Influenza in wild and domestic birds

Based on pathogenicity in chickens, influenza A viruses are classified into two main pathotypes: highly pathogenic avian influenza (HPAI) and low pathogenic avian influenza (LPAI) (Alexander, Reference Alexander2007). Infection with LPAI (includes all subtypes) usually causes little or no disease, but it can sometimes result in high morbidity and low mortality. On the other hand, infection with HPAI viruses results in at least 75% mortality in the infected poultry and it is caused solely by some viruses of H5 and H7 subtypes (Swayne and Halverson, Reference Swayne and Halverson2007). The HA protein of LPAI is characterized by a single arginine (basic amino acid) at the cleavage site and another basic amino acid at position 3 or 4 upstream from the cleavage site (depending on the virus subtype). Thus, the HA protein of LPAI viruses is limited to cleavage by extracellular proteases (trypsin-like) that are secreted by cells or bacteria at the site of infection (e.g. trachea and intestine). On the other hand, HPAI viruses possess multiple basic amino acids at the HA protein cleavage site, making them prone to cleavage by intracellular ubiquitous proteases, such as furin (Stieneke-Grober et al., Reference Stieneke-Grober, Vey, Angliker, Shaw, Thomas, Roberts, Klenk and Garten1992; Vey et al., Reference Vey, Orlich, Adler, Klenk, Rott and Garten1992; Senne et al., Reference Senne, Panigrahy, Kawaoka, Pearson, Suss, Lipkind, Kida and Webster1996).

Wild birds, especially Anseriformes (ducks, geese and swans) and Charadriiformes (gulls, terns and surfbirds), are the natural reservoirs for influenza A viruses. The most recent report showed that these viruses have been isolated from at least 105 wild bird species representing 26 families (Olsen et al., Reference Olsen, Munster, Wallensten, Waldenstrom, Osterhaus and Fouchier2006). In an older study performed by Stallknecht and Shane (Reference Stallknecht and Shane1988), a virus isolation recovery rate of 10.9% (2317 of 21,318 sample) was observed in samples collected from wild birds. A 15.2% recovery rate was reported for Anseriformes, 2.9% for Passeriformes and 2.2% for Charadriiformes.

LPAI viruses of all subtypes have been isolated from wild birds in which they cause little or no disease. Few HPAI viruses have been isolated from wild birds, represented lately by HPAI H5N1 in Asia, Europe and Africa (Swayne and Halverson, Reference Swayne and Halverson2007). Sharp et al. (Reference Sharp, Kawaoka, Wright, Turner, Hinshaw and Webster1993) suggested that the waterfowls do not act as a reservoir for all avian influenza viruses and part of the influenza gene pool is maintained in shorebirds and gulls.

Infection of poultry with influenza A viruses has always been associated with the transmission of the viruses from wild birds (Alexander, Reference Alexander2007; Swayne and Halverson, Reference Swayne and Halverson2007). Poultry infection with influenza A virus was first reported in Italy in 1878. The disease was first diagnosed as fowl cholera and the causative agent was not identified as influenza A virus until 1955 (Swayne and Halverson, Reference Swayne and Halverson2007). Milder diseases (primarily respiratory disease) are caused by LPAI viruses and were first reported in the middle of the last century. The oldest existing LPAI virus (H10N7) was isolated in Germany from chickens in 1960 (Swayne and Halverson, Reference Swayne and Halverson2007).

Chickens and turkeys are not considered natural reservoirs of influenza viruses; however, due to alterations in the natural ecosystems by humans (ways of raising and marketing poultry), new niches have been created for the occurrence and distribution of influenza infections amongst poultry (Swayne and Halverson, Reference Swayne and Halverson2007).

Earlier and recent findings suggest that turkeys are more susceptible to influenza infections than chickens (Alexander, Reference Alexander2000; Pillai et al., Reference Pillai, Suarez, Pantin-Jackwood and Lee2009). For example, the H7N2 LPAI virus isolated from Virginia in 2002 was more infectious in turkeys compared to chickens under laboratory conditions and was transmitted more easily amongst turkeys which also required lower doses to be infected (Swayne and Halverson, Reference Swayne and Halverson2007). Influenza was reported in turkeys for the first time in 1963 in the USA (Alexander, Reference Alexander2000). Since then, various subtypes have been isolated from turkeys in many states within the USA. The isolation of these viruses was mostly reported from states that are heavily populated with turkeys and situated on migratory waterfowl flyways, such as California and Minnesota. In these states, outbreaks occurred every year since 1966, causing tremendous economic losses (Halverson et al., Reference Halverson, Frame, Friendshuh and Shaw1998). In 1995, two major outbreaks were reported in turkeys in the USA. The first in Utah (H7N3) resulted in 40% mortality in young birds and was associated with co-infection with Escherichia coli or Pasteurella multocida. The other outbreak occurred in Minnesota (H9N2) causing economic losses estimated in millions of dollars in Minnesota alone (Halverson et al., Reference Halverson, Frame, Friendshuh and Shaw1998). The elimination of outdoor raising of turkeys in the State of Minnesota in 1998 dramatically reduced the number of incidences of avian influenza.

