Wild Ducks as Long-Distance Vectors of Highly Pathogenic Avian Influenza Virus (H5N1)

Some duck species are potential long-distance vectors; others are more likely to function as sentinels.

T he currently ongoing outbreaks caused by highly pathogenic avian infl uenza virus (HPAIV) of the subtype H5N1 are of concern not only to the poultry industry but also to public health (1,2). This virus, which causes a high fatality rate among infected patients, may adapt to effi cient human-to-human transmission and thus initiate a new human infl uenza pandemic (3). Since 1996, when the ancestor virus was identifi ed in domestic geese from China (4), outbreaks have spread and now encompass countries in Asia, the Middle East, Europe, and Africa (5). This spread of HPAIV among poultry fl ocks is traditionally thought to occur by transport of infected poultry, contaminated equipment, and persons associated with the poultry industry (6). HPAIV has occasionally been detected in wild birds near affected poultry fl ocks, but these birds have had limited or no role in virus dissemination (7). In the current outbreaks, however, wild birds are suspected of playing a major role as long-distance virus vectors.
During the expansion of HPAI (H5N1) outbreaks from Asia to Europe, 2 events implicated wild birds, particularly waterbirds, as long-distance virus vectors (8). First, virus outbreaks in 2005 rapidly spread westward from Russia and Kazakhstan in July and August to Turkey, Romania, and Ukraine in October. Wild waterbirds were suggested as a vector because the virus spread through areas that had no record of any virus presence and coincided with the fall migration of wild waterbirds between these areas. Second, at the beginning of 2006, HPAIV (H5N1) was detected in many wild waterbirds in western Europe, often in areas where no outbreaks had been detected among intensively surveyed poultry (9)(10)(11)(12); this event overlapped with unusual waterbird movements associated with cold weather in the Black Sea area. Quantitative analysis of the global spread of HPAIV (H5N1) also supports the potential role of migratory wild birds in virus spread (13). In June and July of 2007, Germany, France, and the Czech Republic again reported HPAIV (H5N1) in wild waterbirds (14), which illustrates their ongoing involvement in the epidemiology of this viral infection.
The main argument against the view that wild waterbirds are long-distance vectors of HPAIV (H5N1) is that most wild waterbirds in which this virus was identifi ed were either sick or dead, which suggests that they were too severely affected to spread the virus over any substantial distance (15). This argument is supported by experimental evidence that over time HPAIV (H5N1) has become more pathogenic for domestic ducks. Although domestic ducks did not show clinical disease or death from HPAI V (H5N1) isolates from 2001 or before, experimental infection of domestic ducks with an HPAIV (H5N1) isolate from 2002 caused neurologic disease and death (16). This high pathogenicity has been shown to be associated with molecular changes in the polymerase genes PA and PB1 Wild Ducks as Long-Distance Vectors of Highly Pathogenic Avian Infl uenza Virus (H5N1) (17). However, little is known about the pathogenicity and excretion pattern of recent HPAIV (H5N1) isolates in wild waterbird species whose migration patterns correspond with the observed westward expansion of HPAI (H5N1) outbreaks.
To test the hypothesis that wild waterbirds can excrete HPAIV (H5N1) in the absence of debilitating disease and so potentially act as long-distance virus vectors, we experimentally infected 6 species of wild ducks with an avian isolate of HPAIV (H5N1) from Europe, obtained in 2005 (A/ turkey/Turkey/1/2005). We chose ducks because they are an important group in the epidemiology of avian infl uenza in the wild, although other waterbird species, such as geese, swans, and gulls, also play a role (15). We chose these particular duck species because of their abundance, preference for freshwater habitats, and migratory pattern spanning Asia, Europe, and Africa (Table; online Appendix Figure  1, available from www.cdc.gov/EID/content/14/4/600-appG1.htm). All 6 species are listed by the European Union as carrying a higher risk for avian infl uenza (18).

