Lectin-based staining assays were used to determine the variations in sialic acid form and tissue distribution in various poultry species. Trachea, lung, and large intestine from 6 minor poultry species were used to determine the distribution of SA receptors. Ducks were included as a control as it has previously been reported that they show predominantly α2,3 SA in the trachea with increasing α2,6 on epithelial lining farther along the respiratory tract and only minimal α2,6 in the large intestine [10]. All other species were chosen for their presence in live poultry markets across the world.

The results indicate that there is a distinct difference between waterfowl (duck and goose) and land-based poultry (pheasant, turkey, bobwhite quail, and guinea fowl) (Table 1) in terms of presence and distribution of SA receptors, particularly α2,6. There were also age-based differences observed, particularly in turkeys (Table 1).

In the trachea, the ducks showed moderate to high levels of α2,3 SA (Table 1 and Figure 1A, B, C), consistent with previous reports [10,18]. There was no expression of α2,6 SA, consistent with one report [10], but not the other [18]. The geese trachea also showed an abundance of α2,3 SA and absence of α2,6 SA at any age (Table 1 and Figure 1D, E, F). On the contrary, the four land-based species showed both forms of sialic acid at all ages tested with positive staining of mucin-producing cells lining the lumen of the trachea (Table 1 and Figure 1G-R). Farther down the respiratory tract, the lungs (Figure 2) tested positive for both SA forms in all birds of all ages with the only exception being in the goose. Staining was present on cells lining the lumen of the lungs. Strong positive staining for both types of SA receptors was observed in the lungs of turkeys (Figure 2J, K), consistent with the observation of influenza outbreaks in turkeys caused by swine influenza viruses with human-like receptor specificity. The lungs of guinea fowl showed also significant staining for both SA receptors, which is consistent with the circulation in these birds of H9N2 viruses with human-like receptor specificity. At 4 weeks of age, no α2,6 SA was detected in the goose's lung (Figure 2E, F). However, both α2,3 SA and α2,6 SA were seen in the lung samples from geese at weeks 1 and 2 (not shown).

Testing of the large intestine once again showed a divide between the species. All six species tested positive for α2,3 SA in the large intestine in cells facing the lumen (Figure 3). However, duck, goose, and pheasant large intestine also showed minimal positive results for α2,6 SA (Figure 3B, E and 3H) while turkey, guinea fowl and quail tested negative (Figure 3K, N and 3Q; please note that significant α2,6 SA staining was observed on the basolateral side - opposite to the intestinal lumen - of epithelial cells in guinea fowl.)

The birds can be divided into three groups based on the distribution of sialic acids in the tissues examined. The waterfowl, the natural host of avian influenza viruses, show predominantly α2,3 SA in their tissues. α2,6 SA is only seen in the lower respiratory tract and minimally in the large intestine. The land-based birds also express α2,3 SA in all the tissues tested, however, they also express significant levels of α2,6 SA in the upper respiratory tract. This could help explain why these birds are susceptible to AIVs resulting in the emergence of strains with altered receptor specificity, including with human-like receptor binding [19]. This also underscores the potential role of these birds in influenza virus reassortment. Finally, the pheasants showed α2,6 SA in the trachea similar to the other land birds, but also showed α2,6 SA in the large intestine like the aquatic birds. This could make the pheasant more likely than other species to facilitate viral reassortment or to act as a "mixing bowl" species.

While performing the experiments described above a trend was noticed in three species. The ducks and geese showed an increasing expression of α2,6 SA in the large intestine as they aged. Similarly, an increase in α2,6 SA detection was seen in the trachea of turkeys as they aged. The age dependence in turkeys was later reported by Pillai and Lee [18], however, they did not see any increase in α2,6 in Pekin ducks. There was no detection of α2,6 SA in the large intestine of ducks and geese at week 1 (Figure 4J for duck, not shown for geese). However, by week 2 there was a very low level positive reaction and at week 4 this reaction was slightly increased (Figure 4K and 4L arrows). Expression levels of α2,3SA remained relatively constant (Figure 4G-I) at all three time points.

In the turkey trachea this change in expression was even more pronounced. At week 1 (Figure 4D) only minimal α2,6 SA was detected. A week later (Figure 4E) there was a moderate positive response. By week 4 (Figure 4F) there was high level of expression. Again, there was no change in expression α2,3 SA at all time points (Figure 4A-C). No major age-related changes were observed in the other avian species tested for either 2,3 or 2,6 SA expression. This changing receptor pattern could have effects for live attenuated vaccines against viruses with a α2,6 binding preference in young turkeys and in ovo inoculations.