Additionally, swine influenza viruses have been frequently isolated from turkeys in the USA since the early 1980s. The first report of swine influenza (H1N1) in turkeys was in 1981 (Mohan et al., Reference Mohan, Saif, Erickson, Gustafson and Easterday1981). In 2003, a new form of swine influenza virus emerged in turkeys and was recognized as a triple reassortant (TR) H3N2 virus (Choi et al., Reference Choi, Lee, Erickson, Goyal, Joo, Webster and Webby2004). These viruses were first isolated from pigs in 1998 and then from turkeys in 2003. Infections with these viruses elicit no clinical signs, but are associated with a drastic drop or complete cessation in egg production in turkey breeder hens. In our in vivo experimental system, different H3N2 TR strains with high genetic similarity behaved differently in terms of transmission between both species (Yassine et al., Reference Yassine, Al-Natour, Lee and Saif2007). While some viruses were transmitted one or both ways between swine and turkeys, other viruses were not transmitted either way between the two species.

Despite their higher populations compared to domestic turkeys, primary influenza outbreak incidences in chickens are fewer compared to turkeys (Alexander, Reference Alexander2000). Major outbreaks in chickens in North America, like the Pennsylvania outbreak in 1983 and the Mexican outbreak in 1995, were due to HPAI viruses. Both outbreaks were very similar in their progress, where H5N2 was isolated as an LPAI virus prior to the outbreak, and mutated to a HPAI virus causing high mortality in the infected flocks (Alexander, Reference Alexander2000). On the other hand, the H7 subtype viruses were responsible for multiple outbreaks in poultry in the USA and other countries around the globe. From 1870 until the mid-20th century, fowl plague (HPAI caused by H7 subtype strains) occurred in many regions including several parts of Europe, North Africa, the Middle East, Asia, North and South America. Major recent outbreaks were recorded in Pakistan (H7N3; 1995–2003), the Netherlands (H7N7, 2003), Canada (H7N3, 2004) and North Korea (H7N7, 2005) (Alexander, Reference Alexander2000; CIDRAP, Reference Wappes2009a; Pasick et al., Reference Pasick, Berhane, Hisanaga, Kehler, Cottom-Birt, Handel, Argue and Leighton2009). Chickens were the primary host for the disease in all of these outbreaks (Swayne, Reference Swayne2008).

Between 1965 and 1966, chicken, turkeys and other avian species were diseased with clinical signs indistinguishable from those caused by fowl plague (H7 HPAI). Interestingly, viruses from these birds were not inhibited by antisera collected from fowl-plague recovered birds. The disease was first referred to as ‘fowl plague-like’ and was then identified as H5 HPAI (Swayne and Halverson, Reference Swayne and Halverson2007; Swayne, Reference Swayne2008).

Since 1996, H5N1 has been the cause of the major ongoing pandemic of HPAI worldwide. This virus was first isolated from geese in 1996 in Guangdong Province, China (Swayne and Halverson, Reference Swayne and Halverson2007). The virus then spread to infect various poultry species, and a broad range of non-poultry animals including humans (Sims and Brown, Reference Sims and Brown2008). Between 2003 and 2004, outbreaks with H5N1 HPAI were reported almost concurrently in eight countries in South East Asia including China, Cambodia, Indonesia, Japan, Korea, Loa PDR, Thailand and Vietnam (Sims and Brown, Reference Sims and Brown2008). The virus then spread over large portions of Asia, some countries in Europe and Africa. So far, the virus has evolved (genetic drift/genetic shift) to form at least 20 clades/subclades in the phylogenetic tree (Sims and Brown, Reference Sims and Brown2008).

In addition to the above incidences of influenza in poultry, H9N2 LPAI has been a continuous problem in poultry since the mid-1990s worldwide. H9N2 viruses affected chickens, turkeys and domestic ducks in Germany, turkeys in the USA, ostriches in South Africa, pheasants in Ireland and chickens and quails in Italy, South East Asia and the Middle East (Alexander, Reference Alexander2007). Phylogenetic analysis of Asian-H9 subtype influenza isolated from poultry indicated wild bird-origin of the virus (Perez et al., Reference Perez, Lim, Seiler, Yi, Peiris, Shortridge and Webster2003). In Asia, H9N2 viruses have caused multiple outbreaks and established lineages in land-based poultry (Hossain et al., Reference Hossain, Hickman and Perez2008).

Influenza in mammals

In contrast to birds and most mammalian species, humans are prone to infection with all three types of influenza viruses, A, B and C (95). Type C viruses are less common in human populations and cause milder disease compared to types A and B. Infections with type A and type B viruses result in mild to acute disease in human populations. Up to 142,000 hospitalizations are recorded in the USA alone every year as a result of influenza infection, with a mortality rate of ~1% (20,000–36,000) of all deaths in the USA (Thompson et al., Reference Thompson, Shay, Weintraub, Brammer, Cox, Anderson and Fukuda2003, Reference Thompson, Shay, Weintraub, Brammer, Bridges, Cox and Fukuda2004; Subbaroa et al., Reference Subbaroa, Swayne and Olsen2006).