Virus Preparation
A virus stock of infl uenza virus A/turkey/Turkey/1/2005 (H5N1) was prepared by 2 passages in 10-day-old embryonated chicken eggs. The harvested allantoic fl uid had a titer (19)  The Anas species represent 3 clades: the mallard represents the mallard clade, the common teal represents the greenwinged teal clade, and the Eurasian wigeon and gadwall represent the wigeon clade, previously belonging to the genus Strepera (20). For each species, males and females were equally represented. All ducks used for the infection experiments were captive-bred (Dierenhandel Hoogendoorn, Stolwijk, the Netherlands) and housed indoors since hatching to minimize the risk for inadvertent avian infl uenza virus infection. Birds were 8-11 months of age at time of inoculation. Serum samples, cloacal swabs, and pharyngeal swabs were collected from all ducks 1 week before inoculation. Serum was analyzed by using a commercially available infl uenza A virus antibody ELISA kit for the detection of antibodies against nucleoprotein (European Veterinary Laboratory, Woerden, the Netherlands) according to the manufacturer's instructions. Swabs were tested by reverse transcription-PCR (RT-PCR). No duck had antinucleoprotein antibody, except 1 pochard. Its serologic status did not protect it from HPAIV (H5N1) infection; it had the most severe clinical signs of all inoculated pochards and died at 4 days postinoculation (dpi). For 1 teal and 2 pochards, titers were suspected positive. No duck used for the infection experiments was positive for avian infl uenza virus by RT-PCR. We used 8 specifi c-pathogenfree White Leghorn chickens, 4-6 weeks old, as controls for the pathogenicity of the virus stock.

Experimental Design
For each species, 8 birds were housed together in negatively pressurized isolator units. Each bird was inoculated with 1 × 10 4 TCID 50 HPAIV (H5N1), 1.5 mL intratracheally and 1.5 mL intraesophageally. We used this low dose to increase the chance of inducing a subclinical infection and to simulate fi eld circumstances. In addition, 4 birds per duck species, which served as negative controls, were sham inoculated in the same manner with PBS-diluted sterile allantoic fl uid. Each day, a qualifi ed veterinarian scored clinical signs of disease in all birds according to a standardized list. Cloacal and pharyngeal swabs were collected in 1 mL transport medium (21) daily for the fi rst 14 days and every 2 days thereafter. We randomly divided each group of 8 birds into 2 groups of 4. One group was euthanized by exsanguination under isofl urane anesthesia for pathologic examination at 4 dpi; the other group was monitored for virus excretion until 18-21 dpi. Two ducks were removed after inoculation: 1 tufted duck because of concurrent aspergillosis and 1 pochard because of concurrent staphylococcosis. Also, 1 pochard and 3 tufted ducks were were chosen for pathologic examination at 4 dpi because they were dead or moribund. Although this was not random sampling, it does refl ect the fi eld situation because dead ducks no longer actively excrete virus. As expected, by 2 dpi 100% of the positive-control chickens were sick or dead, whereas the negative-control ducks showed no clinical signs and were euthanized at 4 dpi. Animal studies were approved by an independent animal ethics committee and performed under Biosafety Level 3+ conditions.

Pathologic Examination and Immunohistochemical Testing
Necropsies and tissue sampling were performed according to a standard protocol. After fi xation in 10% neutral-buffered formalin and embedding in paraffi n, tissue sections were examined by 1 of 2 methods: hematoxylin and eosin staining for histologic evaluation or an immunohistologic method that used a monoclonal antibody against nucleoprotein of infl uenza A virus as a primary antibody for detection of infl uenza viral antigen (22). The positive control was lung tissue of an HPAIV (H5N1)-infected domestic cat; negative controls were omission of primary antibody, substitution of primary antibody by an irrelevant monoclonal antibody of the same isotype, and testing of tissues from sham-inoculated ducks The following tissues were examined: brain (cerebrum, cerebellum, brainstem), trachea, bronchus, lung, caudothoracic or abdominal air sac, esophagus, proventriculus, duodenum, pancreas, liver, jejunum, ileum, cecum, colon, bursa of Fabricius, spleen, kidney, gonad (testis or ovary), heart, pectoral muscle, and adrenal gland.

