|Home | About | Journals | Submit | Contact Us | Français|
The highly pathogenic H5N1 avian influenza virus emerged from China in 1996 and has spread across Eurasia and Africa, with a continuous stream of new cases of human infection appearing since the first large-scale outbreak among migratory birds at Qinghai Lake. The role of wild birds, which are the natural reservoirs for the virus, in the epidemiology of the H5N1 virus has raised great public health concern, but their role in the spread of the virus within the natural ecosystem of free-ranging terrestrial wild mammals remains unclear. In this study, we investigated H5N1 virus infection in wild pikas in an attempt to trace the circulation of the virus. Seroepidemiological surveys confirmed a natural H5N1 virus infection of wild pikas in their native environment. The hemagglutination gene of the H5N1 virus isolated from pikas reveals two distinct evolutionary clades, a mixed/Vietnam H5N1 virus sublineage (MV-like pika virus) and a wild bird Qinghai (QH)-like H5N1 virus sublineage (QH-like pika virus). The amino acid residue (glutamic acid) at position 627 encoded by the PB2 gene of the MV-like pika virus was different from that of the QH-like pika virus; the residue of the MV-like pika virus was the same as that of the goose H5N1 virus (A/GS/Guangdong [GD]/1/96). Further, we discovered that in contrast to the MV-like pika virus, which is nonpathogenic to mice, the QH-like pika virus is highly pathogenic. To mimic the virus infection of pikas, we intranasally inoculated rabbits, a species closely related to pikas, with the H5N1 virus of pika origin. Our findings first demonstrate that wild pikas are mammalian hosts exposed to H5N1 subtype avian influenza viruses in the natural ecosystem and also imply a potential transmission of highly pathogenic avian influenza virus from wild mammals into domestic mammalian hosts and humans.
Highly pathogenic avian influenza (HPAI) is an extremely infectious, systemic viral disease that causes a high rate of mortality in birds. HPAI H5N1 viruses are now endemic in avian populations in Southeast Asia and have repeatedly been transmitted to humans (9, 14, 27). Since 2003, the H5N1 subtype has been reported in 391 human cases of influenza and has caused 247 human deaths in 15 countries, leading to greater than 60% mortality among infected individuals (38). Although currently incapable of sustained human-to-human transmission, H5N1 viruses undoubtedly pose a serious threat to public health, as well as to the global economy. Hence, preparedness for such a threat is a global priority (36).
Wild birds are considered to be natural reservoirs for influenza A viruses (6, 18, 21, 35, 37). Of the 144 type A influenza virus hemagglutinin-neuraminidase (HA-NA) combinations, 103 have been found in wild birds (5, 7, 17, 37). Since the first HPAI outbreak among migratory wild birds appeared at Qinghai Lake in western China in May 2005 (3, 16, 25, 34, 41), HPAI viruses of the H5N1 subtype have been isolated from poultry throughout Eurasia and Africa. The continued occurrence of human cases has created a situation that could facilitate a pandemic emergence. There is heightened concern that wild birds are a reservoir for influenza A viruses that switch hosts and stably adapt to mammals, including horses, swine, and humans (11, 19, 22, 37).
Despite the recent expansion of avian influenza virus (AIV) surveillance and genomic data (5, 17, 20, 21, 33, 40), fundamental questions remain concerning the ecology and evolution of these viruses. Little is known about how terrestrial wild mammals within their natural ecological systems affect HPAI H5N1 epidemiology or about the virus's effects on public health, though there are many reports of natural and experimental H5N1 virus infection in animals belonging to the taxonomic orders Carnivora (12, 13, 15, 28, 29) and Artiodactyla (15). Herein, we provide the results of our investigation into H5N1 virus infection in wild pikas (Ochotona curzoniae of the order Lagomorpha) within their natural ecological setting. We describe our attempt to trace the circulation of H5N1 viruses and to characterize pika H5N1 influenza virus (PK virus).
