Influenza A viruses belong to the family Orthomyxoviridae and have been isolated from a variety of different species. Further subtyping of influenza A viruses is based on antigenic differences between the two surface glycoproteins haemagglutinin (H1-H16) and neuraminidase (N1-N9) of the influenza A viruses 1 2 . The HA mediates the attachment of the virus to sialic-acid-containing receptors on the host cell surface, as well as the fusion of the virus envelope with the cellular membrane 3 4 . The specificity of the HA towards these molecules differs. Avian and equine influenza viruses preferentially bind the sialic acid α-2,3-galactose (SAα2,3Gal) linkage, while human influenza viruses preferentially bind the SAα2,6Gal linkage 5 6 7 . The highly pathogenic avian influenza virus H5N1 (HPAIV- H5N1) represents an important poultry pathogen and a major havoc to the poultry industry. Furthermore, HPAIV H5N1 infections in poultry constitute a threat to mammals including humans. Apart from humans, the natural infections of H5N1 influenza A have been reported in several mammalian species including domestic cats 8 , tigers and leopards 9 , dogs 10 , pigs 11 and stone marten 12 . Experimentally, H5N1 has also been able to infect mice 13 , ferrets 14 , monkeys 15 and cattle 16 . Infections in either experimental or naturally infected hosts have been fatal except for pigs and cattle where mild to subclinical infections have been detected 11 16 . No naturally occurring cases of H5N1 HPAI have been reported in horses or other members of the Perissodactyla order nor have any experimental studies been published 17 . An equine influenza virus;H3N8 with avian gene pool has been isolated emphasizing that equines may be susceptible to avian influenza viruses of the H3N8 subtype 18 and possibly others. The family Equidae and in particular donkeys may be of great importance in certain endemic countries like Egypt where they are commonly housed together with poultry. Long term endemic influenza virus infections in poultry increase exposure risks to surrounding humans and other mammals and in turn, create opportunities for the emergence of human-adapted strains with pandemic potential 19 20 . Since 2006, H5N1 influenza A virus has been endemic in Egypt producing great economic losses and most importantly hitting humans hard with high case fatality rate; 34/109 (WHO; http://www.who.int/csr/disease/avian_influenza/country/cases_table_2010_04_09/en/index.html).

Here we report the isolation of HPAI H5N1 from donkeys living in contact with diseased birds and demonstrate the presence of H5 seropositive ones in the neighbouring areas.

In this study, we isolated H5N1 form donkeys clinically affected with moderate respiratory distress including cough, fever and serous nasal discharge. The course of the disease was short (72 h) and responsive well after two shots of streptomycin/penicillin antibiotic therapy and one shot of antipyretic (Diclofenac sodium) with no recorded mortalities. The inhibition induced by antibiotic to the possible secondary bacterial invasion, besides the moderate severity of the H5N1 in donkeys may be responsible for the recovery of the infected animals without further complications. The disease recoded on 24^th March 2009, 1 wk after an outbreak of H5N1 infection in poultry in the village, where many donkeys suffered from the same clinical manifestations in an epidemic manner. The virus was isolated from a pool of nasal discharge from three affected animals. It produced haemagglutination only after the 3^rd egg passage. RT-PCR was performed to the full length of both NA and HA genes, where they were sequenced directly after gel purification. Sequences were deposited in GenBank under accession numbers; GU371911 and GU371912 for HA and NA respectively. The HA and NA genes of the investigated equine isolates revealed that they belonged to (5J), (1J) lineages respectively (According to the Influenza A Virus Genotype Tool) 24 . Phylogenetic analyses revealed that the HA of the equine isolate related to sublineages, A (A1) (Fig. 1). The equine isolate showed a typical polybasic cleavage motif with the GERRRKKR*GLF consensus sequence found in clade 2.2 viruses. It also contains amino acid D403 characteristic to sub-clade 2.2.1 (Fig. 2). The haemagglutinin gene was found to be closely related to A/chicken/Egypt/0894-NLQP/2008, A/Egypt/N00605/2009 and A/chicken/Egypt/092-NLQP/2009 while the neuraminidase gene of the current strain is closely related to A/Egypt/N03450/2009 and A/Egypt/N05056/2009. However, none of these strains were isolated from localities near to Beni-Suef.

The equine isolate has amino acids Q226 and G228 (H3 influenza numbering) denoting the preferential binding of α-2,3 linkage, typical for the avian and equine viruses but not human ones 5 25 . This finding demonstrated that the isolates with avian specific receptor binding properties can replicate and cause infection in equines. Different amino-acids that are implicated in receptor specificity Y98, S136, W153, H183, E190, K193 L194 E216 P221 K222 G225, Q226, S227, G228 (H3 influenza numbering) were tested 26 . Interestingly 98 (Y to N), 193 (K to R), 216 (E to K) and 221(P to S) which were found in the examined equine isolate have also been in many other isolates from the middle east in the flu database which raises a question as to what the impact of such substitution on HA binding to human receptors is. Recently, A138V, N186K and S227N mutations were reported to confer α-2,6-linked sialic acid binding to H5N1 virus 27 28 29 30 . None of such substitutions were found in the equine isolate. The HA of the human or avian Egyptian isolates, contains a total of seven potential N-glycosylation signals whereas the equine isolate contains an additional N-glycosylation (Fig. 2).

In addition, the equine isolate lacks aa S145. This deletion is also present in all other viruses grouped into 2.2 sublineage A1, which also includes sequences from human H5N1 isolates (Fig. 2). The significance of this deletion is unknown, but it should be noted that this position is close to a domain modulating receptor interaction. Interestingly, strains with this deletion appear to evolve towards a receptor usage that is similar to that of the seasonal human H1N1 31 .

Analysis of the NA gene revealed the presence of the 20-amino acid deletion (Fig. 3) as well as the presence of amino-acid R at position 110 which is characteristic of clade 2.2 viruses 32 . Three potential N-glycosylation sites were predicted (Fig. 3). The 228 (N to S) substitution is also present and indicative to 2.2.1 virus (2009). Four NA mutations; E119G, H274Y, R292K, and N294S have been reported to confer resistance to NA inhibitors 33 but none were detected in the equine isolate.

Speculating that infected animals may mount an antibody response depending on the interval post infection, we screened H5-specific antibodies 4-6 months after the initial virus isolation. The H5 specific antibodies were detected in naturally affected animals. 27 out of 105 (25.71%) of the examined animals were H5 positive with the highest percentage found in the area where the virus was isolated (Table 1, Fig. 4).

The authors declare that they have no competing interests.

ASA and AEA designed, performed experiments, and analyzed data. ASA generated genetic constructs and drafts the manuscript. AEA and SASS performed the HI analyses and helped in RT-PCR. AEA reviewed the manuscript.