Human influenza A viruses were first isolated in 1933. Historically, only viruses of H1, H2, H3, N1 and N2 subtypes have circulated widely in human populations (Subbaroa et al., Reference Subbaroa, Swayne and Olsen2006). Three main pandemics occurred in the last century in humans, due to influenza A virus infections: Spanish influenza (1918; H1N1), Asian influenza (H2N2; 1957) and Hong Kong influenza (H3N2; 1968). The Spanish influenza was the most severe of all, resulting in deaths of 40–50 million people worldwide. Bacterial infections in the lungs were a prominent trait of fatal cases in the 1918 pandemic. Genetic analysis of H1N1 Spanish influenza viruses revealed avian-like genes with mutations for human adaptation (Taubenberger and Palese, Reference Taubenberger and Palese2006). Pandemics caused by H2N2 (1957) and H3N2 (1968) were less severe involving four million and one million deaths, respectively (Kilbourne, Reference Kilbourne2006). Genetic characterization revealed that both strains arose from reassortments between avian viruses and co-circulating human viruses. The H2N2 virus harbored HA, NA and PB1 of avian influenza and the rest of the genes were of circulating human H1N1 virus. On the other hand, the H3N2 virus harbored the HA and PB1 genes from avian influenza with the remaining genes from circulating H2N2 viruses (Subbaroa et al., Reference Subbaroa, Swayne and Olsen2006).

For the last 15 years, the H5N1 HPAI virus represented one of the highest threats to human health worldwide. The first human disease caused by H5N1 was reported from Hong Kong in 1997, involving 18 cases, including 6 deaths (Peiris et al., Reference Peiris, de Jong and Guan2007). At the time of the preparation of this manuscript, at least 476 cases from 15 countries have been reported with H5N1 infection, resulting in the death of 283 people (WHO, 2009a). Seroepidemiological study performed after the 1997 outbreak in Hong Kong showed that some human infections with H5N1 could be asymptomatic or mildly symptomatic (Peiris et al., Reference Peiris, de Jong and Guan2007). Moreover, avian H7 and H9 subtype viruses have been transmitted to humans in Asia, Europe and North America. After the major outbreak of H7N7 in poultry in the Netherlands in 2003, 89 human cases were laboratory confirmed with the same subtype virus. Most of the infected humans developed conjunctivitis, few developed influenza-like illnesses and one veterinarian died from acute respiratory distress syndrome (ARDS) due to the infection (Koopmans et al., Reference Koopmans, Wilbrink, Conyn, Natrop, van der Nat, Vennema, Meijer, van Steenbergen, Fouchier, Osterhaus and Bosman2004). Two people were also infected with H7N3 subtype virus after an outbreak in Canada in 2004, manifesting self-limited conjunctivitis and influenza illnesses (Tweed et al., Reference Tweed, Skowronski, David, Larder, Petric, Lees, Li, Katz, Krajden, Tellier, Halpert, Hirst, Astell, Lawrence and Mak2004; Belser et al., Reference Belser, Wadford, Xu, Katz and Tumpey2009). Most of the above infections with the three subtypes (H5, H7 and H9) were due to direct contact of humans with infected poultry. Human to human transmission was suggested for three individuals in the Netherlands outbreak (Peiris et al., Reference Peiris, de Jong and Guan2007).

In late April 2009, WHO announced the emergence of a novel influenza in humans (WHO, 2009b). Considering the highly contagious nature of H1N1 subtype influenza virus and the wide geographical prevalence, the first pandemic of the 21st Century was declared on 11 June 2009. As of January 2010, more than 208 countries/territories have been confirmed with pandemic H1N1 (pH1N1) influenza cases in humans (WHO, 2010a). In the USA alone, the Center for Disease Control and Prevention (CDC) estimated 39–80 million infected Americans along with 8–16 thousand deaths with pH1N1 virus, between April and December of 2009 (CNN, 2010). Interestingly, pH1N1 influenza accounted for more than 90% of the specimens tested positive for influenza A virus worldwide (WHO, 2010b). The new pH1N1 influenza virus contains a unique combination of gene segments not previously identified elsewhere. Although the origin of the new H1N1 virus is unclear, the new virus is genetically most closely related to TR swine influenza viruses currently circulating in the pig population. In particular, NA and M genes of the virus are closely related to those of Eurasian lineage of swine influenza viruses and the remaining six genes (HA, NP, NS, PA, PB1 and PB2) are similar to North American lineage viruses (Gibbs et al., Reference Gibbs, Armstrong and Downie2009).

While Spanish and other influenza pandemics caused a fatality rate of >2.5% and <0.1%, respectively, infection with the novel swine-origin virus was less severe with a fatality rate of <0.05% (Taubenberger and Morens, Reference Taubenberger and Morens2006; CIDRAP, Reference Robert2009b; Donaldson et al., Reference Donaldson, Rutter, Ellis, Greaves, Mytton, Pebody and Yardley2009). Analyzing the household transmission of pH1N1 virus in the USA, it was shown that the transmissibility of the 2009 H1N1 influenza virus in households is lower than that seen in past pandemics. It was also shown that most transmissions occur soon before or after the onset of symptoms in a case patient (Cauchemez et al., Reference Cauchemez, Donnelly, Reed, Ghani, Fraser, Kent, Finelli and Ferguson2009).

pH1N1 virus has infected species other than humans including pigs, turkeys, cats and dogs (Swayne et al., Reference Swayne, Pantin-Jackwood, Kapczynski, Spackman and Suarez2009; Ariel Pereda et al., Reference Ariel Pereda, Quiroga, Baumeister, Insarralde, Ibar, Sanguinetti, Cannilla, Franzese, Escobar Cabrera, Craig, Rimondi, Machuca, Debenedetti, Zenobi, Barral, Balzano, Capalbo, Risso and Perfumo2010; AVMA, 2010; Sponseller et al., Reference Sponseller, Strait, Jergens, Trujillo, Harmon, Koster, Jenkins-Moore, Killian, Swenson, Bender, Waller, Miles, Pearce, Yoon and Nara2010). Experimentally, pH1N1 replicates similarly compared to other swine influenza viruses in pigs. On the contrary, turkeys were resistant to a natural route of infection, although increased susceptibility was demonstrated using the intrauterine inoculation route.

