Lack of transmission of H5N1 avian–human reassortant influenza viruses in a ferret model

To establish our ferret transmission model, we first evaluated the ability of two human H3N2 influenza viruses, A/Panama/2007/99 (Pan99) and A/Victoria/3/75 (Vic75), to undergo efficient respiratory droplet transmission by housing ferrets in adjacent cages, each with a perforated side wall that prevented direct contact but allowed spread of virus through the air. Three ferrets were inoculated intranasally (i.n.) with 10⁴ 50% ferret infectious dose (FID₅₀) of Pan99 [10^5.2 50% egg infectious dose (EID₅₀)] or Vic75 (10^4.5 EID₅₀), and 24 h later three naive ferrets were each placed in a transmission cage adjacent to an inoculated ferret. Inoculated ferrets achieved peak mean titers in nasal washes of 6.9 ± 1.5 or 7.8 ± 1.1 log₁₀ EID₅₀/ml, respectively on day 1 postinoculation (p.i.) (Fig. 1 A and B); and exhibited modest signs of illness, including sneezing beginning on day 2 p.i. (Table 1). Both viruses were efficiently transmitted to each of the three contact ferrets by day 3 or 5 postcontact (p.c.) as demonstrated by detection of virus in nasal secretions and seroconversion for hemagglutination inhibition (HI) antibody 14 days p.c. in all contact animals (Fig. 1 A and B and Table 1). These data indicate that both human H3N2 viruses were efficiently transmitted from inoculated to naive ferrets in a situation that did not allow for direct contact between animals or indirect contact with virus-contaminated food or surfaces. These results reflect the general transmissibility of human H3N2 influenza viruses in humans by respiratory droplets.

We next evaluated the ability of avian H5N1 influenza viruses to undergo respiratory droplet transmission between ferrets by inoculating three ferrets i.n. with 10⁴ FID₅₀ (10⁶ EID₅₀) of A/Hong Kong/486/97 (HK486), an H5N1 virus isolated from a human case patient in 1997. Mean nasal wash virus titers of the HK486-infected ferrets reached a maximal of 7.0 ± 1.1 log₁₀ EID₅₀/ml on day 5 p.i. (Fig. 1C), and signs of severe illness were observed (Table 1). The mean maximum weight loss in HK486-infected ferrets was 16.5% on days 2–7 p.i.; sneezing was observed only in one of three ferrets on day 4 p.i. (Table 1), and one of the ferrets was humanely killed on day 7 p.i. because of the onset of neurological symptoms. Virus was not detected in the nasal washes from any of the contact ferrets through day 9 p.c. (Fig. 1C). However, anti-HK486 HI antibody titers of 160 or 320 were detected in two of the three contact ferrets 17 days p.c., indicating that HK486 virus was transmitted to two of the contact ferrets in the absence of detectable virus in nasal secretions (Table 1). To determine whether HK486 virus would transmit more efficiently between cohoused ferrets, we next inoculated two animals i.n. with the same dose of HK486, and 24 h later we placed a naive ferret in the same cage with each inoculated animal. A low level of virus (10^2.3 EID₅₀/ml) was recovered on day 1 p.c. from one of the contact ferrets that had been cohoused with an inoculated animal that shed peak virus titers of 10^6.5 EID₅₀/ml (Fig. 2A), although both contact ferrets seroconverted with titers of 320 or 640 (data not shown). Two additional ferrets inoculated with a lower dose of HK486 virus (10³ EID₅₀) shed peak virus titers approximately three logs lower than the ferrets inoculated with the higher dose and failed to transmit virus to their cage mates, because the contact animals did not shed virus and did not seroconvert (data not shown).

