Avian Influenza H6 Viruses Productively Infect and Cause Illness in Mice and Ferrets

Abstract

Influenza pandemic preparedness has focused on influenza virus H5 and H7 subtypes. However, it is not possible to predict with certainty which subtype of avian influenza virus will cause the next pandemic, and it is prudent to include other avian influenza virus subtypes in pandemic preparedness efforts. An H6 influenza virus was identified as a potential progenitor of the H5N1 viruses that emerged in Hong Kong in 1997. This virus continues to circulate in the bird population in Asia, and other H6 viruses are prevalent in birds in North America and Asia. The high rate of reassortment observed in influenza viruses and the prevalence of H6 viruses in birds suggest that this subtype may pose a pandemic risk. Very little is known about the replicative capacity, immunogenicity, and correlates of protective immunity for low-pathogenicity H6 influenza viruses in mammals. We evaluated the antigenic and genetic relatedness of 14 H6 influenza viruses and their abilities to replicate and induce a cross-reactive immune response in two animal models: mice and ferrets. The different H6 viruses replicated to different levels in the respiratory tracts of mice and ferrets, causing varied degrees of morbidity and mortality in these two models. H6 virus infection induced similar patterns of neutralizing antibody responses in mice and ferrets; however, species-specific differences in the cross-reactivity of the antibody responses were observed. Overall, cross-reactivity of neutralizing antibodies in H6 virus-infected mice did not correlate well with protection against heterologous wild-type H6 viruses. However, we have identified an H6 virus that induces protective immunity against viruses in the North American and Eurasian lineages.

There are 16 known influenza A virus hemagglutinin (HA) subtypes (H1 to -16) and nine neuraminidase (NA) subtypes (N1 to -9), all of which have been isolated from aquatic birds (14, 47). While infection of poultry with some avian influenza (AI) viruses of the H5 and H7 subtypes can be highly pathogenic (HP) and fatal for poultry, less severe infections are seen with all AI virus subtypes, including non-HP H5 and H7 viruses. These viruses are referred to as low-pathogenicity AI viruses. LPAI H9N2 viruses have caused infections in humans that were associated with mild clinical symptoms (6, 33). Additionally, HA sequence analysis indicates that the 1957 and 1968 pandemics were caused by reassortant influenza viruses that derived two or three gene segments from an AI virus and the remaining gene segments from the previously circulating human influenza virus (13, 16, 25, 36). However, the AI viruses that were the source of the novel genes in the 1957 and 1968 pandemic viruses were not HPAI viruses. While human infections by AI viruses have been limited to viruses of the H1, H2, H3, H5, H7, H9, and H10 subtypes, serologic data suggest that poultry and live animal market workers in Asia have also been exposed to other AI virus subtypes, and a recent study demonstrated serologic evidence of infection by LPAI viruses among veterinarians in the United States (31, 38). Influenza pandemic preparedness has largely focused on AI viruses of the H5 and H7 subtypes, which include viruses that are HP in chickens and can cause serious illness and death in humans. However, as it is not possible to predict with certainty which AI subtype will cause the next pandemic, it is prudent to include LPAI subtypes in pandemic preparedness.

Very little is known about the replicative capacity, immunogenicity, and correlates of protective immunity for LPAI viruses in mammals. As we prepare for a potential influenza pandemic, the characterization of AI viruses of all subtypes in animal models is important for the evaluation of antiviral drugs and vaccines in the event that an LPAI virus is a precursor to a new pandemic influenza virus.

A/teal/Hong Kong/W312/97 (H6N1), a virus isolated from a duck in a live poultry market in Hong Kong (HK), was identified as a potential precursor to the HK/97 H5N1 viruses isolated during the 1997 H5N1 outbreak in humans; the H6N1 virus shared greater than 98% homology with the index human H5N1 virus A/HK/156/97 in all six internal protein genes and 97% homology in the NA gene (7, 23). A recent analysis suggested that although the H6N1 teal virus was closely related to the H5N1 viruses isolated in Hong Kong in 1997, it may not be a direct progenitor of the H5N1 viruses (7). However, it is notable that teal-like H6N1 viruses continue to circulate in poultry in southeastern China (8).

The first H6 influenza virus was isolated from a turkey in 1965, and since then H6 viruses have been isolated with increasing frequency from wild and domestic aquatic and terrestrial avian species throughout the world (1, 2, 8, 10, 11, 20, 26, 32, 37, 39, 41-44, 46). An 8-year surveillance study (1998 to 2006) that included more than 36,000 wild birds from Europe and the Americas found that H6 was the most abundantly detected influenza virus subtype and that this subtype had a broader host range than the other subtypes (30). Surveillance in southern China from 2000 to 2005 indicated that H6 viruses were prevalent year-round in terrestrial poultry (7). While the H6 viruses isolated to date have largely caused asymptomatic infections in waterfowl, infection of chickens with H6 viruses has been associated with decreased egg production, upper respiratory tract infection, morbidity, and increased mortality (1, 9, 10, 43, 46). The prevalence of H6 viruses in aquatic and terrestrial birds increases the potential for transmission to humans. Additionally, if these viruses reassort with circulating human influenza viruses and acquire the ability to transmit efficiently from person to person, an H6 virus could potentially spark an influenza pandemic.

