Influenza A Virus (H3N8) in Dogs with Respiratory Disease, Florida

Genetic and antigenic characterization established the uniqueness of this virus circulating in dogs.

I nfl uenza A viruses (family Orthomyxoviridae) are known to cause acute respiratory disease in humans, horses, pigs, and domestic poultry (1,2). Infl uenza A virus subtype H3N8 has recently emerged as a respiratory pathogen in dogs, associated with outbreaks of acute respiratory disease in racing greyhounds (3). The disease is caused by a novel virus closely related to contemporary equine infl uenza A virus subtype H3N8. These viruses share >96% nucleotide sequence identity, which suggests direct transmission of the entire virus from horses to dogs without reassortment with other strains (3).
Canine infl uenza virus (CIV) was fi rst identifi ed in racing greyhounds in Florida in January 2004 and was later associated with respiratory disease outbreaks in racing greyhounds in 9 states from 2004 through 2006 (3,4). Most affected greyhounds had clinical signs associated with virus infection of the upper respiratory tract-cough for 10-30 days, nasal discharge, low-grade fever-followed by recovery. However, some dogs died peracutely with extensive hemorrhage in the lungs, mediastinum, and pleural cavity. Histologic examination showed tracheitis, bronchitis, bronchiolitis, and suppurative bronchopneumonia associated with extensive erosion of epithelial cells and infi ltration with neutrophils. The isolation of 4 closely related infl uenza A subtype H3N8 viruses from dogs that died in different geographic locations over a 25-month period, together with substantial serologic evidence of widespread infection among racing greyhounds in 9 states, suggested sustained CIV circulation in this population by dog-to-dog transmission (3,4).
The fi rst evidence of CIV infection in dogs other than greyhounds came from serologic testing of dogs with acute respiratory disease in shelters, boarding kennels, and veterinary clinics in Florida and New York in 2004 and 2005 (3). Since August 2005, a national syndromic serosurvey for canine infl uenza has been conducted on >5,000 samples collected from nongreyhound dogs with compatible clinical signs (Cornell University College of Veterinary Medicine, http://diaglab.vet.cornell.edu/issues/civ-stat.asp). As of April 2008, seropositive dogs have been identifi ed in 25 states and the District of Columbia.
In April and May 2005, an outbreak of respiratory disease occurred in dogs housed in a shelter facility in northeastern Florida (3). The outbreak involved at least 58 dogs, ranging in age from 3 months to 9 years, and included purebred dogs as well as mixed breeds; 6 were euthanized. In May 2005, a respiratory disease outbreak occurred among ≈40 pet dogs at a veterinary clinic in southeastern Florida; 1 died. We performed molecular analyses on 2 infl uenza

Infl uenza A Virus (H3N8) in Dogs with Respiratory Disease, Florida
A subtype H3N8 viruses isolated from these 7 nongreyhound dogs that died and genetically and antigenically compared them with infl uenza (H3N8) viruses from racing greyhounds.

Specimen Collection
Postmortem examinations were performed on the 6 mixed-breed shelter dogs that died in April and May 2005 and on the 1 pet Yorkshire terrier that died in the veterinary clinic in May 2005. Tissues were fi xed in 10% neutral buffered formalin and embedded in paraffi n; 5-μm sections were stained with hematoxylin and eosin for histopathologic diagnosis. Unfi xed tissues for virologic and molecular analyses were stored at -80 o C.

RNA Extraction
Frozen lung tissues from each dog were thawed and homogenized in lysis buffer containing β-mercaptoethanol by using a disposable tissue grinder (Kendall, Lifeline Medical Inc., Danbury, CT, USA). Total RNA was extracted by using a commercial kit (RNeasy Mini Kit, QIAGEN Inc., Valencia, CA, USA) according to manufacturer's instructions and eluted in a fi nal volume of 60 μL of buffer. Total RNA was also extracted from lung tissue collected from specifi c-pathogen-free dogs without respiratory disease.