To estimate the infection of wildlife with influenza A virus in a natural ecosystem, in China's Western Qinghai Province, from August 2006 to December 2007, Ochotona curzoniae (black-lipped) pikas were caught in the regions around Qinghai Lake (Fig. (Fig.1A),1A), including Bird Island (BI), Garila (GRL), Shenhekou (SHK), Quanwan (QW), and Heimahe (HMH); in Xingxinghai of Maoduo County; and in Longbaotan of Yushu County. The area of each sample collection point was 1 km2. Serum and tissue samples were collected from live-caught pikas and stored in liquid nitrogen.
Following the protocol outlined by the Office International des Épizooties (OIE), antibodies were detected in pika serum samples using an HA inhibition (HI) assay using the A/bar-headed goose/Qinghai/0510/05 (H5N1) virus as an antigen. To remove nonspecific HAs, serum samples were treated with a receptor-destroying enzyme (Takeda Chemical Industries, Osaka, Japan) prior to the HI test. A serum sample was considered to be positive for influenza virus if its HI titer was ≥16. The HI-positive sera were further analyzed for neutralizing antibodies against the representative isolates. Tenfold serial dilutions were screened for antibodies to the representative viruses in a 50% tissue culture infective dose (TCID50) assay as previously described (2).
Virus isolation and identification were performed as described previously (41). Briefly, each tissue sample was ground separately in phosphate-buffered saline (pH 7.2) and propagated for three passages in 10-day-old, specific-pathogen-free (SPF), embryonated chicken eggs. The HA and NA subtypes were determined by conventional HI and neuraminidase inhibition tests according to the OIE manual. The 50% egg lethal doses (ELD50) of the virus isolates were also determined as previously described (2, 41). All experiments with the H5N1 isolates were performed in a biosafety level 3 laboratory.
Analysis of the genomic sequences of the virus isolates was performed as described previously (41). Briefly, viral RNA was obtained from the virus-infected allantoic fluid using the Trizol LS reagent (Invitrogen, Carlsbad, CA). Reverse transcription-PCR was conducted using a one-step reverse transcription-PCR kit (Qiagen, Hilden, Germany) with primers specific for influenza A virus (data not shown). The primer sequences and amplification conditions are available upon request. The PCR products were sequenced using a CEQ data transmission and control system quick-start kit on a CEO 8000 DNA sequencer (Beckman Coulter). Nucleotide and amino acid sequences were aligned using the Clustal W program running within the BioEdit software package (version 5.0.9). With MEGA (version 4), phylogenetic trees of nucleotide sequence were constructed and amino acid substitutions were calculated. Genome sequences of H5N1 influenza viruses from GenBank (data not shown) were used as references.
Pathogenicity tests were performed in accordance with the instructions provided in the OIE manual. Five-week-old, SPF White Leghorn chickens (10 chickens/group, obtained from the Merial Vital Laboratory Animal Technology Co. Ltd., Beijing, China) housed in negative-pressure isolator cages with HEPA-filtered air were inoculated intravenously (i.v.) with 0.2 ml of a 1:10 dilution of bacterium-free allantoic fluid containing a virus isolate to determine the i.v. pathogenicity index.
Eight-week-old, SPF BALB/c mice, obtained from the Experiment Animal Center of Zhejiang University, Hangzhou, China, were divided into five groups with 20 mice per group. The mice in each group were inoculated intranasally (i.n.) under anesthesia with the pika virus isolates BI, SHK, GRL, QW, and HMH at a dose of 0.05 ml of a solution containing 106 ELD50/0.1 ml. One day later, two uninfected mice were placed in direct contact with the inoculated mice. Beginning at 2 days postinoculation (dpi), two inoculated mice were sacrificed each day until 10 dpi for use in virus titration assays, histopathology assays, and immunofluorescence assays (IFA). Tissue samples were homogenized in 1 ml of cold phosphate-buffered saline and were titrated for virus infectivity on MDCK cell monolayers with initial dilutions of 1:10 (lung) and 1:2 (other organs). The remaining inoculated mice were monitored daily for weight loss and mortality. The TCID50 was calculated using the method of Reed and Muench (24). Procedures for histopathology and IFA followed those previously described (2, 41).