Some AI viruses of the H5, H7, and H9 subtypes replicate in the respiratory tracts of mice and ferrets, the two animal models that are commonly used to study the pathogenesis of influenza viruses and to evaluate antiviral drugs and vaccines (15, 18, 19, 40, 48). In this study we evaluated the ability of 14 H6 viruses to productively infect and cause illness in mice and ferrets following intranasal inoculation. We characterized the immunogenicities of these viruses and the cross-reactivities of the neutralizing antibody responses induced in these two animal species and evaluated the extent of cross-protection induced by primary H6 virus infection in mice.

MATERIALS AND METHODS

Viruses.

Fourteen influenza A H6 subtype viruses were obtained from the influenza virus repository at St. Jude Children's Research Hospital, Memphis, TN (Table 1). Virus stocks were propagated in the allantoic cavity of 9- to 11-day-old embryonated specific-pathogen-free hen's eggs (Charles River Laboratories, Wilmington, MA) at 37°C as previously described (24). The 50% tissue culture infectious dose (TCID₅₀) titers were determined in Madin-Darby canine kidney (MDCK) cells (ATCC, Manassas, VA). Virus titers were calculated by the Reed and Muench method (35). The viruses formed plaques in MDCK and chicken embryo kidney cells (CEK), and comparable titers were observed in both cell types.

TABLE 1.

Avian influenza A virus isolates of the H6 subtype included in this study

Virus name	Virus abbreviation	Subtype	Phylogenetic lineagea
A/duck/Hong Kong/182/77	dk/HK/77	H6N9	Eurasian
A/goose/Hong Kong/W217/97	go/HK/97	H6N9	Eurasian
A/teal/Hong Kong/W312/97	tl/HK/97	H6N1	Eurasian
A/goose/Hong Kong/324/98	go/HK/98	H6N2	Eurasian
A/quail/Hong Kong/1721-30/99	qu/HK/99	H6N1	Eurasian
A/pheasant/Hong Kong/fy479/00	ph/HK/00	H6N1	Eurasian
A/chicken/California/465/00	ck/CA/00	H6N2	Eurasian
A/chicken/California/6643-TRA/01	ck/CA/01	H6N2	Eurasian
A/turkey/Massachusetts/3740/65	tk/MA/65	H6N1	North American
A/shearwater/Australia/1/73	sh/Aus/73	H6N5	North American
A/mallard/Alberta/89/85	ma/Alb/85	H6N2	North American
A/laughing gull/Delaware/4/90	lg/DE/90	H6N2	North American
A/mallard/Alberta/232/94	ma/Alb/94	H6N8	North American
A/shorebird/Delaware/127/97	sb/DE/97	H6N2	North American

^aThe phylogenetic analysis based on HA nucleic acid sequences was generated using the neighbor-joining method rooted on A/Ann Arbor/6/60 H2.

The HA gene of each virus was sequenced and amino acid sequences were analyzed with the Sequencher software as described previously (24). Phylogenetic analysis was performed comparing the HA of the 14 selected H6 viruses using the neighbor-joining method rooted to the HA gene of A/Ann Arbor/6/60 (H2N2). Primer sequences for reverse transcription-PCR and sequencing are available upon request.

Animals.

Four- to 6-week-old female BALB/c mice (Taconic Farms, Inc., Germantown, NY) were used in all mouse experiments. Mouse experiments were approved by the National Institutes of Health Animal Care and Use Committee and were conducted at the NIH. Seven- to 12-week-old male ferrets (Triple F Farms, Sayre, PA) were used in the ferret studies. Ferret experiments were approved by the MedImmune Animal Care and Use Committee and were conducted at MedImmune.

Virus replication in mice.

Groups of 20 mice lightly anesthetized with 4% isoflurane were inoculated intranasally (i.n.) with 10⁵ TCID₅₀ of each of the 14 H6 viruses. Each virus was diluted in Leibovitz (L-15) medium (Invitrogen) to a final volume of 50 μl/mouse. Mock-infected controls were inoculated with L-15 medium alone. Four mice per virus were sacrificed on days 1, 2, 3, 4, and 7 postinoculation, and lungs and nasal turbinates (NT) were harvested and stored at −80°C. Organs were weighed and homogenized in L-15 medium containing 1× antibiotic-antimycotic (Invitrogen) to make 10% and 5% (wt/vol) tissue homogenates of lung and NT, respectively. Titers of tissue homogenates clarified by centrifugation at 1,500 rpm for 5 min were determined in MDCK cells as described previously (24). Virus titers were calculated by the Reed and Muench method (35).

Virulence in mice.

Groups of four mice each were inoculated with 10¹ to 10⁷ TCID₅₀ of tl/HK/97, 10³ to 10⁷ of qu/HK/99, or 10⁴ to 10⁶ of mal/Alb/85 or dk/HK/77. The virus inoculation doses were selected based on the presence or absence of morbidity and/or mortality observed in mice inoculated with 10⁵ TCID₅₀ of each virus and the titer of the virus stock. Mice were observed daily for 14 days for clinical signs of illness, including weight loss, ruffled fur, and hunching. In accordance with our animal study protocol, mice were sacrificed if they lost ≥20% of their original body weight. The 50% mouse lethal dose (MLD₅₀) was calculated for lethal viruses using the Reed and Muench method (35). To determine if lethality of H6 viruses was associated with systemic spread of virus, groups of eight mice were inoculated with 10⁵ TCID₅₀ of tl/HK/97, qu/HK/99, or dk/HK/77. Four mice in each group were sacrificed on days 2 and 4 postinoculation, and lungs, NT, brain, and spleen were harvested and stored at −80°C. A 10% (wt/vol) concentration of homogenates of lungs and brains and 5% (wt/vol) homogenates of spleen and NT were generated, and virus titers in these organs were determined in MDCK cells.