Real-Time Reverse Transcription-PCR
A single-step quantitative real-time reverse transcription-PCR (RT-PCR) was performed on total RNA extracted from the canine tissue samples by using the QuantiTect Probe RT-PCR Kit containing ROX as a passive reference dye (QIAGEN). Briefl y, 2 primer-probe sets were used for detection of infl uenza A sequences in each sample (Table  1). One primer-probe set was selective for canine infl uenza subtype H3 gene sequences. The other primer-probe set targeted a highly conserved region of the matrix (M) gene of type A infl uenza virus. The sequence of the M probe contained 3 locked nucleic acids. For each real-time RT-PCR, 5 μL of total RNA was added to a reaction mixture containing 12.5 μL of 2× QuantiTect Probe RT-PCR Master Mix, 0.25 μL of QuantiTech RT Mix (both QIAGEN), forward and reverse primers (0.4 μmol/L fi nal concentration for each), probe (0.1 μmol/L fi nal concentration), and RNasefree water in a fi nal volume of 25 μL. Real-time PCR for eukaryotic 18S rRNA was performed by using commercially available assay reagents (VIC/TAMRA; TaqMan, Applied Biosystems, Foster City, CA, USA), according to manufacturer's instructions for detection of endogenous 18S rRNA, as an internal control for RNA extraction from the tissues.
Quantitative one-step real-time RT-PCR was performed on the reaction mixtures in an Mx3000P QPCR System (Stratagene, La Jolla, CA, USA). Cycling conditions were a reverse transcription step at 50°C for 30 min, an initial denaturation step at 95°C for 15 min to activate the HotStarTaq DNA Polymerase (QIAGEN), and amplifi cation for 40 cycles. Each amplifi cation cycle included denaturation at 94°C for 15 s, followed by annealing/extension at 60°C for 1 min. The FAM (emission wavelength 516 nm) and VIC (emission wavelength 555 nm) fl uorescent signals were recorded at the end of each cycle. The threshold cycle (Ct) was determined by setting the threshold fl uorescence at 1,000 for each individual experiment. The Mx3000P version 2.0 software program (Stratagene) was used for data acquisition and analysis. The results were normalized by calculating H3 Ct or M Ct to 18S rRNA Ct ratios for each sample. Samples were considered positive for infl uenza A virus when the normalized H3 or M Ct ratio was 3 U smaller than the normalized Ct ratio for lung tissues from dogs without respiratory disease. The positive control was amplifi ed RNA that had been extracted from A/canine/Florida/242/2003 (H3N8) virus grown in MDCK cells (3).

Inoculation of Cell Culture
Frozen lung tissues from each of the 7 dogs were thawed and homogenized in 10 volumes of Dulbecco modifi ed Eagle medium (DMEM) supplemented with 0.5% bovine serum albumin and gentamicin and ciprofl oxacin. Solid debris was removed by centrifugation, and supernatants  were inoculated onto MDCK cells cultured in DMEM supplemented with 1 μg/mL TPCK (L-1-tosylamido-2-phenylethyl chloromethyl ketone)-treated trypsin (Sigma-Aldrich Corp., St. Louis, MO, USA), 0.35% bovine serum albumin, and antimicrobial drugs. Cells were grown in 25-cm 2 fl asks at 37°C in a humidifi ed atmosphere containing 5% CO 2 . The cultures were observed daily for morphologic changes and harvested 3 days after inoculation. The harvested cultures were clarifi ed by centrifugation, and the supernatants were inoculated onto fresh MDCK cells as described for the initial inoculation; up to 3 additional passages were performed for samples that did not show evidence of infl uenza virus by hemagglutination or RT-PCR. Hemagglutination activity in the clarifi ed supernatants was determined by using 0.5% turkey red blood cells as described (5,6). RT-PCR was performed as described below.