Pigs are susceptible to infection with both human and avian influenza viruses and have been hypothesized to serve as the intermediate host for interspecies transmission of these viruses. H1 and H3 subtypes are the most common influenza subtypes in pigs worldwide, but sporadic infections with other subtype viruses have been recorded (Ninomiya et al., Reference Ninomiya, Takada, Okazaki, Shortridge and Kida2002; Ma et al., Reference Ma, Vincent, Gramer, Brockwell, Lager, Janke, Gauger, Patnayak, Webby and Richt2007; Yu et al., Reference Yu, Hua, Wei, Zhou, Tian, Li, Liu and Tong2008). Influenza was initially recognized in pigs in 1918, in a temporal and spatial coincidence with Spanish flu in humans (Subbaroa et al., Reference Subbaroa, Swayne and Olsen2006). The first isolation of the virus (H1N1) was in 1930 and it was later recognized as classical H1N1 lineage of swine viruses (cH1N1). Recent research has indicated that swine and human viruses of 1918 were closely related to each other, both genetically and antigenically (Reid et al., Reference Reid, Taubenberger and Fanning2001; Kilbourne, Reference Kilbourne2006). cH1N1 swine viruses in the USA remained genetically and antigenically stable until the 1990s, when variants of cH1N1 viruses were isolated (Vincent et al., Reference Vincent, Lager, Ma, Lekcharoensuk, Gramer, Loiacono and Richt2006; Wright et al., Reference Wright, Neumann and Kawaoka2007). In Europe, classical H1N1 viruses disappeared in the late 1970s with the introduction of avian-like viruses. Avian-like viruses reassorted with human viruses in pigs resulting in new viruses harboring human-like HA and NA genes and avian-like internal genes (Campitelli et al., Reference Campitelli, Donatelli, Foni, Castrucci, Fabiani, Kawaoka, Krauss and Webster1997; Dunham et al., Reference Dunham, Dugan, Kaser, Perkins, Brown, Holmes and Taubenberger2009).

Reassortants between swine and human viruses have been reported in more than one incident (e.g. Japan (H1N2, 1978) and Europe (H1N2, 1987)) (Wright et al., Reference Wright, Neumann and Kawaoka2007). H1N2 viruses are still widely circulating with H1N1 and H3N2 subtypes in pigs in Europe. In the USA, a new form of reassortant viruses (H3N2) emerged in swine populations in 1998. These viruses harbored genes from human (HA, NA and PB1), avian (PB2 and PA) and swine (NP, M and NS) lineage viruses (Webby et al., Reference Webby, Swenson, Krauss, Gerrish, Goyal and Webster2000). Interestingly, HA and PB1 genes share the same origin (human lineage), in a similar pattern to the 1957 and 1968 human pandemic viruses (Kawaoka et al., Reference Kawaoka, Krauss and Webster1989). Human virus genes in the H3N2 viruses were similar to those contained in human vaccine strains used in the mid-1990s (Olsen, Reference Olsen2002). Further reassortments occurred between H3N2 TR and cH1N1 viruses resulting in the emergence of H1N2 viruses. Since then, double and/or triple reassortants of H1N1, H1N2, H3N1 and H3N2 have been isolated from the swine population in the USA (Lekcharoensuk et al., Reference Lekcharoensuk, Lager, Vemulapalli, Woodruff, Vincent and Richt2006; Wright et al., Reference Wright, Neumann and Kawaoka2007; Yassine et al., Reference Yassine, Al-Natour, Lee and Saif2007). Interestingly, more than one of these viruses has been isolated from species other than swine, including humans and turkeys. H1N1 TR viruses have been isolated from humans in several locations in the USA including Ohio, Wisconsin and Texas (Olsen, Reference Olsen2002; Olsen et al., Reference Olsen, Brammer, Easterday, Arden, Belay, Baker and Cox2002; Gray et al., Reference Gray, McCarthy, Capuano, Setterquist, Olsen and Alavanja2007; Newman et al., Reference Newman, Reisdorf, Beinemann, Uyeki, Balish, Shu, Lindstrom, Achenbach, Smith and Davis2008; Schnirring, Reference Schnirring2008). An H3N2 TR virus was also isolated from a child in Canada in 2007 (Robinson et al., Reference Robinson, Lee, Patel, Bastien, Grimsrud, Seal, King, Marshall and Li2007).

Additionally, pigs have been infected with avian influenza viruses worldwide. Influenza viruses of H4 and H9 subtypes have been isolated from pigs in Canada and China, respectively (Karasin et al., Reference Karasin, Olsen, Brown, Carman, Stalker and Josephson2000; Cong et al., Reference Cong, Pu, Liu, Wang, Zhang, Zhang, Fan, Brown and Liu2007). Although H5N1 influenza viruses were isolated from pigs (Shi et al., Reference Shi, Gibbs, Zhang, Zhang, Zhao, Jin, Zhu, Yang, Yang, Cui and Ji2008; Zhu et al., Reference Zhu, Yang, Chen, Cao, Zhong, Jiao, Deng, Yu, Yang, Bu, Kawaoka and Chen2008), inoculation and feeding domestic pigs with H5N1 viruses from different origins (human, chicken, duck and whooper swan) or virus contaminated feed did not result in a severe disease. Virus replication was restricted to the respiratory tract and tonsils of pigs, which is in contrast to mouse and ferret where some of the viruses were highly pathogenic and replicated systemically. Additionally, the virus replication level was remarkably low compared to the replication of swine influenza viruses (Lipatov et al., Reference Lipatov, Kwon, Sarmento, Lager, Spackman, Suarez and Swayne2008).