Replication and virulence in ferrets.

Groups of ferrets negative for serum antibodies against H1N1, H3N2, and H6 viruses were lightly anesthetized with isoflurane and inoculated intranasally with 10^7.0 PFU of the H6 viruses. Two ferrets from each group were sacrificed on day 1, 3, or 5 postinoculation, and their lungs and NT were harvested. A 10% suspension of lung homogenate was prepared, and virus titers were determined based on the 50% egg infectious dose per gram of tissue (EID₅₀) as described previously (22). Virus titers in the NT suspensions were determined by plaque assay in CEK cells and expressed as log₁₀ PFU per gram of tissue. The body weight and temperature of these animals were measured prior to virus inoculation (day zero) and daily for 3 days. Fever was defined as a core body temperature of >103.8°F.

Immunogenicity in mice and ferrets.

Neutralizing antibody titers against homologous virus and 13 heterologous viruses were determined in postinfection mouse and ferret sera by microneutralization assay. Mouse sera collected prior to inoculation (prebleed) and at 28 and 42 days postinfection (dpi) and ferret sera collected prior to inoculation and at 21 dpi were used. Serial twofold dilutions of heat-inactivated serum were prepared, and equal volumes of serum and virus were mixed and incubated for 60 min at room temperature. The residual infectivity of the virus-serum mixture was determined in MDCK cells in four replicates for each dilution. The neutralizing titer was defined as the reciprocal of the highest dilution of serum that completely neutralized the infectivity of 100 TCID₅₀ of the virus as determined by the absence of cytopathic effect at day 4. A neutralizing antibody (NtAb) titer that was fourfold lower than the homologous neutralizing antibody titer was considered significantly different and indicative of a lack of cross-reactivity between viruses, as previously described by Archetti and Horsfall (3).

Efficacy of cross-protection induced by selected H6 viruses.

Groups of 20 mice were inoculated with 10⁵ TCID₅₀ of dk/HK/77 or mal/Alb/85 or a sublethal dose of 10³ TCID₅₀ (0.3 MLD₅₀) of tl/HK/97. NtAb responses to the primary infection were determined in sera collected on day 36 postinoculation. At 42 dpi, groups of four mice were challenged with 10⁵ TCID₅₀ of dk/HK/77 or mal/Alb/85, 30 MLD₅₀ of tl/HK/97, or 3 MLD₅₀ of qu/HK/99. The doses of the challenge viruses were chosen based on the results of the infectivity and virulence studies and the titer of the virus stock. Mice were sacrificed 4 days later, lungs and NT were harvested, and virus titers were determined.

RESULTS

Selection of avian influenza H6 viruses.

Representative H6 viruses isolated from different avian species on two continents over a span of 36 years were selected for evaluation (Table 1). Phylogenetic analysis was performed based on the nucleotide sequence of the HA gene of the 14 selected H6 viruses and the viruses grouped into two lineages, North American and Eurasian, as previously described (Table 1). The lineages generally corresponded to the geographic locations of the animals from which the viruses were isolated, with the exception of the two H6N2 viruses isolated from chickens in California (ck/CA/00 and ck/CA/01) (45). As previously reported, these two viruses grouped with the Eurasian lineage (44, 46). All of the H6 viruses selected for this study contain the sequence PQIETR↓G and had a single basic amino acid at the HA cleavage site. Additionally, each of the H6 viruses was able to agglutinate red blood cells derived from chicken, turkey, sheep, and horses, suggesting that they bind to both α-2,3- and α-2,6-linked sialic acid receptors (data not shown).

Replication of avian influenza H6 viruses in mice.

To evaluate the ability of H6 subtype AI viruses to replicate in mice, groups of mice were inoculated i.n. with 10⁵ TCID₅₀ of each of the 14 H6 viruses. Eleven of the 14 viruses replicated in the lungs, and 10 of them replicated in the NT of mice without prior adaptation (Fig. 1A and B). Virus titers detected in the lungs of infected mice were 10- to 1,000-fold higher than those detected in the NT, with a range of 10^1.6 to 10^7.3 TCID₅₀/g in the lungs and 10^1.9 to 10^3.9 TCID₅₀/g in the NT. Peak virus titers were detected between 2 and 4 dpi in the upper and lower respiratory tracts of mice. Three of the 14 viruses (go/HK/98, ph/HK/00, and sh/Aus/73) did not replicate in mice; virus levels were at or below the limit of detection in the lungs and NT of mice that received these viruses (data not shown). A correlation between NA subtype or phylogenetic lineage and the ability to replicate in the respiratory tract of mice was not apparent. Sequence analysis indicated that all of the H6 viruses tested contain glutamic acid at amino acid residue 627 in PB2 and not lysine, which has been associated with increased virulence in mice (21).

FIG. 1.