Inoculation of Embryonated Chicken Eggs
Frozen lung tissues were homogenized as described above for inoculation of MDCK cells by using phosphatebuffered saline instead of DMEM, and a volume of 0.2 mL was inoculated into the allantoic sac of 10-day old embryonated chicken eggs. After 48 h of incubation at 35°C, the eggs were chilled at 4°C overnight before the allantoic fl uid was harvested. Hemagglutination activity in the clarifi ed allantoic fl uid supernatants was determined by using 0.5% turkey red blood cells as described (5,6). RT-PCR was performed as described below. Harvested allantoic fl uid samples lacking evidence of infl uenza virus were reinoculated up to 2 more times in embryonated eggs and evaluated as described.

RT-PCR, Nucleotide Sequencing, and Phylogenetic Analyses
Virus RNA was extracted from MDCK culture supernatant or allantoic fl uid by using the QIAamp Viral RNA Mini Kit (QIAGEN) according to manufacturer's instructions. The virus RNA was reverse transcribed to cDNA by using the QIAGEN OneStep RT-PCR Kit according to manufacturer's instructions. PCR amplifi cation of the coding region of the 8 infl uenza virus genes in the cDNA was performed as described (7) by using universal gene-specifi c primer sets (3). The resulting DNA amplicons were used as templates for automated sequencing in the ABI PRISM 3100 automated DNA sequencer by using cycle sequencing dye terminator chemistry (Applied Biosystems). Nucleotide sequences were analyzed by using the Lasergene 6 Package (DNASTAR, Inc., Madison, WI, USA). The PHYLIP Version 3.5 software program was used to estimate phylogenies and calculate bootstrap values from the nucleotide sequences (8). Phylogenetic trees were compared with those generated by neighbor-joining analysis with the Tamura-Nei model implemented in the MEGA3 program (9) and confi rmed by the PAUP 4.0 Beta program (Sinauer Associates, Inc., Sunderland, MA, USA). The complete genome sequences from the 2 new reported CIV isolates (A/ canine/Miami/2005 and A/canine/Jacksonville/2005) were deposited in GenBank under accession nos. EU402407-EU402408 and EU534193-EU534204.

Hemagglutination Inhibition Assay
Convalescent-phase immune serum samples were obtained from horses and dogs naturally infected with infl uenza A virus (H3N8) in 2005. Antiserum from ferrets infected with A/canine/Florida/43/2004 (H3N8) was prepared as described (3). All serum samples were incubated with receptor-destroying enzyme (DENKA SEIKEN Co., Ltd., Tokyo, Japan) (1 part serum: 3 parts receptor-destroying enzyme) for 16 h at 37°C before heat inactivation for 30 min at 56°C. Infl uenza A/canine/Jacksonville/2005 virus (H3N8) was grown in MDCK cells for 72 h at 37°C in 5% CO 2 . Virus culture supernatants were harvested, clarifi ed by centrifugation, and stored at -80°C. All other canine and equine viruses used in the hemagglutination inhibition (HI) assay were grown in 10-day old embryonated chicken eggs from which allantoic fl uid was collected after 72 h and stored at -80°C. The HI assay was performed as described (6). On the basis of assay results from serum of uninfected specifi c-pathogen-free dogs with HI titers <4, HI titers >32 were considered as evidence of previous exposure to CIV.