RT-PCR and Virus Titration
Tissue samples were weighed and homogenized in 3 mL of transport medium by use of a homogenizer (Kinematica Polytron, Lucerne, Switzerland). RNA isolation and RT-PCR were performed as described (23). Briefl y, RNA from swabs and tissue suspensions was isolated by using a MagNaPure LC system with the MagNaPure LC Total Nucleic Acid Isolation Kit (Roche Diagnostics, Almere, the Netherlands). Real-time RT-PCR assays were performed on an ABI Prism 7000 Sequence Detection System machine (Applied Biosystems, Foster City, CA, USA) by using the TaqMan EZ RT-PCR Core Reagents Kit (Applied Biosys-tems, Nieuwerkerk a/d IJssel, the Netherlands) according to the manufacturer's instructions. The test used a hybridization probe (5′-6-FAM-TTT-ATT-CAA-CAG-TGG-CGA-GTT-CCC-TAG-CAC-T-TAMRA-3′) and specifi ed primers (forward: 5′-GAG-AGG-AAA-TAA-GTG-GAG-TAA-AAT-TGG-A-3′ and reverse: 5′-AAG-ATA-GAC-CAG-CTA-CCA-TGA-TTG-C-3′) to detect the hemagglutinin gene of HPAIV (H5N1). For each run the samples were prepared and processed in parallel with several negative and positive control samples. Virus titers were determined by serial 10-fold dilution of the homogenized tissue samples and swabs on MDCK cells, as described (19). Virus titrations were performed in quadruplicate.

Results
Despite the low doses of virus used to inoculate the ducks, rates of productive infection in the 6 species were high: 76% according to virus isolation and 93% according to RT-PCR (Table). HPAIV (H5N1) infection caused clinical signs of disease in only tufted ducks and pochards, both of which are diving ducks in the genus Aythya (Table). In contrast, the remaining 4 species-all dabbling ducks belonging to the genus Anas-were clinically unaffected. Clinical signs, which were more severe in tufted ducks than in pochards, developed at 3 to 4 dpi and consisted of labored breathing, increased recumbency, and neurologic signs (torticollis [ Figure 1, panel A], circling, loss of balance, and head tremors). Severely affected birds died or were euthanized in a moribund state at 4 dpi. Mildly affected birds recovered by 7 or 8 dpi.
Severe neurologic signs in tufted ducks and pochards were associated with multifocal viral encephalitis. Although no gross brain lesions were noted (online Appendix Table 1, available from www.cdc.gov/EID/content/14/4/ 600-appT1.htm), according to laboratory analysis these birds had multiple foci of infl uenza virus antigen expression ( Figure 1, panel B; online Appendix Table 2, available from www.cdc.gov/EID/content/14/4/600-appT2.htm) associated with severe necrosis and infl ammation (Figure 1, panel C) and high virus titers (10 3.5 to 10 5.2 TCID 50 per g tissue) in the brain (online Appendix Table 3, available from www.cdc.gov/EID/content/14/4/600-appT3.htm). The only other ducks with evidence of HPAIV (H5N1) infection of the brain were gadwalls, none of which showed neurologic signs. Gadwalls had only focal infl uenza virus antigen expression (Figure 1, panel D; online Appendix Table 2), mild encephalitis (Figure 1, panel E), and low virus titers (10 1.5 TCID 50 per g tissue) in the brain (online Appendix Table 3). No other species had evidence of HPAIV (H5N1) infection in the brain according to immunohistochemical testing, histologic examination, or virus isolation (Figure 1, panels F and G; online Appendix Tables 2 and 3), although  individual animals did have positive RT-PCR results for  the brain (online Appendix Table 4, available from www. cdc.gov/EID/content/14/4/600-appT4.htm).
Pharyngeal excretion of HPAIV (H5N1) varied signifi cantly among the 6 duck species (1-way analysis of variance of area under pharyngeal excretion curve up to 4 dpi, p<0.001), by virus isolation (Figure 2, panel A) and by RT-PCR (Figure 2, panel C). The ducks could be divided into a high-excretion group consisting of tufted ducks, pochards, and mallards, and a low-excretion group consisting of teals, wigeons, and gadwalls (Figure 2, panels B and D). Pharyngeal excretion also varied substantially among individuals within species (online Appendix Figures 2 and  3, available from www.cdc.gov/EID/content/14/4/600-app G2.htm and www.cdc.gov/EID/content/14/4/600-appG3. htm, respectively.). This fi nding was most extreme in tufted ducks and pochards, the species in which the individuals with the highest excretion level were also those showing severe clinical signs (Figure 2, panels B and D).
Pharyngeally excreted HPAIV (H5N1) likely originated from lung, air sac, or both, because these were the only tissues in the respiratory tract that had immunohistochemical evidence of virus replication (Figure 2, panels E and F; online Appendix Table 2) and because virus was frequently detected in these tissues by virus isolation (online Appendix Table 3) and RT-PCR (online Appendix Table 4). The histologic lesions corresponding to infl uenza virus antigen expression in these tissues were bronchointerstitial pneumonia ( Figure 2, panel F) and lymphocytic airsacculitis (Figure 2, panel H). Despite frequent isolation of HPAIV (H5N1) from trachea and extrapulmonary bronchus (online Appendix Table 3), these tissues had neither histopathologic nor immunohistochemical evidence of HPAIV (H5N1) replication (online Appendix Table 2), which suggests that virus isolated from these sites at 4 dpi originated from elsewhere in the respiratory tract.
Cloacal excretion of HPAIV (H5N1) was uncommon; virus was detected in cloacal swabs of only 7% of ducks by virus isolation and 59% by RT-PCR (Table). Cloacal excretion was exceeded by pharyngeal excretion in all 6 duck species, according to virus isolation (Figure 3, panels A and B; online Appendix Figure 2) and RT-PCR ( Figure  3, panels C and D; online Appendix Figure 3).
Cloacally excreted virus likely originated from pancreas, liver, or both, on the basis of the signifi cant association between virus antigen expression in these tissues (online Appendix Table 2) and virus detection in cloacal swabs by RT-PCR (online Appendix Table 4); 7 of 8 birds with virus antigen expression in liver, pancreas, or both had PCR-positive cloacal swabs between 1 and 4 dpi, in contrast to 0 of 15 birds without virus antigen expression in these tissues (Fisher exact test, p<0.00001). In the pancreas, tufted ducks and pochards had multifocal necrosis (Figure 3, panel E), which was the most prominent gross lesion associated with HPAIV (H5N1) infection in this study (online Appendix Table 1). Virus antigen expression (Figure 3, panel F) occurred at the transition between normal parenchyma and these necrotic foci (Figure 3