As with human and swine diseases, equine influenza has been a problem for many years. Two main subtypes have been recognized in horses, equine-1 (H7N7) and equine-2 (H3N8) (Timoney, Reference Timoney1996). The disease is usually more associated with H3N8 viruses, with symptoms similar to those observed in human and swine influenza infections, including fever, coughing, loss of appetite and muscular soreness (Wright et al., Reference Wright, Neumann and Kawaoka2007). H7N7 was first isolated in 1956 and the last confirmed outbreak was reported in 1979. On the other hand, H3N8 was first isolated in 1963 and is still widespread in horse populations. It caused major outbreaks in the USA (1963), China (1989) and widespread epidemics in Europe and around the world (Timoney, Reference Timoney1996; Wright et al., Reference Wright, Neumann and Kawaoka2007).

Dogs and other canine species can be infected with influenza A viruses. Equine H3N8 viruses were isolated from dogs (greyhounds) with respiratory disease in Florida, suggesting interspecies transmission of these viruses (Crawford et al., Reference Crawford, Dubovi, Castleman, Stephenson, Gibbs, Chen, Smith, Hill, Ferro, Pompey, Bright, Medina, Johnson, Olsen, Cox, Klimov, Katz and Donis2005). Serologic studies indicated that H3N8 virus has spread to the general dog population (Crawford et al., Reference Crawford, Dubovi, Castleman, Stephenson, Gibbs, Chen, Smith, Hill, Ferro, Pompey, Bright, Medina, Johnson, Olsen, Cox, Klimov, Katz and Donis2005). Avian H3N2 influenza viruses were also isolated from dogs with severe respiratory disease in Korea, in 2007 (Song et al., Reference Song, Kang, Lee, Jung, Ha, Kang, Park, Park and Oh2008).

Cats (domestic cats, tigers and Owston civets) were shown to be infected with influenza A virus. H5N1 virus was isolated from all three species and probable virus transmission was reported amongst tigers (Thanawongnuwech et al., Reference Thanawongnuwech, Amonsin, Tantilertcharoen, Damrongwatanapokin, Theamboonlers, Payungporn, Nanthapornphiphat, Ratanamungklanon, Tunak, Songserm, Vivatthanavanich, Lekdumrongsak, Kesdangsakonwut, Tunhikorn and Poovorawan2005). Pandemic H1N1 viruses were also isolated from dogs and cats with influenza-like illness in the USA and around the globe (oregonvma, 2010). In the USA, death was associated with some cat infections with the pandemic H1N1 influenza A virus.

Minks are also prone to infection with influenza viruses, naturally and experimentally. Both human and avian viruses can be transmitted to and infect minks. In 1985, a mink died in Sweden due to an infection with the avian-like H10N4 subtype. In addition to avian viruses, minks were also experimentally infected with human, swine and equine viruses and the viruses were transmitted to cage mates (Wright et al., Reference Wright, Neumann and Kawaoka2007).

Finally, influenza viruses were isolated from marine mammals including seals and whales and were in most cases avian-like viruses, indicating a possible transmission from sea-shore birds and water fowls (Hinshaw et al., Reference Hinshaw, Bean, Geraci, Fiorelli, Early and Webster1986; Nielsen et al., Reference Nielsen, Clavijo and Boughen2001).

The role of surface glycoproteins

Influenza host-range restriction is a multigenic trait, where most of the genes play a role in influenza virus adaptation to a specific host. The availability of a large number of virus sequences and the utilization of reverse genetics, site-directed mutagenesis (SDM) and other molecular techniques permitted the identification at the molecular level of the determinants of host-range specificity of influenza viruses.

The HA protein is the major surface glycoprotein that mediates two important functions in virus replication: binding to the appropriate receptor on the host cell and fusion between viral and endosomal membranes. Four hundred to 500 trimeric HA spikes are found on each virus which enables simultaneous interaction of RBD with multiple copies of cell-surface receptors. The RBD that contains the SA binding residues is a shallow depression on the protein surface located on the membrane's distal end of the globular head (Matrosovich et al., Reference Matrosovich, Klenk and Kawaoka2006b). Some residues at the RBD are rather conserved because of their essentiality in virus interaction with the receptor and others are variable within different subtypes and different lineage viruses. Residues 98, 131–134, 136–138, 153–155, 183–190, 194–195, 218–220, 224 and 226–228 have been postulated to be key amino acids that constitute or comprise the RBD. However, the exact location of the RBD depend on the individual HA subtype or specific strain. Amino acids at 134, 136 and 153 are preserved throughout the evolution of influenza viruses in different hosts. On the other hand, those at positions 190, 225, 226 and 228 are more variable and more prone to mutations. While human viruses usually have Asp190 (H1N1 and H3N2), Asp225 (H1N1), Leu/Ile/Val226 and Ser228 (H3N2 and H2N2), avian viruses harbor Glu190, Gly225, Gln226 and Gly228 at these positions (Matrosovich et al., Reference Matrosovich, Klenk and Kawaoka2006b). In addition, avian viruses have conserved amino acids Ala138 and Leu194. Mutations at these positions could affect the binding of the avian viruses to their corresponding receptors. Interestingly, an H7N7 virus was isolated from domestic chickens, with a deletion in residues 221–228 of the HA molecule. Moreover, a laboratory generated mutant with deletion of residues 224–230 was still able to bind receptors and grow in embryonated chicken eggs (ECE) (Daniels et al., Reference Daniels, Jeffries, Yates, Schild, Rogers, Paulson, Wharton, Douglas, Skehel and Wiley1987; Suarez et al., Reference Suarez, Garcia, Latimer, Senne and Perdue1999). This indicates a less critical role of residues at this site (220–230) for receptor binding, at least for some strains.