The HK486 virus is genetically distinct from H5N1 viruses isolated since 1997; therefore, we also evaluated a 2003 and two 2005 H5N1 viruses for their ability to transmit between ferrets. A/Hong Kong/213/03 (HK213) contains a mutation at residue 223 (H5 numbering; residue 227 by H3 numbering) within the receptor-binding domain of the HA and was shown to bind to both α2,6 and α2,3 SA in vitro (13). Therefore, three ferrets were inoculated i.n. with 10⁶ EID₅₀ of HK213, and the next day three naïve ferrets were placed in adjacent transmission cages. HK213-infected ferrets achieved mean virus titers of 5.5 ± 0.3 log₁₀ EID₅₀/ml on days 1 and 3 p.i. (Fig. 1D) and exhibited some weight loss and lethargy but no sneezing (Table 1). Virus was not detected in the nasal washes from the contact ferrets nor was seroconversion detected by HI analysis of convalescent sera. A/Indonesia/5/05 (Indon05) and A/Vietnam/HN30408/05 (VN30408) are representative of the two distinct genetic clades of H5N1 viruses (clades 2 and 1, respectively) (14) that circulated in Asia during 2004–2005 and were isolated from human cases that had been associated with family clusters of H5N1 illness (4). In a contact transmission experiment, ferrets infected with 10⁶ EID₅₀ of Indon05 virus exhibited severe illness with substantial weight loss (18.8% mean maximum 7 days p.i.) and dyspnea but no sneezing; none of the ferrets survived past day 7 p.i. (Table 1). Although high titers of infectious virus were detected in the upper respiratory tract of Indon05-inoculated ferrets, virus was not detected in any of the nasal washes from the contact ferrets through day 9 p.c. (Fig. 2B), and convalescent sera collected from the contact animals lacked anti-Indon05 HI antibodies (Table 1). Ferrets infected with the same dose of VN30408 virus exhibited less severe illness and also failed to transmit the virus to contact animals (Table 1 and Fig. 2C). Taken together, these data demonstrate the lack of efficient transmission of H5N1 viruses in ferrets.

To evaluate the pandemic potential of H5N1 reassortant viruses and better understand the relative contribution of avian or human virus surface and internal protein genes to transmissibility, reassortant viruses containing various gene constellations from the avian H5N1 virus, HK486, and the human H3N2 virus, Vic75, were generated by using plasmid-based reverse genetics (rg). The parental viruses, containing all eight genes of either Vic75 (rgVic) or HK486 (rg486), were generated and their genetic identities were confirmed; the parental rg viruses exhibited virulence and transmissibility properties in ferrets similar to their WT counterparts (Table 2 and data not shown). To assess the contribution of the influenza virus internal protein genes, we first generated the reassortant virus, rgVic:486HANA, containing all six human virus internal protein genes and the avian virus H5 and N1 surface protein genes, and the reciprocal, rg486:VicHANA, containing avian virus internal protein genes and human H3 and N2 surface protein genes, and evaluated their infectivity for Madin-Darby canine kidney (MDCK) cells and eggs (Table 2). Although these two reassortant viruses had comparable infectivity titers in eggs, rgVic:486HANA achieved a titer that was 100-fold lower than that for the parental rg486 in MDCK cells, whereas rg486:VicHANA replicated as efficiently as the parental rgVic. An additional H5N1 reassortant virus, rg486:VicRNP, which contained the human virus ribonucleoprotein (RNP) genes (PB2, PB1, PA, NP), was generated and found to replicate to a higher titer in MDCK cells than rgVic:486HANA, although the infectivity of the two reassortants in embryonated eggs was similar (Table 2). Reassortant viruses, rg486:VicHANA and rg486:VicRNP, also exhibited higher infectivity for ferrets than rgVic:486HANA virus, and therefore had lower FID₅₀ values (1.5 and 1.5 compared with 3.0, respectively; expressed as the log₁₀ EID₅₀ required to give 1 FID₅₀). For comparison, WT Vic75 had an FID₅₀ value of 0.5, whereas HK486 had an FID₅₀ of 2.0. The data suggest that the infectivity for ferrets and MDCK cells of reassortant viruses bearing the HK486 HA and NA genes may be enhanced by the presence of the HK486 M and NS gene. In contrast, the reassortant virus bearing the Vic HA and NA genes replicated efficiently even when all internal genes, including the M and NS genes, were of avian virus origin. Certain gene constellations, such as rgVic:486HANAPB1 and rgVic:486HAPB1, which reflect the gene constellations of the viruses responsible for the 1957 and 1968 pandemics, respectively, could not be rescued even after multiple attempts, although the same plasmids were used to successfully rescue other reassortant combinations. These results suggest that certain combinations of genes from the avian and human influenza viruses used in this study may not be compatible for virus viability.