, panel G) and
Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 14, No. 4, April 2008 603 corresponded with high virus titers (10 1.5 to 10 6.2 TCID 50 per g tissue) in the pancreas (online Appendix Table 3). In the liver, widespread virus antigen expression (Figure 3, panel H) was associated with necrotizing hepatitis ( Figure  3, panel I) and variable virus titers (no virus isolated to 10 6.2 TCID 50 per g tissue) in the liver (online Appendix Table  3). Virus produced in pancreas and liver could potentially have reached the intestinal lumen through pancreatic and bile ducts, respectively. Although virus antigen expression was detected in several other tissues, virus originating from these sites likely did not contribute to virus excretion. It is unlikely that cloacally excreted virus originated from the intestinal, urinary, or genital tracts, although all 3 tracts empty into the cloaca. In the intestine, no virus antigen expression was found in the intestinal epithelium of any of the 23 ducks examined. Virus antigen expression was found in neurons and satellite cells in the peripheral nervous system (submucosal and myenteric plexi, mesenteric ganglia) of the small intestine (online Appendix Figure  4, panel A, available from www.cdc.gov/EID/content/14/4/ 600-appG4.htm), in association with necrotizing ganglioneuritis (online Appendix Figure 4, panel B), and in myocytes in the lamina muscularis mucosae of the colon, without associated histologic lesions. However, these tissues do not empty into the intestinal lumen. No virus antigen expression was found in tissues of urinary tract (kidney) or genital tract (testis or ovary) (online Appendix Table 2). The occasional isolation of HPAIV (H5N1) from kidney samples (online Appendix Table 3) may be explained by inadvertent sampling of overlying air sac wall, which did express infl uenza virus antigen.
Evidence of HPAIV (H5N1) replication was found sporadically in tissues other than those of the respiratory, digestive, and nervous systems (online Appendix Table 2). Virus antigen expression was detected in multiple foci of medullary and cortical cells of the adrenal gland (online Appendix Figure 4, panel C) and was associated with necrotizing adrenalitis (online Appendix Figure 4, panel D). Virus antigen expression was also detected in multiple foci of myocytes of the heart and was associated with myocardial necrosis.