Replication and pathogenicity of selected avian influenza H6 subtype viruses in mice. (A and B) Groups of mice were inoculated i.n. with 14 different H6 viruses. Lungs and NT were harvested on the indicated days postinoculation. Virus titers in the lungs (A) and NT (B) were determined in MDCK cells. The lower limits of detection were 10^1.8 and 10^1.5 TCID₅₀ per gram for the NT and lungs, respectively. The means and standard errors from groups of four mice are presented. (C and D) Groups of four mice were inoculated with 10¹ to 10⁷ TCID₅₀ or 10³ to 10⁷ TCID₅₀ of tl/HK/97 (C) or qu/HK/99 (D), respectively. Mice were weighed daily for 14 days and were sacrificed if a weight loss of ≥20% from original weight was observed. Percent survival data are based on a 20% weight loss cutoff. The * denotes a mouse that died when a cage was dropped and not due to influenza virus infection.

Morbidity and mortality in mice.

To investigate whether H6 viruses that replicated to moderate or high titers in the respiratory tracts of mice are associated with morbidity and/or mortality, groups of mice were inoculated with serial 10-fold dilutions of tl/Hk/97, qu/HK/99, ma/Alb/85, and dk/HK/77 and observed for 14 days for clinical signs of illness, including weight loss, ruffled fur, and hunching. Significant weight loss and illness were observed in mice inoculated with tl/HK/97 and qu/HK/99 (Fig. 1C and D, left panel), but not in mice inoculated with ma/Alb/85 or dk/HK/77 (data not shown). Clinical signs of illness, including ruffled fur and hunching, were observed as early as 1 dpi in mice inoculated with 10⁶ to 10⁷ TCID₅₀ and by 3 to 6 dpi in mice inoculated with 10³ to 10⁵ TCID₅₀ of tl/HK/97. A transient weight loss of 5 to 10% was observed from 4 to 9 dpi in mice inoculated with 10² and 10³ TCID₅₀ of tl/HK/97, with weight returning to baseline by 11 dpi (Fig. 1C). Inoculation with 10⁴ to 10⁷ TCID₅₀ of tl/HK/97 resulted in significant, dose-dependent weight loss that rapidly progressed to ≥20% of body weight. Mice that lost >20% of body weight were euthanized on days 4, 6, 7, and 8 following infection with 10⁴, 10⁵, 10⁶, and 10⁷ TCID₅₀ of virus, respectively; the MLD₅₀ of tl/HK/97 was 10^3.5 TCID₅₀. Infection with qu/HK/99 was also associated with morbidity and mortality in mice (Fig. 1D); transient weight loss of 5 to 12% was observed at 2 to 8 dpi in mice infected with 10³ and 10⁴ TCID₅₀ of qu/HK/97, with a return to baseline by 13 to 14 dpi. Clinical signs of illness were observed as early as 1 to 3 dpi in mice inoculated with 10⁵ to 10⁷ TCID₅₀ of qu/HK/99 and by 6 dpi in mice inoculated with 10⁴ TCID₅₀ of virus. Mildly ruffled fur was observed in mice inoculated with 10³ TCID₅₀ of virus. Weight loss of ≥20% was observed in mice inoculated with 10⁵ and 10⁶ to 10⁷ TCID₅₀ of virus 4 and 6 dpi, respectively; the MLD₅₀ of qu/HK/99 was 10^4.5 TCID₅₀.

Lethality of HPAI viruses of the H5 and H7 subtypes in mice is associated with systemic spread of virus to extrapulmonary organs, including the brain, spleen, and kidneys (5, 15, 19, 24, 29). To determine if systemic spread of virus occurs in mice inoculated intranasally with lethal doses of tl/HK/97 and qu/HK/99, groups of eight mice were inoculated with 3 and 30 MLD₅₀ of each virus, respectively. Lungs, brain, and spleen were harvested from four mice per group at 2 and 4 dpi, and virus titers in these organs were determined. Despite high virus titers in the lungs at 2 and 4 dpi (10⁸ and 10⁷ TCID₅₀/g in mice inoculated with tl/HK/97 and qu/HK/99, respectively), virus was not detected in the brain or spleen of mice infected with either virus (data not shown). These data indicate that lethal infection with these H6 viruses is not associated with extrapulmonary spread of the virus.

Replication and pathogenicity of selected H6 viruses in ferrets.

Ferrets are a widely accepted animal model for evaluation of the immunogenicity and efficacy of human influenza virus vaccines. However, limited data exist on the susceptibility of ferrets to most subtypes of AI. We evaluated the ability of tl/HK/97, qu/HK/99, ma/Alb/85, dk/HK/77, and ph/HK/00 (viruses that replicated to varying levels in mice) to replicate in the respiratory tract of ferrets (Table 2). Virus titers in the lungs of ferrets inoculated with tl/HK/97 and qu/HK/99 peaked at 10^7.2 and 10^6.7 TCID/g, respectively, at 5 dpi, and peak titers in the NT were at least 10-fold lower than lung titers. Virus titers in the lungs of ferrets inoculated with dk/HK/77 peaked at 10⁷ EID₅₀/g at 3 dpi and remained at similar levels at 5 dpi, while mal/Alb/85 reached peak titers of 10⁷ EID₅₀/g in the lungs of ferrets at 1 dpi and decreased to 10^5.8 EID₅₀/g at 5 dpi. Interestingly, virus titers in the NT were 100-fold lower at 1 dpi and fell below the limit of detection by 5 dpi, suggesting that mal/Alb/85 may replicate less efficiently in ferrets than the other H6 viruses. Of the viruses tested in ferrets, only ph/HK/00 did not replicate efficiently, as indicated by a 100-fold decrease in lung virus titers from 10^5.7 to 10^3.2 EID₅₀/g from 1 to 5 dpi. Similarly, virus titers in the NT decreased from 10^4.3 EID₅₀/g at 1 dpi to below the limit of detection for one of the two animals by 3 dpi. ph/HK/00 also failed to replicate in the lungs and NT of mice.

