In this study, a lectin histochemistry technique was used to extensively assess the expression of α-2,3 and α-2,6 receptors in the respiratory and intestinal tracts of seven domestic avian species. The staining for these two types of receptors was not performed in a single assay and, therefore, it is not possible to accurately determine whether cells co-express both α-2,3 and α-2,6 receptors, but to infer about the expression of these receptors in a particular tissue or cell type. In the chicken, common quail, red-legged partridge, turkey, and golden pheasant, both α-2,3 and α-2,6 receptors were expressed in at least one segment of the respiratory and intestinal tracts. The α-2,6 receptors were not observed in the respiratory tract of the ostrich, nor in the intestinal tract of mallards. The PVA of an avian-origin H4N5 and a human-origin H1N1 influenza virus was also evaluated on the same tissues and compared with the lectin histochemistry distribution pattern. There was a great variation among the tissues and avian species studied (Tables 1 and 2), and, with a few exceptions, the results obtained with the lectin histochemistry were in agreement with the virus histochemistry (Tables 3 and 4).
Virus histochemistry has shown to be a useful assay to study the pattern of virus attachment in different tissues, and the PVA and the lectin histochemistry results were comparatively consistent. When both techniques were compared, the results obtained with the PVA and with the lectin histochemistry were either equal or varied only one grade (Tables 3 and 4). There were only two cases where the lectin histochemistry was graded as strong (+++) and the PVA was graded as low (+), and only three cases where the lectin histochemistry was graded as low (+) and the PVA was graded as strong (+++) (Table 3).
The results obtained in the chicken were in agreement with previous reports, in which a limited number of tissues were evaluated [21, 24, 27]. However in the present study we observed a low level of staining for the α-2,6 receptor on the epithelial cell in the jejunum-ileum and cecum, whereas Liu et al.  did not detect α-2,6 receptors in the intestinal tract of chickens, and Kuchipudi et al.  only detected this receptor in the large intestine. These differences could be attributed to the signal amplification methodology used in our study. On the contrary, the presence of α-2,6 receptors and the binding of the human-origin H1N1 virus to the intestinal tract of chickens was in accordance with the results of an experimental study where a human-origin influenza A virus was able to bind in vitro to chicken colon cells .
In common quails, our results were in accordance with previous reports in common quails  and Japanese quails [23, 25] in which both α-2,3 and α-2,6 receptors were observed in the respiratory and intestinal tracts. Another study with bobwhite quail, however, did not detect an α-2,6 receptor in the intestinal tract . This difference in receptor expression could be related to interspecies differences. Both viruses used for the PVA were able to bind to the respiratory and intestinal tracts of quails, particularly the H4N5 virus in trachea.
In red-legged partridges, low levels of expression of both α-2,3 and α-2,6 receptors were observed in respiratory and intestinal tracts. Assuming this pattern of receptor distribution is seen in other partridge species, it could explain the susceptibility of chukar partridges (Alectoris chukar) to experimental infection with both avian-origin and swine-origin influenza viruses .
In turkeys, both α-2,3 and α-2,6 receptors were detected on epithelial cells along the intestinal tract and the entire respiratory tract, in agreement with recent studies [18, 19, 36]. Regarding the PVA, attachment of the human-origin H1N1 virus was observed in the respiratory (nasal cavity and lung) and intestinal tracts, while the attachment of the avian-origin H4N5 virus was restricted to the respiratory tract. This was consistent with the fact that turkeys are also susceptible to swine-origin H1N1 and H3N2 influenza viruses [37–41], to reassortant viruses with human, swine, and avian influenza genes (H1N2) [42, 43], as well as to the pH1N1 virus, as demonstrated in August 2009 in two turkey flocks naturally infected in Chile , indicating the potential of avian species that express both α-2,3 and α-2,6 receptors, to be susceptible to mammal-origin influenza viruses and potentially offer an adequate environment for the emergence of reassortant viruses. Turkeys have also been successfully infected with the pH1N1 virus by experimental inoculation via the intrauterine route, with subsequent oropharyngeal and cloacal virus shedding, but were not infected when the intranasal route was used . This observation shows that other factors rather than the tissue distribution of receptors and the affinity of virus binding may determine the outcome of an exposure to certain influenza A viruses.