Clinical Findings
Among the 58 affected shelter dogs, the most common clinical signs were low-grade fever, purulent nasal discharge, and cough for 10-21 days. Paired acute-and convalescent-phase serum samples were collected from 5 dogs and tested for CIV-specifi c antibodies by using the HI assay. All the dogs seroconverted to A/canine/Florida/43/2004 (H3N8) and had an increase in the geometric mean antibody titer (GMT) from 37 in the acute phase to 626 in the convalescent phase. Single serum samples were collected from another 18 dogs that had had clinical disease for at least 7 days, and 17 (94%) were seropositive for A/canine/ Florida/43/2004 (H3N8); HI antibody titers ranged from 32 to 2,048 and GMT was 533. Pneumonia developed in at least 10 dogs, of which 6 were euthanized and submitted for postmortem examination. These 6 mixed-breed dogs were 3 males and 3 females ranging in age from 4 months to 3 years. The duration of clinical signs at the time of euthanasia was 2-10 days.
Among the ≈40 affected pet dogs at the veterinary clinic, the most common clinical signs were low-grade fever, purulent nasal discharge, and cough for 10-30 days. Paired acute-and convalescent-phase serum samples were collected from 19 dogs and tested for CIV-specifi c antibodies by using the HI assay. Of these, 11 (58%) dogs seroconverted to A/canine/Florida/43/2004 (H3N8), and the GMT increased from 9 in the acute phase to 329 in the convalescent phase. Single serum samples were collected from another 9 dogs that had had clinical disease for at least 7 days, and 6 (67%) were seropositive for A/canine/ Florida/43/2004 (H3N8); HI antibody titers ranged from 64 to 512 and GMT was 228. Pneumonia developed in 3 dogs; 1, a 9-year-old male Yorkshire terrier, died 3 days after onset of clinical signs and was submitted for postmortem examination.
Postmortem examinations showed that all 7 dogs had tracheitis and bronchitis with infl ammatory changes that involved submucosal glands. Tracheitis and bronchitis were characterized by surface and glandular epithelial necrosis and hyperplasia with infi ltration by lymphocytes, neutrophils, and macrophages. Suppurative bronchopneumonia was found in 2 of the shelter dogs and the pet dog; histologically identifi ed bacteria in the lesions indicated a bacterial contribution to the pneumonic lesions. None of the dogs had intrathoracic or pulmonary hemorrhage.

Laboratory Diagnosis
Lung tissues from the 7 dogs were analyzed by quantitative real-time RT-PCR assays that detect the M gene of infl uenza A and the H3 gene of CIV. The M and H3 genes were amplifi ed from the lungs of all 7 dogs, which confi rmed the presence of CIV (Table 2). Lung tissue from the specifi c-pathogen-free dogs did not show evidence of amplifi cation of infl uenza virus genes. MDCK cultures were inoculated with these lung homogenates to identify viable infl uenza virus. Pathologic cell damage was noted in the MDCK cell monolayer on the third passage of the lung homogenate from 1 of the shelter dogs that died after 3 days of pneumonia. Infl uenza A virus (later identifi ed as subtype H3N8) was recovered from the supernatant, and the isolate was named A/canine/Jacksonville/2005. After 2 passages in embryonated chicken eggs, infl uenza A virus (later identifi ed as subtype H3N8) was recovered from the lung homogenate of the pet dog that also died of pneumonia 3 days after onset. This virus was named A/canine/ Miami/2005. These isolates provided virologic evidence of CIV infection in nongreyhound dogs.

Amino Acid Sequence of the CIV H3 Hemagglutinin
To identify changes with possible functional significance with regard to antigenicity or receptor binding, we compared the deduced amino acid sequences of the H3 hemagglutinins from the 6 available CIV isolates. All 6

Antigenic Analyses of the Canine Infl uenza A (H3N8) Isolates
HI tests were performed by using an antigen panel of previously circulating and contemporary equine infl uenza viruses and the available CIVs as well as convalescentphase immune serum from horses and dogs naturally infected with infl uenza virus (H3N8) in 2005 (Table 4). An antiserum from ferrets infected with A/canine/Florida/43/2004 was also included in the analyses. HI antibody titers in equine serum were 2-fold to 16-fold higher with contemporary equine viruses (1999)(2000)(2001)(2002)(2003) than with older isolates . The heterologous titers of equine serum to canine viruses were generally similar to the homologous values for the contemporary equine viruses. The canine serum failed to substantially inhibit hemagglutination by the older equine infl uenza viruses , but the antibody titers to contemporary equine isolates (1999)(2000)(2001)(2002)(2003) and the canine isolates were comparable. Similar results were observed for serum from ferrets infected with CIV. These patterns of inhibition demonstrated the antigenic similarity between the CIVs and contemporary equine infl uenza viruses and were consistent with the phylogenetic analyses. The antibody titers to the A/canine/Miami/2005 isolate in equine, canine, and ferret serum were similar to those for the 2003 and 2004 canine isolates, which indicates that the amino acid substitutions in the isolates did not result in measurable antigenic drift. The antibody titers to the A/canine/Jacksonville/2005 isolate were, in general, 2to 4-fold lower than those to A/canine/Florida/43/2004 and other canine viruses, which suggests a potential antigenic difference.