TABLE 2.

Replication and pathogenicity of selected avian influenza H6 viruses in ferrets

Virus	Subtype	dpi	Mean titera		Mean peak % change in body wt ± SEM (day)b	Peak tempe (°F) ± SEM (day)
			Lung (log10 EID50/g ± SEM)	NT (log10 PFU/g ± SEM)
tl/HK/97	H6N1	1	6.0 ± 0.2	5.9 ± 0	−8.9 ± 4.3 (3)	104.2 ± 0.3 (1)
		3	6.3 ± 0.2	6.1 ± 0.3
		5	7.2 ± 0.4	5.4 ± 0.5
qu/HK/99	H6N1	1	6.2 ± 0	5.8 ± 0.4	−10.5 ± 5.2 (2)	104.8 ± 0.2 (1)
		3	6.0 ± 0.5	6.2 ± 0.1
		5	6.7 ± 0.2	5.6 ± 0.1
ma/Alb/85	H6N2	1	7.0 ± 0.2	5.0 ± 0.6	−4.3 ± 0.2 (3)	103.8 ± 0.4 (1)
		3	6.5 ± 0	3.9 ± 0.1
		5	5.8f	2.0
dk/HK/77	H6N9	1	6.7 ± 0.5	5.8 ± 0.5	−2.2 ± 0 (3)	102.7 ± 0.4 (2)
		3	7.0 ± 0.5	5.4 ± 0.1
		5	6.0 ± 0.5	6.1 ± 0
ph/HK/00	H6N1	1	5.7 ± 0.2	4.3 ± 0.1	−1.9 ± 2.3 (1)	104.2 ± 0.5 (1)
		3	3.0 ± 0.2	4.3; <3.1
		5	3.2 ± 0.3	3.4; <3.1
Mock			NDh	ND	+7.5 ± 2.8 (3)	102.1 ± 0.3 (1, 3)

^aGroups of six ferrets were inoculated i.n. with 10⁷ PFU of virus or were mock inoculated with medium alone. Virus titers were determined in lungs and NT harvested from two ferrets per virus on days 1, 3, and 5 postinoculation. Bold type indicates day of peak replication.

^bIn a separate study, groups of three ferrets were inoculated with 10⁷ PFU of virus, and body weights and temperatures were measured on days 0, 1, 2, and 3 postinoculation. Mean peak percent change in body weight and peak temperature are shown.

c A temperature of >103.8°F was classified as a fever.

d Results for a single ferret.

^eTiters from two individual ferrets.

^fND, not determined.

Additional studies were performed to determine if infection of ferrets with the selected H6 viruses was associated with clinical illness, including weight loss and fever. Infection with tl/HK/97, qu/HK/99, and ma/Alb/85 was associated with transient weight loss that reached a nadir at an average of 8.9, 10.5, and 4.3%, respectively (Table 2). Ferrets inoculated with tl/HK/97, qu/HK/99, and ma/Alb/85 had transient elevated body temperatures at or above the 103.8°F threshold; temperatures peaked at 1 dpi but returned to normal by 2 dpi. Although ferrets inoculated with dk/HK/77 did not exhibit weight loss, unlike the mock-inoculated animals, they did not gain weight over the duration of the experiment. Elevated body temperature was not observed in ferrets inoculated with dk/HK/77.

Homologous neutralizing antibody response to H6 viruses does not correlate with the level of virus replication in mice and ferrets.

The immunogenicities of the selected H6 viruses in mice and ferrets were compared using sera collected at 42 and 21 dpi, respectively, in microneutralization assays. Previous studies performed in our laboratory determined that the NtAb response developed earlier in ferrets than in mice (24). High NtAb titers were detected in the sera of mice inoculated with viruses that replicated to moderate to high titers in the lungs (tl/HK/97, qu/HK/99, ma/Alb/85, tk/MA/65, and ck/CA/00) (Table 3). However, high NtAb titers were also detected in mice inoculated with go/HK/98, although we were unable to detect replication of this virus in the respiratory tract of mice.

TABLE 3.

Homologous neutralizing antibody titers in mice and ferrets^a

Virus	Serum NtAb titer
	Miceb		Ferretsc
	GMTd	Range	GMT	Range
tl/HK/97	987	905-1280	854	806-905
qu/HK/99	570	508-640	2,281	2,032-2,560
ma/Alb/85	1,497	640-3,020	1,280	1,280
tk/MA/65	523	254-1,016	2,416	1,810-3,225
ck/CA/00	508	226--905	4,832	3620-6,451
ma/Alb/94	147	<10-905	9,123	8,127-10,241
dk/HK/77	185	127-254	380	226-640
ck/CA/01	67	20-226	226f
sb/DE/97	39	25-57	640	453-905
go/HK/97	226e		1,016	1,016
go/HK/98	3,835	1,810-6,451	8,611	7,241-10,241
ph/HK/00	123	63-202	1,918	1,810-2,032
sh/Aus/73	36	<10-320	640f

^aNeutralizing antibody titers were determined in sera collected from mice and ferrets in a microneutralization assay in MDCK cells.

^bSera were collected from four mice/virus at 42 days postinoculation.

^cSera were collected from two ferrets/virus at 21 days postinoculation.