Generally, avian viruses are thought to preferentially bind to the N-acetylneuraminic acid α2,3-galactose (NeuAcα2,3Gal) form of SA receptors and human viruses preferentially bind to NeuAcα2,6Gal SA receptors. Hinshaw et al. (Reference Hinshaw, Webster, Naeve and Murphy1983) were the first group to show the contribution of HA in determining host range specificity. They demonstrated that a reassortant between human (H3N2) and avian (H2N2) viruses, with HA being from human strain and the rest of the genes from the duck strain, failed to replicate in ducks. The introduction of double mutations (Leu226Gln and Ser228Gly) into the HA protein resulted in the replication of the above reassortant strain in duck intestine. A single Leu226Gln mutation was not sufficient to initiate replication in duck intestine. The Gln226 was found to be responsible for the binding to α2,3-linked SA and Gly228 enhanced the binding affinity (Naeve et al., Reference Naeve, Hinshaw and Webster1984; Vines et al., Reference Vines, Wells, Matrosovich, Castrucci, Ito and Kawaoka1998; Matrosovich et al., Reference Matrosovich, Tuzikov, Bovin, Gambaryan, Klimov, Castrucci, Donatelli and Kawaoka2000). Residues 228 and 229 were shown to be highly conserved in ducks and to a lesser extent residues 252 and 223 (Kawaoka et al., Reference Kawaoka, Chambers, Sladen and Webster1988; Matrosovich et al., Reference Matrosovich, Klenk and Kawaoka2006a).

Variations in receptor binding specificity were also observed within avian species themselves. Although gull viruses have higher binding affinity to α2,3-linked SA, few gull viruses were shown to infect and replicate in ducks. Interestingly, H13 subtype, which is the most prevalent subtype in gulls, has never been isolated from ducks (Sivanandan et al., Reference Sivanandan, Halvorson, Laudert, Senne and Kumar1991; Laudert et al., Reference Laudert, Sivanandan, Halvorson, Shaw and Webster1993).

Various subtypes isolated from chickens showed variability in receptor binding specificity. An H8N4 isolated from chicken in Ontario (1968) had duck-like receptor-binding specificity, while an H9N2 isolated from chicken in Wisconsin (1966) had human-like receptor specificity (Nobusawa et al., Reference Nobusawa, Aoyama, Kato, Suzuki, Tateno and Nakajima1991). The latter subtype (H9N2) has spread in many countries in Asia, Europe, Middle East and Africa since the early 1990s, affecting poultry production in all of these countries. Similar viruses were isolated from human and pigs in South East Asia, raising concerns about interspecies transmission of these viruses (Peiris et al., Reference Peiris, Yuen, Leung, Chan, Ip, Lai, Orr and Shortridge1999; Lu et al., Reference Lu, Lu, Chen, Jiang, Tan, Ling, Zheng and Sui2008; Yu et al., Reference Yu, Hua, Wei, Zhou, Tian, Li, Liu and Tong2008). Avian H9N2 viruses isolated from Hong Kong were shown to carry a human-like amino acid at residue 226 of the RBD (Leu instead of Gln) plus additional mutations at residues 183, 190 and 225 that are supposed to be conserved in avian viruses. These viruses showed an ability to bind human-like receptors (α2,6-linked SA) that also exist in chicken tracheal epithelial cells (Lin et al., Reference Lin, Shaw, Gregory, Cameron, Lim, Klimov, Subbarao, Guan, Krauss, Shortridge, Webster, Cox and Hay2000; Matrosovich et al., Reference Matrosovich, Krauss and Webster2001; Ha et al., Reference Ha, Stevens, Skehel and Wiley2002). Viruses isolated from pigs in China in 2006 also harbored Leu226, indicating a significant role played by this residue for the adaptation of avian H9N2 viruses in mammalian species (Guo et al., Reference Guo, Wen, Wang, Li, Zhang and Guo2004). In a recent study, using ferrets as a transmission model (Sorrell et al., Reference Sorrell, Wan, Araya, Song and Perez2009), two mutations (at residues 226 and 198) occurred in the HA protein upon the adaptation of avian H9N2 in ferrets. Both mutations were shown to be required for transmission of the virus amongst ferrets. The same group (Wan and Perez, Reference Wan and Perez2007) has previously shown that Leu226 on the RBD determines cell tropism and replication of H9N2 in human airway epithelial cells.