Discussion
The genetic and phylogenetic analyses of infl uenza A subtype H3N8 viruses recovered from racing greyhounds affected by respiratory disease outbreaks and fatal pneumonia in 2003 and 2004 have been described (3). These greyhound infl uenza viruses were most homologous to contemporary equine infl uenza A subtype H3N8 viruses isolated from horses in 2002 and 2003. Although serologic evidence of infl uenza virus infection in nongreyhound dogs was reported (3), whether these infections were caused by the same virus that infected the greyhound dogs is not clear. The infl uenza subtype H3N8 viruses that we describe in this report came from nongreyhound dogs involved in fatal canine infl uenza outbreaks independent of any known outbreaks in greyhounds.
Because viral hemagglutinin is a critical determinant of the host specifi city of infl uenza virus (10), we compared the nucleic acid and deduced amino acid sequences for the canine and equine H3 to identify residues that may be associated with effi cient replication in different species or dog breeds. The 5 conserved amino acid substitutions in all 906 Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 14, No. 6, June 2008 the canine viruses differentiated them from contemporary equine viruses. The substitution of lysine for asparagine at position 54 maintains the positive charge of the residue and is of unknown functional signifi cance. Position 83 located within antigenic site E of human H3 has been implicated in antigenic drift (11). The substitution of serine for asparagine at position 83 in canine H3 maintains the polar nature of the residue, but the functional signifi cance in the evolution of canine infl uenza is not immediately apparent. The substitution of leucine for tryptophane at position 222 represents a nonconservative change adjacent to the sialic acid-binding pocket, which suggests a potential modulator function in adaptation of equine infl uenza virus to canine sialic acid receptors. This leucine substitution has also been reported in avian infl uenza subtype H4 infection of pigs (12) and subtype H9 infection of humans (13). The strictly conserved isoleucine at position 328 near the cleavage site of the equine H3 hemagglutinin has been replaced by threonine in all the canine isolates, which suggests the potential importance of threonine for recognition of the hemagglu-  .  from humoral immune responses (11,15,16). Although no evidence supports high variability at positions 118 and 261, neighboring positions 121 (antigenic site D) and 262 (antigenic site E) are sites of frequent positive selection in human H3 genes (11). The 2005 isolates from nongreyhound dogs differed from each other by substitution of proline for serine at position 107 in the HA1 subunit of A/canine/ Jacksonville/2005. Serine is conserved at position 107 in all other equine and canine isolates except for A/equine/ Jilin/1/89, which has a threonine substitution (17). A/canine/Jacksonville/2005 has potentially signifi cant antigenic variation from the other canine isolates, which may be partially related to the proline substitution at position 107. However, no serologic evidence indicates that residue 107 modulates the antigenicity of human, porcine, or equine H3 hemagglutinins.
The HI results showed antigenic similarity between the canine and contemporary equine infl uenza viruses and were in agreement with the phylogenetic clustering of canine hemagglutinin sequences near the equine counterparts. The amino acid and antigenic differences between canine and equine hemagglutinin illustrate the ongoing process of divergent evolution of the canine viruses from the ancestral equine viruses. These differences also support the hypothesis that CIV is a separate virus lineage, which is effi ciently maintained in the dog population by horizontal dog-to-dog transmission.
With the introduction and sustained transmission of infl uenza virus (H3N8) in dogs, the H3 hemagglutinin became the most broadly distributed subtype in mammalian species, including humans, swine, horses, and dogs. Continued virologic and serologic surveillance will be important for monitoring the evolution of CIV and its health effects in dogs as well as the possible transmission to other species, including humans.