^dGMT, geometric mean titer.

^eSera from four mice were pooled.

^fData are for one ferret.

NtAb titers detected in the sera of ferrets generally correlated with those in mice, but in neither model did the immunogenicity correlate fully with the ability of the viruses to replicate in the respiratory tract (Table 3). For example, high NtAb titers were detected in ferrets inoculated with ph/HK/00, although the virus replicated less efficiently in the respiratory tract of ferrets.

The cross-reactivity of the NtAb response in mouse sera does not correlate with that of ferret sera.

To determine the antigenic relatedness of the selected H6 viruses, the cross-reactivities of NtAbs were measured in a checkerboard fashion. In those cases where insufficient quantities of mouse sera collected 42 dpi were available, sera collected 21 dpi were used. A NtAb titer that was fourfold or lower than the homologous NtAb titer was considered to be significantly different and indicative of a lack of cross-reactivity between viruses, as previously described by Archetti and Horsfall (3).

Mouse sera.

The levels of cross-reactive antibodies induced in mice by the H6 viruses are summarized in Table 4. dk/HK/77, ck/CA/00, and ph/HK/00 elicited broadly cross-reactive antibodies that neutralized 10 of the 11 viruses evaluated, and tk/MA/65, tl/HK/97, and go/HK/97 induced NtAb against 6 or 8 of the viruses evaluated. Antibodies induced by ma/Alb/85, qu/HK/99, and go/HK/98 were poorly cross-reactive with the other H6 viruses despite high titers of NtAb against the homologous viruses. The NtAb titers induced by ph/HK/00, a virus that did not replicate in the respiratory tract of mice, were significantly lower than those induced by the other viruses. Antigenic relatedness as determined by NtAb cross-reactivity generally correlated with phylogenetic relatedness of the viruses based on the HA amino acid sequence (data not shown).

TABLE 4.

Titers of cross-neutralizing antibodies in mouse sera induced by infection with H6 viruses^a

Virus	Titer of mouse antisera raised against indicated virusb
	dk/77	ck/00	ph/00	tk/65	tl/97	go/97	sb/97	ma/85	qu/99	sh/74	go/98
dk/HK/77 (H6N9)	226	202	50	160	127	10	10	10	57	160	403
ck/CA/00 (H6N2)	254	226	80	320	226	63	113	160	127	10	905
ph/HK/00 (H6N1)	127	226	113	127	403	57	10	226	113	10	508
tk/MA/65 (H6N1)	226	101	113	640	640	80	10	80	806	10	10
tl/HK/97 (H6N1)	10	10	113	113	640	10	10	10	226	10	10
go/HK/97 (H6N1)	320	320	50	320	127	226	50	28	101	57	2032
sb/DE/97 (H6N2)	202	320	101	320	226	160	113	403	101	10	403
ma/Alb/85 (H6N2)	160	320	10	508	127	10	63	806	80	10	508
qu/HK/99 (H6N1)	202	320	160	640	806	160	40	101	806	NDc	80
sh/Aus/74 (H6N5)	226	202	226	453	254	113	40	113	160	254	80
go/HK/98 (H6N2)	1016	2560	640	1016	453	2560	320	905	640	80	12902

^aHomologous neutralizing antibody titers are indicated in bold. Underlined titers indicate cross-reactive antisera. Sera are arranged from most cross-reactive to least cross-reactive.

^bMouse antisera were collected at 28 days (dk/77, tl/97, qu/99, ck/00, tk/65, and ma/85) or 42 days (go/97, go/98, ph/00, sh/73, and sb/97) postinoculation. Antibody titers were determined in pooled sera from three to four mice per virus by microneutralization assay in MDCK cells.

^cND, not determined.

Ferret sera.

As shown in Table 5, the H6 viruses elicited poorly cross-reactive NtAbs in ferrets. dk/HK/77, ph/HK/00, and tl/HK/97 induced the most broadly cross-reactive antibodies in ferrets, but each of these antisera only neutralized 4 to 6 of the 11 viruses evaluated. Interestingly, high titers of homologous NtAbs were induced by ck/CA/00 and sh/Aus/73, but these antisera did not cross-react with the other viruses. Only 62% of the cross-neutralizing titers were concordant between postinfection mouse and ferret sera (Table 6).

TABLE 5.

Titers of cross-neutralizing antibodies in ferret sera induced by infection with H6 viruses

Virus	Titer of ferret antisera raised against indicated virusb
	dk/77	ck/00	tk/65	ph/00	tl/97	go/97	sb/97	ma/85	qu/99	sh/73	go/98
dk/HK/77 (H6N9)	226	226	50	113	101	101	20	10	80	101	453
ck/CA/00 (H6N2)	101	1280	32	57	10	10	10	10	10	10	63
tk/MA/65 (H6N1)	50	160	905	113	50	32	20	40	40	63	320
ph/HK/00 (H6N1)	28	32	113	508	640	10	10	10	453	10	10
tl/HK/97 (H6N1)	10	10	80	453	453	10	10	10	453	10	10
go/HK/97 (H6N1)	101	32	50	10	10	806	10	32	20	40	5120
sb/DE/97 (H6N2)	10	20	10	10	10	10	160	10	10	10	20
ma/Alb/85 (H6N2)	10	57	28	10	10	28	10	226	10	10	40
qu/HK/99 (H6N1)	80	160	453	640	1613	10	10	98	1613	63	20
sh/Aus/74 (H6N5)	226	160	80	160	226	905	57	10	101	1613	6451
go/HK/98 (H6N2)	202	57	254	101	63	403	57	202	50	127	6451

a Homologous neutralizing antibody titers are indicated in bold. Underlined titers indicate cross-reactive antisera.

