for the human receptor binding profile. These findings may be attributed to a particular introduction of a new virus that has been circulating within poultry species.
Remarkably, residue at position 198 within the receptor binding pocket has been reported to influence the affinity of virus binding to SA receptor; high affinity to the human like receptor being with V at position 198, intermediate with T and low with A [38]. It can be predicted that the Tunisian A/Ck/TUN/12/10 isolate has a weaker affinity binding to human like receptor, but this finding need to be further confirmed by experimental studies.
Analysis of potential glycosylation site motif N-X-S/T in the HA1 protein of Tunisian H9N2 isolates, revealed some sites at positions 82, 105, 141, 298, 305. Additional glycosylation site at position 168 was observed compared with representative reference strains. It has been suggested that the alteration in the glycosylation pattern influences the adaptation of avian influenza viruses to poultry by altering their pathogenecity and antigenicity and helps to the evasion from the host antibody response [39][40][41].
The NA of the new Tunisian isolates carried substitutions in the HB site, similar to those of other avian H9N2 viruses isolated in Asia and Middle Eastern during human pandemic H2N2 and H3N2 that bind to α -2, 6 -linked receptors [42]. These viruses were showed to be under a positive selection pressure, resulting in compatible combinations of HA and NA [38].
Based on the deduced amino acid sequence, the Tunisian H9N2 isolates displayed Glu (E) at position 627 of the PB2 protein. Likewise, a single aa substitution in PB2 protein (E627K) can dramatically alter the virulence and enhance viral replication in mice [43] and other mammals [44][45][46]. Two avian virus-like aa at positions 661 (A) and 702 (K), as seen in H5N1 strains, were identified in the PB2 protein. These specific aa are located in the functional domain responsible for interaction with other polymerase components [47].
The main function of the NP is encapsidation of the viral genome to form a ribonucleoprotein particle for transcription and packaging; it interacts with other viral PB1, PB2, M1 and cellular proteins (Importin α, F-actin, CRM1/exportin 1) for viral transcription and nuclear transportation controls [48]. The NP aa sequence of the Tunisian isolates retained conserved G1-lineage defining residues. Interestingly, a new mutation (S402F), which has not been seen before, was detected in the NP protein. This mutation was previously found in the H5N1 subtype instead of Ser in all other H9N2 subtypes compared in this study (Table 5).
It has been reported that aa residues at positions 15, 115, 121, 137, 240 in virus matrix protein are linked with increased replication in mammals or increased pathogenecity in small animal models [49,50]; an Ile substitution was present in Tunisian H9N2 viruses at aa position 15 (V15I). No one contained substitutions at aa positions 26, 27, 30, 31 or 34 within the transmembrane domain of M2 protein, maintaining a genotype associated with sensitivity to adamantadine M2 blocker antiviral drugs [51]. It has been known that amantadine binds to the ion channel region of the M2 protein and prevents the release of viral RNA into cells [52]. Tunisian H9N2 strains harbor two human virus-like amino acids at positions 16 (Gly, G) and 28 (Ile, I) which are related to the ion channel domain and associated with host range (49,50), but the role of these substitutions on increased replication efficiency in mammalian cells is not yet known.
A Val substitution at position 46 was also detected for the first time in our strains; however, all other H9N2 and other IA subtypes have leu at this position (Table  6). Interestingly, this substitution should be considered.
Recent studies have suggested that NS1 protein suppresses the host antiviral defenses at multiples levels and correlation between NS1 and virulence has been reported [53][54][55]. Molecular analysis showed that our isolates contained an NS1 protein, with 230 aa in lengh, typical of H9N2 viruses. A recent study has shown that increasing the length of the 2009 H1N1 NS1 protein to 230 aa does not increase virus replication in human and pig cells [56]. Other study showed that viruses containing NS1 truncations were found to induce more interferon than viruses with full-length NS1 proteins and were correspondingly more attenuated in mice [57].
In the RNA-binding domain of NS1, A/Ck/TUN/12/ 2010 isolate contained R38 and K41, which are shown to be critical for RNA binding. as well amino acid residues P31, D34, R35, G45, R46, T49 and D55, which also mediate NS1-dsRNA interaction. Residue 55 is located within the third alpha-helix (residues 54-70) of the dsRNA-binding domain (RBD (residues 1-73) of NS1 [58]. It is documented that variation of NS1-55 from Asp (D) to gly (G) represents loss of changed aa which may stabilize the coiled-coiled helical structure. In addition, our strains exhibited no differences in the second nuclear localization sequences (NLS2) motif, or change in the amino acid D at position 92. Some studies reported a mutation of asp (D) to Glu (E) to be related to virulence of H5N1 in mammalian species and cytokine resistance [59]. Nevertheless, their effector domain carried Leu at position 103 and Gly at residue 184. In fact, F103L and M106I mutations are adaptive genetic determinants of growth and virulence in both human and avian NS1 genes in the mouse model [60]. Likewise, it has been demonstrated that, in addition to its contribution to cleavage and polyadenylation specificity factor (CPSF) binding, Gly 184 strongly influence viral virulence by an unknown mechanism which does not involve the INF system [61].
An Asp (N) was found at residue 217 only in Ck/ TUN/12/2010 strain that differs from the other H9N2 strains which have K at this position, but similar to those of H5N1 strain CK/HK/8761/01 (Table 4). However, the biological significance of this substitution is not yet known.
Also, Ck/TUN/12/2010 didn't exhibit a five amino (80TIAS84) deletion observed in 2001 in poultry in Hong Kong and has, since, became the most common sequence found in the HP viruses. Long et al, demonstrated that viruses containing NS1 with 5 amino acid deletion (80TIAS84) residues showed increased virulence in both mouse and poultry [62] but the biological significance of these deletions is not fully understood yet.
Fortunately, previous study, performed on large scale sequence analysis of viruses isolated from different birds and mammalian species, have identified that the C-terminal domain of NS1 functions as a species-specific virulence domain: the vast majority of avian influenza viruses have an NS1 protein with a PDZ ligand (PL) C-terminal ESEV domain, while typical human viruses have a conserved RSKV domain. NS1 proteins with C-terminal ESEV, KSEV, and EPEV domains were shown to bind to PDZ domains containing cellular proteins [63,64]. Soubies et al demonstrated that RSKV motif, which lacks a PDZ-binding domain, replicated to higher titers than ESEV in humans and ducks cells, suggesting the ability of NS1 to interact with PDZ containing proteins does not contribute to virulence in the host species [65]. Nevertheless, it has been showed that insertion of four C-terminal aa, either ESEV, KSEV, or EPEV, into avirulent viruses resulted in an increase in virus virulence and caused severe disease in mice [52]. The H9N2 viruses recently isolated in Tunisia, have a PL motif "GSEV" previously found in Dubai strains during 2001-2003 (data not published); but the biological signification of this motif is unknown. Interestingly, the E227G mutation in NS1 introduces an S70I mutation into nuclear export protein.

Conclusion
In the present study, we characterized two Tunisian H9N2 strains isolated from disease outbreaks in commercial chicken and migratory bird from different geographic areas. Our findings indicate that CK/TUN/ 12/10 isolate has evolved and reassorted with other influenza viruses and has acquired an affinity for the human receptor binding profile. Otherwise, the biological significance of new detected substitutions founded in some internal genes and the PDZ ligand polymorphism of the NS protein should be tested.
Actually and face to the high mortality, we are preparing for an active program to develop and evaluate vaccines against H9N2 to control the further spread of the disease in the industrial poultry production sectors. Interestingly, continuous monitoring of viral genetic changes throughout the years is warranted to monitor variations of Influenza viruses in the field.