Thirty-day-old New Zealand White rabbits, purchased from the Experimental Rabbit Farm of Zhejaing Provincial Academy of Agriculture, Zhejiang, China, were divided into three groups, with 17 rabbits per group. The rabbits in each group were inoculated i.n. under anesthesia with 106.0 ELD50 of the PK virus isolates BI and QW/0.1 ml in a volume of 1 ml. To determine virus transmissibility, five rabbits inoculated i.n. with 1 ml normal saline were introduced into each group. Daily, during the experimental period, all rabbits were weighed, their body temperatures were recorded, and each was observed for signs of disease. The rabbits were monitored for the presence of infectious virus from dpi 1 to dpi 21 by collecting swabs from the nasal cavity, pharynx, and rectum. Tissues were collected regularly at 7, 14, and 21 dpi from each rabbit for virus isolation, histopathology, and IFA. The collected samples from the infected rabbits were titrated for infectivity by determining the ELD50. As before, after histopathological analyses, the frozen sections for IFA were stained using a primary mouse monoclonal antibody to influenza A virus matrix protein and fluorescein-labeled affinity-purified goat anti-mouse immunoglobulin G (KPL) as the secondary antibody. To detect HI and neutralizing antibodies to H5N1 virus, rabbits were bled and sera collected regularly each day after inoculation from days 5 to 44.
The genomes of the H5N1 viruses in this study have been deposited in GenBank (accession no. GQ227357 to GQ227398).
Of the 82 pika serum samples tested for antibodies against the A/bar-headed goose/Qinghai/0510/05 (H5N1) virus, only 11 samples (13.41%) were positive by HI. Eleven samples (HI titer > 16) with antibodies against the H5 subtype of influenza virus were used to neutralize the A/bar-headed goose/Qinghai/0510/05 (H5N1) virus; neutralizing antibody titers were 1:37 to 1:47. These data suggest that free-ranging wild pikas have been exposed to the H5 subtypes of the influenza virus.
The virus isolates were detected from the brains, lungs, and rectums of the live-caught pikas. The samples yielded five virus isolates (3.4%) obtained from 147 pikas collected in April 2007 and December 2007. Subtype analysis showed that the five isolates belonged to the H5N1 subtype of influenza A virus. The samples were given the names A/PK/QH/BI/0704/2007 (BI virus), A/PK/QH/SHK/0712/2007 (SHK virus), A/PK/QH/GRL/0712/2007 (GRL virus), A/PK/QH/QW/0712/2007 (QW virus), and A/PK/QH/HMH/0712/2007 (HMH virus). After i.v. injection, all five virus isolates killed inoculated chickens within 18 to 31 h, with i.v. pathogenicity index values of 2.93 to 3.0, indicating that all five PK viruses are highly pathogenic to chickens.