^bFerret antisera were collected at 21 days postinoculation. Antibody titers were determined in pooled sera from two ferrets per virus by microneutralization assay in MDCK cells.

TABLE 6.

Discordance of cross-neutralization assays between mouse and ferret sera postinfection

Virus	Results with antisera raised against indicated virusa
	dk/77	ck/00	tk/65	ph/00	tl/97	go/97	sb/97	ma/85	qu/99	sh/74	go/98
dk/HK/77 (H6N9)	+	−	−	−	+	+	+	+	+	−	+
ck/CA/00 (H6N2)	+	+	−	−	−	−	−	+	+	+	+
tk/MA/65 (H6N1)	−	−	+	−	−	−	+	−	+	+	+
ph/HK/00 (H6N1)	−	−	+	+	+	−	+	+	+	+	+
tl/HK/97 (H6N1)	+	+	+	+	+	+	+	+	+	+	+
go/HK/97 (H6N1)	+	−	−	−	+	+	−	+	+	+	−
sb/DE/97 (H6N2)	−	−	−	−	−	−	+	−	+	+	+
ma/Alb/85 (H6N2)	−	−	−	+	+	+	−	+	+	+	+
qu/HK/99 (H6N1)	+	−	+	+	+	−	−	−	+	ND	+
sh/Aus/74 (H6N5)	+	−	−	+	+	+	+	+	+	+	−
go/HK/98 (H6N2)	+	−	+	−	−	+	+	+	−	+	+

^a+, concordant results; −, discordant results; ND, not determined.

Cross-neutralizing antibodies in postinfection mouse sera do not accurately predict protection against virus challenge.

To determine whether a robust NtAb response correlated well with protection, we infected mice with selected H6 viruses and subsequently challenged the mice with heterologous H6 viruses selected to be antigenically related or unrelated using mouse antisera. Mice were infected with dk/HK/77, ma/Alb/85, or a sublethal dose of tl/HK/97. Six weeks after primary virus infection, the mice were challenged with the homologous virus or heterologous H6 viruses dk/HK/77, tl/HK/97, ma/Alb/85, or qu/HK/99. Primary infection resulted in a significant homologous NtAb response (Fig. 2A). Virus titers in the lungs of mock-immunized mice challenged with dk/HK/77, tl/HK/97, ma/Alb/85, and qu/HK/99 were similar to titers seen in the previous experiments (Fig. 1 and 2B). Primary infection with dk/HK/77 provided complete protection from replication of the homologous virus dk/HK/77 and heterologous virus ma/Alb/85 and a significant reduction in the titers of tl/HK/97 and qu/HK/99 in the lungs. Interestingly, postinfection sera from mice infected with dk/HK/77 did not cross-react with tl/HK/97 (Table 4). Primary infection with tl/HK/97 induced complete protection from replication of all four challenge viruses in the lungs of mice despite poor cross-reactivity of NtAbs. Infection with ma/Alb/85 provided complete protection from replication of homologous virus ma/Alb/85 and heterologous virus dk/HK/77 and significantly reduced the titers of tl/HK/97 and qu/HK/99 viruses in the lungs, efficacies that would not have been predicted based on the cross-reactivity of postinfection mouse antisera. These results demonstrate that the cross-reactivity of NtAbs in mouse sera is not an accurate predictor of protective efficacy in the mouse model and that other correlates of protection must be sought. Although a similar study in ferrets would have been interesting, such a large study was not feasible for logistical reasons.

FIG. 2.

Homologous and heterologous virus challenge in mice following primary infection with selected H6 viruses. Mice were inoculated with medium alone or with dk/HK/77, ma/Alb/85, or tl/HK/97. (A) The geometric mean and range of neutralizing antibody titers against the homologous viruses in sera collected 42 dpi from 20 mice are presented. (B) Mice were challenged with dk/HK/77, ma/Alb/85, tl/HK/97, or qu/HK/99, and virus titers were determined in lungs (black bars) and NT (gray bars) harvested on day 4 postchallenge. Each bar represents the mean and standard error of titers from four mice. Statistical significance was determined using the nonparametric Mann-Whitney test. *, P ≤ 0.05. The lower limits of detection are represented by the gray (NT) and black (lungs) lines.

DISCUSSION

Avian influenza viruses of the H6 subtype have become endemic in both aquatic and terrestrial avian species in China and are the most abundantly detected influenza virus subtype in migratory birds in North America and Europe (30). Although human infection with this virus subtype has not yet been reported, low levels of H6-specific antibodies were detected in the sera of poultry and live animal market workers in China, and more recently in veterinarians exposed to birds in the United States, suggesting that clinically inapparent or unreported human infections with H6 viruses have occurred (31, 38). The ability of H6 AI viruses to infect humans is further supported by a study in which healthy human volunteers were experimentally inoculated with H6N1 and H6N2 viruses isolated from ducks (4). Evidence of viral replication associated with mild upper respiratory symptoms was observed in 2 of 11 volunteers inoculated with the H6N1 virus, though these findings were not accompanied by a rise in hemagglutination-inhibiting antibody titers. In contrast, a rise in Ab titer was detected in volunteers inoculated with an H6N2 virus in the absence of detectable virus replication. Taken together, these findings suggest that infection of humans with H6 avian influenza viruses is possible. The H6 viruses evaluated in this study are able to agglutinate red blood cells derived from chicken, turkey, sheep, and horse, suggesting that they can bind to both α-2,3- and α-2,6-linked sialic acid receptors (data not shown). Additionally, both α-2,3- and α-2,6-linked sialic acid receptors have been detected in the upper airway of ferrets (28). The continuing circulation of teal-like H6 viruses, which are closely related to and are a putative precursor of the 1997 H5N1 viruses, and the prevalence of H6 viruses in avian populations raise the possibility that after reassortment with a circulating human influenza virus, an H6 AI virus could pose a pandemic threat.