In golden pheasants, the observed expression of both α-2,3 and α-2,6 receptors in the respiratory and intestinal tracts was in accordance with previous reports [19, 26]. This may explain why pheasants can be infected with avian-origin influenza viruses that have specificity to α-2,6 receptors, as observed for some H9N2 influenza virus isolates . In pheasants, the PVA of the avian-origin H4N5 was observed throughout the respiratory and intestinal tract, while the PVA of the human-origin H1N1 virus was restricted to the trachea and intestinal tract. In addition, the fact that ring-necked pheasants could not be experimentally infected with human and swine influenza viruses  indicates that, as mentioned above, besides the hemagglutinin receptor binding site, other factors may be involved in the restriction of inter-species transmission .
In ostriches, the influenza receptor expression was almost exclusively restricted to α-2,3 receptors on epithelial cells throughout the respiratory and intestinal tracts, as with the PVA of the avian-origin H4N5 virus. The expression of α-2,6 receptors was low on GALT lymphocytes, and the PVA of the human-origin H1N1 virus was restricted to a few lining epithelial cells of the small intestine. In fact, there are no reports of ostrich infection with mammal-origin influenza viruses.
In mallards, both α-2,3 and α-2,6 receptors were observed in the respiratory tract, as previously reported [19, 22, 26]. Regarding the intestinal tract, α-2,3 receptors were moderately expressed in mallards, while α-2,6 receptors were absent. A recent study, however, reported a very minimal expression of α-2,6 receptors in the large intestine of the Pekin duck . The absence of abundant α-2,6 receptors in the mallard intestinal tract, associated with the fact that the PVA of the human-origin H1N1 virus was low and restricted to the trachea and small intestine, correlate with the fact that ducks are resistant to the infection with human influenza A viruses under natural and experimental conditions [5, 48].
When interpreting the distribution of influenza receptors based on lectin histochemistry or virus histochemistry it is important to take several factors into consideration. For instance, studies of influenza virus infections in ex vivo cultures indicated that there are alternative receptors that cannot be identified with the lectins used in the present study . Furthermore, ongoing viral mutations and adaptation may change host receptor affinity over time, further limiting lectin staining . Additionally, besides the distribution of receptors, other host and viral factors intervene in the process of viral replication and adaptation of any influenza viruses in a new species, such as host immune response, and viral glycoproteins and internal proteins [47, 51].
Using glycan micro arrays it has been shown that not all α-2,3 or α-2,6 receptors bind to influenza hemagglutinin (HA) proteins equally well; one glycan terminating in α-2,3 might not bind HA while another may bind exceedingly well . Therefore, determining the influenza virus-binding profile in tissues of different animal species is a condition fundamental to better understanding the role of these receptors . However, care must be taken when interpreting and extrapolating PVA results, since the PVA may vary between and within influenza A strains.
In summary, this study demonstrates that although both α-2,3 and α-2,6 receptors are expressed in domestic birds, there is marked variation among species. This information helps to understand the effect of host pressure on virus evolution and indicates that "mixing vessels" is not only restricted to pigs, since other species could also have an important role in the transmission and adaptation of influenza viruses of avian-origin to the mammal host. Hence these poultry species could pose a greater threat to humans, since avian viruses with affinity for α-2,6 receptors may be selected, amplified and transmitted. This could explain why some avian H9N2 strains acquired affinity for α-2,6 receptors in quail after continual circulation in the field [46, 53, 54]. In this respect, it is important to determine the role of concomitantly co-expressed α-2,3 and α-2,6 receptors in the emergence of new viral strains, especially those with pandemic potential.