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 , pigeons , turkeys , ducks [25–27], geese, pigs , 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  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 . 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 . 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 . 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 [39–41].
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 . These viruses were showed to be under a positive selection pressure, resulting in compatible combinations of HA and NA .
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  and other mammals [44–46]. 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 .
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 . 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 . 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 . 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 [53–55]. 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 . 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 .
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 . 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 . 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 . 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 .
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  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 . 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 . 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.