In this study, we established the utility of mice and ferrets as models for the evaluation of selected H6 virus vaccines and antiviral drugs. Prior to this study, 11 H6 viruses isolated from 1997 and 2003 from ducks and chickens had been evaluated in mice (10, 23, 27). Without prior adaptation, only 4 of these 11 viruses replicated in the respiratory tract of mice and only one of the viruses (A/teal/HK/W312/97) replicated to high titers and was associated with significant morbidity and mortality. We have extended the number of H6 viruses tested in mice to include an additional 10 viruses collected from aquatic and terrestrial birds on two continents spanning 36 years and included an evaluation of the viruses in ferrets. We identified 10 H6 viruses that productively infect mice; replication of two viruses (tl/HK/97 and qu/HK/97) was associated with significant morbidity and mortality in mice. Our observation that the tl/HK/97 virus caused significant morbidity and mortality in the absence of systemic virus spread is consistent with the report of Hoffmann et al., and the behavior of this H6 virus differs from the association between the extrapulmonary spread of virus and lethality seen with H5N1and HPAI H7 virus infections in mice (5, 23). Peak virus titers in the lungs of mice ranged from 10^2.9 to 10^7.3 TCID₅₀/g. Interestingly, although the highest titers in the lungs of mice were seen following infection with the two viruses that were lethal for mice, ma/Alb/85 and tk/MA/65 also replicated to peak titers within 5- to 10-fold of the lethal qu/HK/99 virus without causing notable morbidity, indicating that the virus titer in the lung is not the sole determinant of virulence.

Ferrets showed a similar pattern of response to infection as mice, although the viruses replicated to higher titers in ferrets than in mice and four of the five viruses replicated to similar peak titers in the lungs of the ferrets and were associated with morbidity, while the peak titers in the lungs of mice differed by 1,000-fold and only two viruses caused morbidity in mice. Interestingly, ph/HK/00, which did not replicate in mice, also replicated poorly in ferrets. Our findings that both mice and ferrets are susceptible to some avian influenza H6 viruses without prior adaptation demonstrate that mammals can be productively infected with H6 viruses, as they have been for some H5 and H7 viruses (5, 12, 15, 18, 24, 29). In our study there was no clear correlation between the source of the virus (aquatic or terrestrial birds) and the ability of the virus to replicate in the respiratory tract of mammals, although it has been suggested that avian influenza viruses capable of infecting humans may first need to adapt to terrestrial avian species (34).

The establishment of mouse and ferret models for H6 virus infection provided us the opportunity to compare the immunogenicity and antigenic relatedness of H6 viruses and to examine the breadth of the protective immune response induced in these animals. H6 virus infection induced similar patterns of NtAb responses in mice and ferrets, although, as with H7 viruses, the NtAb response developed earlier in ferrets than in mice (24). Interestingly, the levels of NtAbs detected in mice and ferrets did not consistently correlate with the level of virus replication in the respiratory tract; several of the viruses replicated poorly in the respiratory tract but induced a high-titer NtAb response. We also observed species-specific differences in the cross-reactivities of the antibody responses. Significantly greater cross-reactivity was observed with mouse sera than with ferret sera; however, this difference may be because sera were collected at different time points postinfection. Mouse sera were collected at 24 or 42 dpi and ferret sera were collected 21 dpi. We have observed that the NtAb response induced by H7 viruses also peaked at day 42 in mice (24). The discrepancies in our findings in the mouse and ferret models suggest that viruses from each avian influenza virus subtype should be independently and systematically evaluated in different animal models.

Cross-reactivity of neutralizing antibodies did not accurately predict cross-protection in mice; we saw evidence of cross-protection in the absence of cross-reactive neutralizing antibodies. The three viruses administered to mice each provided some degree of protection from heterologous challenge, regardless of phylogenetic lineage, geographic origin, or avian species of origin. These data suggest that although the microneutralization assay can be used to identify viruses that induce a high-titer antibody response, neutralization assay data may underestimate the breadth of the cross-protective immune response. Because the primary infection was induced by intranasal administration of virus, it is likely that mucosal and cellular immune responses were elicited and that they play a role in cross-protection. Of the H6 viruses evaluated, tl/Hk/97 is the best choice for a candidate vaccine, because infection with tl/Hk/97 completely protected against subsequent challenge with H6 viruses from both the North American and Eurasian lineages.

We have provided the first characterization of two animal models for avian influenza virus H6 subtype replication and immunogenicity and have identified an H6 virus that induces broad protective immunity against H6 viruses from different lineages. Studies examining the immunogenicity and efficacy of H6 vaccine derived from teal/HK/97 are currently under way.