We next assessed the ability of the avian–human reassortant viruses to undergo respiratory droplet transmission in ferrets, because this is a key property of a pandemic virus. Three ferrets were inoculated i.n. with 10⁴ FID₅₀ of rgVic:486HANA or rg486:VicRNP (10⁷ or 10^5.5 EID₅₀, respectively), and the next day, three naïve ferrets were placed in adjacent transmission cages. Although all ferrets inoculated with either rgVic:486HANA or rg486:VicRNP shed virus in the upper respiratory tract for at least 5 days p.i. (Table 2), there was no evidence of transmission because contact animals did not shed virus or seroconvert (Table 3). In fact, the mean virus titers of each reassortant in nasal washes from ferrets were lower than titers observed with either parental strain (Table 2). Furthermore, the inoculated ferrets exhibited only minor weight loss, indicating that both reassortant viruses were substantially attenuated for ferrets compared with the rg486 parental strain, which caused a mean weight loss of 21% (Table 2).

Because none of the reassortants possessing the HA and NA of the avian HK486 virus demonstrated the ability to transmit, we next determined whether the human virus HA and NA were required for transmission. Ferrets were inoculated with the same dose of the reassortant virus, rg486:VicHANA [10⁴ FID₅₀ (10^5.5 EID₅₀)], and virus was recovered from all three infected animals with a mean peak virus titer of 5.2 ± 0.6 log₁₀ EID₅₀/ml, which was similar to titers achieved in ferrets inoculated with the rg parental virus, rgVic (Table 2). Nevertheless, the rg486:VicHANA virus failed to transmit as efficiently as rgVic; virus was not detected in the nasal washes collected from any of the rg486:VicHANA contact ferrets, and only one of three contact ferrets exhibited a modest rise in serum HI antibody on day 20 p.c., indicating that some transmission of the reassortant had occurred (Table 3). Interestingly, the contact animal that seroconverted was exposed to an inoculated ferret that had shed the highest level of virus detected from any of the reassortant-inoculated ferrets evaluated (10^5.5 EID₅₀/ml). These data suggest that the avian internal protein genes of rg486:VicHANA virus reduce the transmission efficiency of the human H3N2 virus in ferrets without affecting the replication efficiency.

The acquisition of mutations, such as those that confer the human α2,6 SA receptor-binding preference, is likely to be important for an avian influenza virus or an avian–human reassortant virus to transmit efficiently among humans and is thought to occur during adaptation in the new host (10). To determine whether an H5N1 reassortant virus could acquire mutations that would enhance transmissibility through multiple rounds of replication in a mammalian host, we passaged rg486:VicRNP virus (the H5N1 reassortant that had replicated most efficiently in ferrets) five times in ferrets and tested the resulting virus, rg486:VicRNPF5, for its ability to undergo respiratory droplet transmission. Although peak nasal wash titers of rg486:VicRNPF5-inoculated animals were similar to those for ferrets inoculated with rgVic (5.1 ± 0.5 log₁₀ EID₅₀/ml) (Table 2), transmission of rg486:VicRNPF5 was not detected (Table 3). Sequence analysis of the complete rg486:VicRNPF5 genome revealed only a single amino acid change (S353F) within the NP gene. These data and the observation that most avian viruses isolated from humans retain their receptor preference for α2,3 SA (13, 26) suggest that there is a level of stability of receptor specificity among circulating H5N1 viruses.

