Transmission of Equine Influenza Virus to English Foxhounds

We retrospectively demonstrated that an outbreak of severe respiratory disease in a pack of English foxhounds in the United Kingdom in September 2002 was caused by an equine influenza A virus (H3N8). We also demonstrated that canine respiratory tissue possesses the relevant receptors for infection with equine influenza virus.

I nfl uenza A viruses are divided into subtypes according to the serologic reactivity of the surface glycoproteins hemagglutinin (H1-H16) and neuraminidase (N1-N9). Aquatic birds are regarded as the natural reservoir for infl uenza A viruses; a few mammalian hosts are infected by a limited number of virus subtypes. The fi rst evidence of the H3N8 subtype, which currently circulates in horses, crossing species barriers was reported after an outbreak of respiratory disease among racing greyhounds in Florida in 2004. Isolation of virus from 1 case and detection of specifi c antibodies in other cases identifi ed equine infl uenza virus as the cause of the outbreak (1). This information led us to reexamine an outbreak of severe respiratory disease that occurred in a pack of 92 English foxhounds in the United Kingdom in September 2002.

The Study
The outbreak was signaled by a sudden onset of coughing. Some hounds became lethargic and weak; in some, these signs progressed to loss of consciousness. One hound died and 6 were euthanized. Postmortem examination of the hound that died (case 1) and 1 that was euthanized (case 2) showed subacute broncho-interstitial pneumonia; virus was suspected as the cause. When they were puppies (≈8 weeks of age), the hounds had been inoculated with commercially available vaccines against the major canine respiratory and enteric viruses. Postmortem tissue samples submitted to a canine infectious diseases laboratory were negative for known canine viral pathogens (e.g., canine herpesvirus, adenovirus, parainfl uenza virus). The diagno-sis as to the cause of the pneumonia, returned in 2002, was "unknown, suspected viral etiology." In January and March 2005, serum samples were obtained from the hounds affected by the respiratory disease outbreak in 2002 (pack 1). Serum samples were obtained from another 3 packs of foxhounds in the same region of the United Kingdom during December 2004 through February 2005. Samples were collected from 31-33 hounds (equivalent numbers of males and females) in each pack, ranging in age from 9 months to 9 years. The serum was screened for antibodies by using the single radial hemolysis assay (2). None of the samples contained antibodies to the strains that were included in the assay to control for nonspecifi c reactivity: equine H7N7 subtype strain A/equine/ Prague/56 and the human infl uenza virus strain A/Puerto Rico/8/34 (H1N1). Antibodies to the H3N8 subtype strains A/equine/Newmarket/1/93 and A/equine/Newmarket/2/93 were, however, detected in 9 of the samples obtained during the fi rst visit to pack 1 (Table). Of these, 8 were from hounds that had survived the outbreak in 2002; however, 1 was from a hound (no. 22) born after the outbreak in another part of the United Kingdom, which suggests that the 2002 outbreak might not have been the only incident of equine infl uenza to have infected hounds in the United Kingdom. Another 3 positive serum samples were obtained during a second visit to pack 1, and a repeat sample from hound no. 22 again had positive results. The specifi city of the antibodies for equine infl uenza A (H3N8) strains was confi rmed by hemagglutination inhibition assays that included human infl uenza (H3N2) strain A/Scotland/74 (data not shown).
An immunohistochemical test to detect infl uenza A virus that used equine infl uenza-specifi c rabbit polyclonal antiserum was applied to formalin-fi xed paraffi n-embedded (FFPE) tissues from the 2 hounds that were examined postmortem in 2002 (3). Immunostaining of lung tissue showed positive staining in areas of pneumonic change; infected cells had the morphology of epithelial cells and macrophages ( Figure 1). Immunostaining of visceral tissues (lung, liver, spleen, myocardium, intestine, pancreas, and oropharynx) was negative.
Deparaffi nization of the FFPE lung tissue from the 2 hounds was performed as described previously (4) with a few modifi cations. RNA was extracted from the sample pellets obtained using the QIAamp viral RNA Mini Kit (QIAGEN, Hilden, Germany) according to the manufacturer's instructions. Ten different primer pairs designed to amplify short (<250-bp) products from the matrix hemagluttinin and neuraminidase genes were used (details available from the authors on request).  European lineage strains isolated since 1998 have valine at 242.

Conclusions
An important factor in interspecies transmission is the ability of the hemagglutinin protein of the virus to bind to certain receptors on the host cells before the virus is internalized. Although all infl uenza A viruses recognize cell surface oligosaccharides with a terminal sialic acid, their receptor specifi city varies; it is thought that species-specifi c differences in the distribution of linkages on respiratory epithelial cells infl uences the ability of infl uenza A viruses to transmit between species. Respiratory tract tissue samples were obtained within 2-4 hours of death from a horse and a greyhound, each euthanized for reasons other than this study, and rinsed extensively to remove surface mucous. The tissues were stained by immunofl uorescence by using the lectins Sambucus nigra (SNA, specifi c for SAα2,6 galactose(Gal)/N-acetylgalactosaminide) and Maackia amurensis (MAA, specifi c for SAα2,3) as previously described (7). The MAA lectin bound strongly to the equine tracheal epithelium (Figure 2, panel A), which confi rms the fi nding that the NeuAc2,3Gal linkage preferentially bound by equine infl uenza viruses is found on sialyloligosaccharides in the equine trachea (7). The MAA lectin also bound strongly to the canine respiratory epithelium (Figure 2, panel B) at all levels of the respiratory tract examined (distal, medial and proximal trachea; primary and secondary bronchi), which suggests that receptors with the required linkage for recognition by equine infl uenza virus are available on canine respiratory epithelial cells, although further subtle differences in receptor specifi city may exist. The SNA lectin, specifi c for SAα2,6Gal, which did not bind to the equine tracheal epithelium, showed some binding to the canine epithelium (data not shown).
Because the hounds infected in 2002 were housed near horses, it is possible that the virus was transmitted from infected horses by the usual (aerosol) route. However, during the week before onset of clinical signs, the hounds had been fed the meat of 2 recently euthanized horses from independent sources. That viral antigen expression was confi ned to the lungs indicates a respiratory rather than oral route of infection. It is possible that eating respiratory tissue from an infected horse led to inhalation of suffi cient virus particles to initiate a respiratory infection. Consumption of infected bird carcasses has been implicated in the transmission of highly pathogenic avian infl uenza virus of the H5N1 subtype to tigers and leopards (8) and a dog (9) and was demonstrated experimentally by feeding virus-infected chicks to domestic cats (10).
Although the mechanism remains unclear, we have demonstrated transmission of equine infl uenza virus to dogs in the United Kingdom, independent of that in the United States. We have also shown that canine respiratory tissue displays the relevant receptors for infection with equine infl uenza virus.