The transmission of avian viruses to humans is not uncommon. Researchers speculate that an avian H1N1 virus with human adapted mutations was the cause of the Spanish influenza in 1918. Since 1996, the HPAI H5N1 virus has been isolated from humans in several countries over the globe. Few mutations at the RBD in both subtypes were shown to alter the receptor-binding specificity between α2,3 and α2,6-linked SA receptors. For the H1N1 (1918) viruses, Asp190 and Asp225 were important for binding of H1N1 (1918) virus to α2,6-linked SA receptors and also for the transmission of these viruses in the ferret model. A single mutation (Asp190Glu) decreased the ability of the virus to bind α2,6-linked SA receptors, while another single mutation (Asp225Gly) increased the ability of the virus to bind α2,3-linked SA receptors. Double mutations at the above two positions dramatically decreased the virus ability for binding human-like receptors, and affected transmission but not replication in the ferret model. Interestingly, swine H1N1 isolates from the 1930s expressed Asp190 at the RBD, emphasizing the role of Asp at this position for replication in mammals (Gamblin et al., Reference Gamblin, Haire, Russell, Stevens, Xiao, Ha, Vasisht, Steinhauer, Daniels, Elliot, Wiley and Skehel2004).

Since 1996, researchers have been working to understand the mechanisms of the transmission of HPAI H5N1 from poultry to humans. Although the virus replicates efficiently in the infected humans, often leading to death, the virus has not yet adapted for human to human transmission. In 2003, H5N1 viruses were isolated from father and son in Hong Kong where the viruses had a replacement at residue 227 of the RBD. Such a mutation was shown in the same study to alter the receptor-binding specificity, where the human virus was able to bind both α2,3 and α2,6-liked SA receptors (Shinya et al., Reference Shinya, Hatta, Yamada, Takada, Watanabe, Halfmann, Horimoto, Neumann, Kim, Lim, Guan, Peiris, Kiso, Suzuki, Suzuki and Kawaoka2005). A similar mutation (Ser227Asn) enabled an isolate from Vietnam-2004 to bind both α2,3 and α2,6-linked SA receptors (Stevens et al., Reference Stevens, Blixt, Tumpey, Taubenberger, Paulson and Wilson2006).

In a study by Yamada et al. (Reference Yamada, Suzuki, Suzuki, Le, Nidom, Sakai-Tagawa, Muramoto, Ito, Kiso, Horimoto, Shinya, Sawada, Kiso, Usui, Murata, Lin, Hay, Haire, Stevens, Russell, Gamblin, Skehel and Kawaoka2006), various avian and human isolates (H5N1) were analyzed for sequence similarity. Mutations were introduced based on the similarities/differences between isolates and receptor-binding specificity to either α2,3 and α2,6-liked SA receptors was evaluated. In vitro analysis revealed that when similar mutations were applied to viruses from different clades (I and II), variable results were obtained. Single mutations as Gln192Arg, Gly139Arg, Asn182Lys, or Asn193Lys, enabled a virus from clade I to switch binding specificity to α2,6-linked SA receptors. Double and triple mutants in residues 75, 123 and 167 (with or without Asn193Lys) were required to enhance binding to 2,6-linked SA receptors. On the other hand, only Gln192Arg and Asn193Lys mutations enabled a virus from clade II to support binding to α2,6-linked SA receptors. Other mutations, Gly139Arg and Asn182Lys, were not sufficient to enable binding to human receptors and abolished binding to α2,3-linked SA receptors. The specific structure/geometry of the HA molecule and the proximity/type of amino acid residues at the RBD, play a role in which mutations might be suitable for one strain, but not the other, in determining receptor-binding specificity.

In summary, different HA subtypes require various mutations at the RBD to switch binding specificity between α2,3 and α2,6-linked SA receptors. For H2 and H3, Gln226Leu and Gly228Ser are required to switch from avian- to human-binding specificity. In H1, Glu190Asp is critical for adaptation to human-like receptor binding. In H5, contradictory findings were recorded depending on the type of strain used in each study; however, mutations close to those in H3 might play a role in elevating the binding ability to α2,6-linked SA without abolishing the binding ability to avian-like receptors.

Receptor affinity can also be modulated by glycosylation and sialylation of the HA head domain. Human H1 viruses show hyperglycosylation compared to viruses from aquatic birds and pigs (Inkster et al., Reference Inkster, Hinshaw and Schulze1993). Hyperglycosylation of the HA head was associated with increased virulence of H7 subtype viruses in chickens in Australia and Italy. It has been suggested that hyperglycosylation of HA, combined with compensating NA stalk deletion, modifies the progenitor aquatic bird virus prior to the development of virulence during virus evolution in chickens (Perdue et al., Reference Perdue, Latimer and Crawford1995; Matrosovich et al., Reference Matrosovich, Zhou, Kawaoka and Webster1999; Banks et al., Reference Banks, Speidel, McCauley and Alexander2000, Reference Banks, Speidel, Moore, Plowright, Piccirillo, Capua, Cordioli, Fioretti and Alexander2001; Baigent and McCauley, Reference Baigent and McCauley2003).

The role played by the NA protein in determining host-range specificity is minor compared to that of the HA molecule. However, functional compatibility of receptor-binding properties with NA cleavage specificity, along with the stalk length of the NA protein, is critical. As is the case with the HA protein, the NA protein can specifically cleave SA in the form of α2,3 and/or α2,6 linkage. The mechanism of this specificity is not well characterized, but in one study it was shown that the specificity of the NA (N8) is associated with the amino acid at position 275, as well as glycosylation at residue 144 (Saito and Kawano, Reference Saito and Kawano1997). As another form of compatibility between HA and NA proteins, a reduction in NA specific activity is required in human viruses to maintain optimal balance with the HA protein that shows less affinity to its receptor compared to avian viruses (Baigent and McCauley, Reference Baigent and McCauley2003).