In order to analyze the diversity of the five H5N1 viruses isolated from wild pikas, their complete genomes were sequenced and maximum-likelihood phylogenetic trees were constructed with sequences available from public databases. Eight genes of the isolated PK viruses (QW, HMH, GRL, and SHK viruses) shared 99.3 to 100% nucleotide homology with sequences available from public databases, except for the BI virus (data not shown). Except for the PA gene, which displayed 96.0% homology with that of the HMH virus, the seven other genes of the BI virus showed more than 99.2% nucleotide identity to A/duck (DK)/Korea (KR)/Asan5/06. The four other PK viruses shared the greatest nucleic sequence homology with other H5N1 viruses within the HA fragment (97.4 to 97.7% homology; A/CK/Guangxi [GX]/2461/04), the NA fragment (97.5 to 97.8%; A/CK/Guiyang [GY]/3055/05), the NS fragment (98.7 to 98.9%; A/DK/Hunan [HN]/303/04), the NP fragment (98.2 to 98.5%; A/DK/Fujian [FJ]/1734/05), the PA fragment (97.8 to 98.0%; A/DK/FJ/1734/05), the PB1 segment (98.5 to 98.7%; A/bar-headed goose [BHGS]/QH/12/05), and the PB2 segment (98.1 to 98.3%; A/DK/GX/2775/05). However, their MA fragments revealed less than 90% homology with that of the previously reported H5N1 virus, excepting the GS/Guangdong (GD)/1/96 virus (90.3% homology). Similar findings were also obtained with the encoded protein of the PK viruses. According to the report of the WHO/OIE/FAO H5N1 Evolution Working Group (10), phylogenetic analysis (Fig. (Fig.1B1B and data not shown) revealed that seven genes of the BI virus belong to clade 2.2 (QH-like lineage), but its PA gene lies in a branch of the SHK virus. In contrast, the HA genes of the four other PK viruses are distributed in clade 2.3.2 (mixed/Vietnam sublineage, MV-like lineage), which comprises the H5N1 viruses circulating mainly in chickens and waterfowl in China's southern provinces and southern Asia. The phylogeny of the NA gene was similar to that of the HA gene. Similarly, a phylogenetic tree for the NA gene and the genes for the internal proteins PB2, PB1, PA, NP, M, and NS also revealed that the BI virus belongs to the wild bird H5N1 lineage and that the four other PK viruses formed another distinct evolutional clade, indicating that the five isolated PK viruses have descended via two different evolutionary paths, a QH-like lineage and an MV-like lineage.
The QH-like and MV-like PK viruses, similar to the H5N1 viruses within clade 2 (39, 41), maintained the 325RRRKKR330 motif of multiple basic amino acids at the HA cleavage site, characteristic of HPAI viruses. In addition, the 20-amino-acid stretch CNQSIITYENNTWVNQTYVN, conventionally located between amino acids 49 and 68 of neuraminidase, was absent in these viruses; a similar deletion in the H5N1 virus is associated with its extreme virulence (2, 23). However, MV-like viruses had lost two amino acid residues, 327R and 328K. With the exception of the PB2 protein, unique amino acid substitution sites were observed in all MV-like virus-encoded proteins, including 9 amino acid substitutions in the HA protein, 7 amino acid substitutions in the NA protein, single amino acid substitutions in the NP, NS, and PB1 proteins, 8 amino acid substitutions in the PA protein, and 12 amino acid substitutions in the M protein (data not shown). In particular, mutations are more abundant in the MA protein, suggesting that the PK viruses of the MV lineage possess a unique molecular characteristic compared with those of previously reported H5N1 viruses.
To assess the pathogenicity of the PK viruses in mice, 8-week-old, SPF BALB/c mice were i.n. inoculated. Virus replication, morbidity, and mortality were assessed. The BI virus-infected mice suffered weight loss from the beginning of dpi 3 and 65% mortality within 14 dpi, while the remaining PK virus-infected mice did not exhibit weight loss or death during the experiment period (Fig. (Fig.2).2). The mean time to death of BI virus-infected mice was 7.5 days (range, 3 to 10 days). Necropsy examination revealed that the BI virus-infected mice were more severely affected, displaying hemorrhaging in 100% of the lung area and having swollen, yellow livers. Microscopic lesions of BI virus-infected mice showed severe hemorrhagic and histiocytic pneumonia, with infiltration of neutrophils, an infiltration focus of lymphocytes in the cerebral cortex, congestion of the glomerular capillary bed, and moderate hepatic granular degeneration. In contrast, no gross or microscopic lesions were observed in the remaining PK virus-infected mice. We were able to reisolate the PK BI virus from the lung (105.35 ± 0.7 TCID50/0.1 ml) and brain (102.96 ± 0.16 TCID50/0.1 ml) tissues of mice that died but not from tissues of the liver, spleen, heart, kidney, pancreas, duodenum, or rectum of the same mice. In contrast, the other PK H5N1 viruses appeared to replicate only to a low titer (101.43 to ~102.18 TCID50/0.1 ml) in the lungs of the mice that were inoculated at 2 to 4 dpi. No virus was isolated from the feces of any infected or cohabitating mice.