Glycan micro arrays have shown that not all α2,3 SA or α2,6 SA bind to influenza HA proteins equally well [15]. One glycan terminating in α2,3 SA might not bind HA while another may bind exceedingly well [15]. Unfortunately, both will show a positive reaction to the lectin-binding assays. Thus, determining the influenza virus-binding profile in tissues of different animal species is a condition sine qua non to better understand the role of these receptors.

Three H3N2 influenza viruses were selected to determine the correlation between lectin binding and virus binding using 3 prototypic H3N2 viruses to ensure differences were due to receptor specificity and not differences between subtypes. To determine the binding affinity of each virus, hemaglutinin agglutination assays were performed for each virus. According to previous reports, horse red blood cells (RBCs) express solely α2,3 SA on their surface while pig RBCs express predominantly α2,6 SA[20]. By comparing HA titers determined with each blood type, a binding preference can be ascertained. A/Dk/HK/7/75 (A/Dk) is a typical AIV duck isolate that bound horse RBCs twice as readily as pig RBCs, indicating a strong α2,3SA preference (Table 2). A/Tk/OH/313053/04 (A/Tk) was isolated from a turkey and bound pig RBCs slightly higher than horse RBCs, indicating a slight preference for α2,6SA (Table 2). A/Memphis/31/98 (A/Mem) is a human origin virus that shows no α2,3SA binding[21]. Accordingly, A/Mem only showed HA titer with the pig RBCs (Table 2). Using these three viruses we were able to determine the accuracy and resolution of the lectin binding results.

The trachea of the duck and geese showed no α2,6 SA. The virus-binding assay showed no binding to the A/Mem or the A/Tk viruses (Figure 5A, D, J and 5M). Additionally, there was minimal binding of A/Dk to the duck trachea (Figure 5G) and no virus binding of the A/Dk to the goose trachea despite ample expression of α2,3SA (Figure 5P). This is not unexpected as the typical route of infection in waterfowl is through the cloacae. In contrast, pheasant and turkey trachea exhibited the ability to bind all three viruses (Figure 6A, D, G, J, M and 6P). Based on fluorescent intensity and distribution of the fluorescent signal, in the pheasant the A/Dk virus showed the lowest levels of binding while the turkey showed equal binding between the three viruses. The quail trachea showed low binding with A/Dk and A/Tk, and no binding of the human A/Mem virus (Figure 7J, M, and 7P). The guinea fowl, on the other hand showed low levels of binding with A/Mem but no binding with A/Dk or A/Tk (Figure 7A, D and 7G).

To visualize the virus binding in the lungs, we imaged transversal sections of the parabronchi to minimize variations from section to section and from species to species. Whenever virus was seen in these sections, it was seen binding to the smooth atrial muscles lining the parabronchi regardless of bird species or virus. The lungs of ducks showed moderate binding of A/Tk and A/Mem but no binding to A/Dk (Figure 5B, E and 5H). The goose lung however showed binding with A/Dk (Figure 5K) but no binding with the other two viruses (Figure 5N and 5Q). Pheasants showed no binding of any virus in the parabronchi (Figure 6B, E and 6H). Turkey showed low to moderate binding of A/Dk and A/Tk but no binding of A/Mem (Figure 6K, N and 6Q) while the guinea fowl had A/Dk binding but neither of the other two viruses (Figure 7B, E and 7H). Finally the quail were the only species to show binding of all three viruses in the lungs (Figure 7K, N and 7Q).

Despite the fact that all birds expressed α2,3 SA in the intestines, only the ducks and the geese showed any ability to bind A/Dk in the intestines. The four land based poultry species showed no binding despite showing expression of α2,3SA. The duck, goose and pheasant intestines also showed minor α2,6 SA expression. However, only A/Tk was able to bind and only in the intestines of the geese (Figure 5L). These results highlight the complexities associated with understanding the host range of influenza viruses. Although many studies, including ours, have looked at the expression of SA receptors in tissues of several animal species, these receptors are not necessarily capable of binding influenza viruses (at least not under the conditions tested in this report). More studies are needed to better ascertain to which extent different animal species are likely hosts of influenza viruses and which minimal changes in receptor binding are needed to establish productive infections in these hosts.