The human H3N2 influenza A viruses Vic75 and Pan99 and the avian H5N1 influenza A viruses HK486, HK213, Indo05, and VN30408 were used in this study. Virus stocks were prepared in 10-day-old eggs as described (29) except that human viruses were incubated at 33.5°C for 48 h, whereas reassortant viruses were incubated at 37°C for 48 h. All research with H5N1 viruses or reassortant viruses was conducted under biosafety level 3 containment, including enhancements required by the U.S. Department of Agriculture and the Select Agent Program (see interim guidance at www.cdc.gov/flu/h2n2bsl3.htm).

Plasmid-based rg was used to generate the reassortant viruses used in this study. As described (30), all eight Vic75 genes were amplified and cloned into a viral RNA (vRNA) expression plasmid, and Vic75 PB2, PB1, PA, and NP genes were cloned into an mRNA expression plasmid. All eight genes of HK486 were amplified from vRNA by RT-PCR with a One-Step RT-PCR kit (Qiagen, Valencia, CA) and cloned into a bidirectional plasmid (pBD) that possesses vRNA expression elements from pPolISapIRib (31, 32) and mRNA expression elements from pCI (Promega, Madison, WI). Reassortant viruses were rescued as described (33) by using the 12-plasmid Vic75 system, the eight-plasmid HK486 system, or a combination of the two. The genetic makeup of each reassortant virus was confirmed by sequencing as described (29).

Male Fitch ferrets, 6–12 months of age (Triple F Farms, Sayre, PA), that were serologically negative by HI assay for currently circulating influenza viruses were used in this study. Ferrets were housed throughout each experiment in cages within a Duo-Flo Bioclean mobile clean room (Lab Products, Seaford, DE). Baseline serum, temperature, and weight measurements were obtained before infection. Temperatures were measured with an s.c. implantable temperature transponder (BioMedic Data Systems, Seaford, DE). Ferrets were inoculated, and nasal washes were collected and analyzed as described (29). FID₅₀ titers were determined by inoculating groups of nine ferrets i.n. with serial 10-fold dilutions of virus, calculated by using the method of Reed and Muench (34), and expressed as the EID₅₀ value corresponding to 1 FID₅₀.

For the respiratory droplet transmission experiments, ferrets were housed in adjacent transmission cages, each modified so that a side wall was replaced with a stainless-steel, perforated wall with holes 1–5 mm in diameter and spaced 3 mm apart to facilitate the transfer of respiratory droplets through the air while preventing direct contact between ferrets and indirect contact with the bedding and food of neighboring ferrets. The use of the term “respiratory droplet transmission” throughout this article refers to transmission in the absence of direct or indirect contact and does not imply an understanding of the droplet size involved in virus spread between ferrets. A total of six ferrets were used for each respiratory droplet transmission experiment. Three ferrets were inoculated with 10⁴ FID₅₀ of virus, unless otherwise stated, and each was placed in a separate cage. Twenty-four hours later (day 1 p.i. for the inoculated ferrets and day 0 p.c. for the contact ferrets), three naive ferrets were each placed in a cage adjacent to an inoculated ferret. To prevent inadvertent physical transmission of virus by the investigators, the contact ferrets were always handled first, and all items that came into contact with the ferrets or their bedding were decontaminated between each ferret. For contact transmission experiments, two or three ferrets were each placed in an unmodified cage with solid walls and inoculated i.n. with 10⁴ FID₅₀ of virus, unless otherwise stated. Twenty-four hours later, a naive ferret was placed in the same cage with each inoculated ferret. Clinical signs were monitored daily in all ferrets for at least 14 days p.i./p.c. If any ferret lost >25% of its body weight or exhibited neurologic symptoms, it was humanely killed.

HI analysis was conducted as described (35) by using convalescent sera collected from ferrets 14–33 days p.i./p.c. Sera collected from ferrets included in the transmission experiments with Vic75, Pan99, or reassortant H3N2 viruses were tested for H3-specific antibodies by using turkey RBCs and virus bearing the homologous HA. Ferret sera from the transmission experiments with HK486, Indon05, VN30408, or reassortant H5N1 viruses were tested for H5-specific antibodies by using virus bearing the homologous HA and horse RBCs as a means of more sensitive detection (35).