When the rabbits were inoculated i.n. with the BI and QW viruses, they showed no signs of sickness, body weight loss, or death. As shown in Fig. Fig.3A,3A, the body temperature of the BI and QW virus-infected rabbits was elevated (P < 0.05 or P < 0.01) between 4 to 8 dpi, with individual rabbits having temperatures over 40°C at 4 dpi and persistent high fevers for 2 to 3 days. Serological analyses (data not shown) revealed that the rabbits seroconverted at 9 dpi for 80% and at 11 dpi for 100% of the BI virus-inoculated rabbits. However, for QW virus-inoculated rabbits, HI antibodies were 80% seroconverted at 11 dpi and 100% at 13 dpi. In both BI and QW virus-infected rabbits, anti-H5 HI antibodies appeared to peak (1:23 to 1:25) 2 to 10 days after seroconversion and subsequently maintained a lower titer for at least 44 dpi. The contact and mock-infected rabbits did not exhibit significant body temperature changes, nor did they display any anti-H5N1 HI antibody seroconversion. These data show that BI and QW H5N1 viruses isolated from pikas result in rabbit infections under experimental conditions.
The inoculated rabbits began to shed virus from the nasal cavity at 3 dpi and stopped shedding at 10 dpi. The virus shedding cycles were 7 days for the BI virus and 8 days for the QW virus (Table (Table1).1). However, no virus was detected in the oropharyngeal or rectal swabs of infected rabbits, indicating that the virus was shed only from the upper respiratory tracts of the infected rabbits. Necropsy evaluations showed gross lesions in the lungs of infected rabbits, with hemorrhagic spots, at 7 (BI virus and QW virus) and 14 (BI virus) dpi. Histological lesions (Fig. (Fig.3B)3B) observed in the infected rabbits included mainly desquamation of mucosal epithelial cells of turbinate bone and the trachea, mild interstitial pneumonitis with a focal infiltration of lymphocytes around the bronchi, and hyperemia of renal glomerular and heart muscle capillaries. When the frozen tissue sections were examined via indirect immunofluorescence, the viral antigens were detected at 7 dpi in turbinate bone (Fig. (Fig.3B)3B) but not in other tissues of the infected rabbits. The BI and QW viruses were reisolated mainly from turbinate bone of all infected rabbits at 7 dpi but were occasionally identified in pancreas (7 dpi), trachea, and lung (14 dpi) tissues of infected individual rabbits, demonstrating that the PK virus replicates mainly in the upper respiratory tracts of infected rabbits.
After the first H5N1 outbreak in the migratory wild bird population at Qinghai Lake (16, 41), China, HPAI spread to European and African poultry populations. One year after the QH05 outbreak, fatal H5N1 viruses that infected more species of wild birds were reemerging in some areas of the Qinghai Province and Tibet Autonomous Region (34). Considering Qinghai Lake's geographical status in bird migration, the role of migratory birds, possibly as carriers in the circulation of the viruses along the flyway, has been debated extensively (11, 19, 22). The H5N1 virus in migratory birds has become a grave concern. In this study, we first investigated PK virus infections in the Qinghai Province of China. The analysis of anti-AIV HI antibodies confirms that the H5 AIV infection has a high frequency in pikas. Subsequently, five isolates of the H5N1 influenza virus were identified in samples from live-caught pikas, and two kinds of PK viruses, MV-like and QH-like lineage viruses, were identified (Fig. (Fig.1B1B and data not shown), demonstrating that two lineages of H5N1 viruses currently appear in pikas. Therefore, we confirm that free-ranging wild pikas are a mammal host natively infected by the H5N1 influenza A virus in the natural environment. However, whether pikas are a native mammal carrier of H5N1 virus or are a host infected by H5N1 virus from other animals needs further investigation.