Discussion
Our study shows that of the 6 wild duck species studied, the mallard is the prime candidate for being a longdistance vector of HPAIV (H5N1) because it was the only species to show abundant virus excretion without clinical or pathologic evidence of debilitating disease (Table; Fig ing areas; and it is found on nearly every type of wetland and is very tolerant of human presence, thus forming a potential link between wild waterfowl, domestic animals, and humans. Pochards and tufted ducks are less likely candidates as long-distance virus vectors because those individuals that excreted the most virus also developed severe neurologic disease (Figure 2, panels B and D) and therefore would not have been able to fl y far before succumbing. Instead, they are more likely to act as sentinels for HPAIV (H5N1) in wild bird populations, as do mute swans (Cygnus olor) (9). However, pochards cannot be ruled out as potential vectors because some birds excreted abundant virus in absence of severe clinical signs (Figure 2, panels B and D). Our results correspond with fi eld observations of pochards and tufted ducks involved in wild bird die-offs from HPAIV (H5N1) infection in France, Germany, and Sweden early in 2006 (14). Some of these birds showed clinical signs of neurologic disease, e.g., compulsively swimming around in circles (28). Therefore, close surveillance of these 2 Aythya species for unusual illness, particularly neurologic disease, or death should provide early warning for HPAIV (H5N1) infection in an area. Redheads (Aythya americana), which are diving ducks indigenous to North America, experimentally infected with a 2005 isolate of HPAIV (H5N1) neither showed clinical signs nor died (29). Of the 6 species tested, the 3 remaining Anas species-gadwall, teal, and wigeonare the least likely candidates as long-distance virus vectors because they had limited virus excretion (Figure 2, panels B and D).
HPAIV (H5N1) infection in these wild ducks contrasts in pattern of excretion with that of low-pathogenicity avian infl uenza virus infection in wild ducks and contrasts in pattern of disease with that of HPAIV infection in chickens. Both contrasts can be explained by the specifi c tissue tropism of HPAIV (H5N1) in wild ducks. With regard to pattern of excretion, low cloacal excretion was associated with lack of evidence for HPAIV (H5N1) replication in intestinal epithelium of any of the 23 ducks examined (online Appendix Table 2), in contrast to most low-pathogenicity avian infl uenza viruses for which intestine is the main replication site (30). Instead, HPAIV (H5N1) replicated preferentially in the respiratory tract (online Appendix Tables 2 and 4), which corresponds with high pharyngeal excretion. How this preferential pharyngeal excretion of HPAIV (H5N1) affects its spread and persistence in a wild duck population remains to be determined.
Severe clinical disease in the HPAIV (H5N1)-infected tufted ducks and pochards manifested itself mainly as neurologic signs at about 4 dpi, although pathologic examination also showed virus-induced lesions in organs other than the brain. These fi ndings differ substantially from those of HPAIV (H5N1)-infected chickens, which are characterized mainly by widespread hemorrhage and edema and death by about 2 dpi (31). Again, this contrast can be explained by differences in tissue tropism. Whereas the cardiovascular lesions in poultry are associated with widespread replication of HPAIV (H5N1) in endothelium lining the blood vessels (31), no such endotheliotropism was detected in any of 23 ducks examined.
The knowledge gained from this study has several implications for surveillance in wild ducks. Active surveillance (sampling of apparently healthy wild birds) should give priority to mallards and, to a lesser degree, pochards. Sampling should not be limited to cloacal swabs, as is the custom in surveillance for low-pathogenicity avian infl uenza virus, but should include pharyngeal swabs. Passive surveillance (sampling of diseased or dead birds), should pay extra attention to tufted ducks and pochards, particu-    Organ  1  2  3  1  2  3  4  1  2  3  4  1  2  3  4  1  2  3  4  1  2  3  4  Nervous  Brain  14 15 14  36  15 24  34  37  34  35 20 34  Respiratory  Trachea  16 24