Notably, in our study, seven genes of the BI virus share the greatest homology to genes of A/DK/KR/Asan5/06, which is related to the QH-like viruses responsible for the HPAI H5N1 outbreak in Eurasian and African poultry populations (1, 4), but its PA gene (96.0% homology) appears to have greatest identity to the MV-like PK viruses (data not shown), indicating that the BI virus is a reassortant H5N1 virus closely associated with the migration of wild birds. Thus, pikas are infected not only with the QH-like H5N1 virus but also with an H5N1 virus circulating in migratory birds that has rearranged with PK virus. These data confirm that the QH-like PK virus circulates between migratory birds and pikas. The pika, which belongs to the Ochotona genus of the Ochotonidae family of the order Lagomorpha, is a small, wild, nonmigrating herbivorous mammal in the natural ecosystem and a major food source for all raptorial birds and carnivorous mammals (26). The wild birds, such as bar-headed geese and ruddy shelducks, forage weeds as food. An interesting question is how the QH-like H5N1 virus circulating in wild birds transmits between pikas and wild birds. One possible hypothesis is that pikas are infected by H5N1 virus circulating in wild birds at the common weed-foraging sites. Therefore, these data imply that the H5N1 virus circulating in migratory birds has caused environmental pollution in the regions around Qinghai Lake of western China.
Unlike with the QH-like PK virus, seven of eight genes for the MV-like H5N1 viruses share high nucleotide homology to genes in the different H5N1 sublineages that have circulated in poultry in the Chinese Provinces of Guangxi, Guizhou, Hunan, and Fujian since 2004 (data not shown), confirming that the MV-like H5N1 viruses are closely related to the H5N1 viruses circulating in poultry in China's southern provinces, especially in waterfowl. Recently, reports by two research groups from the United States and China on flyways of some migratory birds at Qinghai Lake show only that wild birds migrate from Qinghai Lake to Mongolia, Eastern Asia, and the Bay of Bengal (32). The data do not show evidence of the migration of wild birds from Qinghai to China's southern provinces. Also, the published data show no evidence of H5N1 virus transmission by migratory birds to the pika, a nonmigratory small mammal living on the plateau, and poultry raised in different regions of China's southern provinces. Therefore, why the MV-like PK viruses have a close relation to the H5N1 virus circulating in southern China needs further investigation.
Our concern is whether H5N1 viruses isolated from pikas are pathogenic to mammals and poultry. As a model, we evaluated the pathogenicity of PK virus in mice and in pika-related rabbits. After experimental infection, the QH-like PK virus is fatal to mice while the MV-like PK viruses are hardly pathogenic. However, infected rabbits showed temperature elevations and mild or moderate interstitial pneumonia, similar to the lung damage of human H5N1 virus cases with a long disease duration (8, 30, 31), but suffered no clinical signs or death. Virus reisolation and immunofluorescence analysis revealed that the viral antigens were distributed mainly in turbinate bone of rabbits (Fig. (Fig.3B).3B). These results indicate that the MV-like PK virus establishes an infection in the respiratory tracts of rabbits but is not pathogenic. In view of the fact that no dead pikas were found in our field investigations, we believe that the PK virus is nonpathogenic or only subtly pathogenic to pikas. Notably, in our study, the virus was shed persistently from the noses of infected rabbits at 3 to 10 dpi. Therefore, extensive surveillance of pikas and other mammalian species infected by H5N1 avian influenza virus is imperative.
This work was supported by grants from the State Forestry Administration of the People's Republic of China, from the National Science Foundation of China (30625030), from the National Key Technology R & D Program of China (2004BA519A33, 2004BA519A63, 2006BAD06A03), from the National Modern Agro-Industry Technology Research System of China, and from the State Key Laboratory of Infectious Diseases for Diagnosis and Treatment.
We thank the Wildlife Department of the State Forestry Administration and the Monitoring General Station of Wildlife Infectious Diseases of the State Forestry Administration, People's Republic of China, for administrative help, and we also thank Yunsheng Hou of Qinghai Lake National Natural Reserve for animal sample collections.
Published ahead of print on 24 June 2009.