Changes in the NA protein upon the transmission of influenza virus to a new host usually occurs to accommodate compatibility with the HA protein. For example, many of the NA stalk deletions have been recorded upon the transmission of wild bird viruses (especially duck) into chickens. This has been recorded for three major subtypes (H5, H7 and H9) that cause disease in chickens (Banks et al., Reference Banks, Speidel, Moore, Plowright, Piccirillo, Capua, Cordioli, Fioretti and Alexander2001; Sorrell and Perez, Reference Sorrell and Perez2007; Steensels et al., Reference Steensels, Van Borm, Boschmans and van den Berg2007; Hossain et al., Reference Hossain, Hickman and Perez2008). Types of gangliosides (with short and long sugar chains) that are distributed in the guts of chicken and ducks, and to which influenza viruses bind, may play a role in NA proteins acquiring stalk deletions when the virus is transmitted from ducks to chickens. NA stalk deletion was also observed in a swine H3N2 TR isolate from North Carolina in 2003 (Yassine et al., Reference Yassine, Lee, Suarez and Saif2008). A study by Luo et al. (Reference Luo, Chung and Palese1993) demonstrated that deletions up to 28 amino acids and insertions up to 41 amino acids of the NA stalk protein did not abolish the formation of progeny viruses in vivo. In a separate study (Mitnaul et al., Reference Mitnaul, Matrosovich, Castrucci, Tuzikov, Bovin, Kobasa and Kawaoka2000), a human mutant influenza virus with a 24-amino acid NA stalk deletion, was shown to replicate like the wild-type strain in MDCK cells but not in ECE. Upon passaging several times in ECE, the virus acquired competence in two mechanisms: some viruses acquired insertions in the NA stalk from PB1, PB2, or NP protein, in a rare RNA/RNA recombination process, while others expressed mutations in the HA protein, affecting its binding to SA-linked receptors. Both mechanisms of adaptation require interaction between the NA and HA proteins, emphasizing balanced HA/NA activities for an efficient replication of influenza A viruses. On the other hand, the treatment of infected individuals with NA inhibitors has resulted in multiple mutations in the NA protein (e.g. Glu119V, Arg292Lys and His274Tyr) (Herlocher et al., Reference Herlocher, Carr, Ives, Elias, Truscon, Roberts and Monto2002; Ives et al., Reference Ives, Carr, Mendel, Tai, Lambkin, Kelly, Oxford, Hayden and Roberts2002). Such mutations (especially Arg292Lys mutation) significantly compromised virus infectivity, pathogenicity and transmissibility (Herlocher et al., Reference Herlocher, Carr, Ives, Elias, Truscon, Roberts and Monto2002).

The role of ribonucleoproteins (RNP)

Contribution of RNPs in determining host-range specificity is maintained through three main mechanisms: (i) increasing genetic diversity from which variants with high adaptation and transmission potential are selected in the new host; (ii) proper interaction with host proteins, and thus generation of a suitable environment for efficient virus replication; and (iii) generation of escape mutants to avoid the innate immune system of the new host.

As is the case for the surface glycoproteins, adaptation through RNP occurs either directly in a new host, or in an intermediate host between the donor and recipient species (Naffakh et al., Reference Naffakh, Tomoiu, Rameix-Welti and van der Werf2008). Genetic analysis for the NP protein of the 1918 Spanish influenza revealed that its sequence is divergent from that of the wild bird viruses, indicating that the virus (H1N1) evolved for a long time in an intermediate host (distinct from wild birds) before emerging in the human population. In contrast, when a new PB1 gene was introduced (along with HA and NA genes) to the existing human virus genetic background by genetic reassortment in the 1957 (H2N2) and 1968 (H3N2) pandemics, its sequence was closer to that of avian viruses than the concurrent circulating human (H1N1) viruses (Naffakh et al., Reference Naffakh, Tomoiu, Rameix-Welti and van der Werf2008). Generally, the PB2, PA and NP proteins seem to share similar evolutionary pathways, suggesting co-evolution due to strong physical and functional interaction with each other (Naffakh et al., Reference Naffakh, Tomoiu, Rameix-Welti and van der Werf2008).

Out of 52 host-associated genetic signatures identified throughout the influenza viral genome (Chen et al., Reference Chen, Chang, Mok, Lo, Kung, Huang, Shih, Wang, Chiang, Chen and Shih2006), 35 are linked with the RNP (2, 8, 10 and 15 for PB1, PB2, PA and NP, respectively). The contribution of RNPs in determining the host range of influenza virus was first shown in 1977 (Almond, Reference Almond1977). Generating reassortant viruses using the traditional plaque assay method, it was revealed that the extended host range for one strain was conferred on the PB2 segment. Utilizing the same technique, reassortants were generated between human and avian viruses to identify genes and their amino acids' roles in replication of influenza viruses in different host systems. It was shown that the HA and NA genes alone (of human virus lineage) did not confer good replication of the virus in squirrel monkeys if the other six genes were of avian origin (Naffakh et al., Reference Naffakh, Tomoiu, Rameix-Welti and van der Werf2008). Later on, it was shown that the most significant genetic signature for host adaptation is located at residue 627 of PB2.

In human viruses, Lys is usually encoded at that residue, while most avian viruses have Glu at this position. Although the PB2 was of avian origin in the 1918 Spanish influenza virus, Lys was recognized at the PB2-627 residue (Taubenberger et al., Reference Taubenberger, Reid, Lourens, Wang, Jin and Fanning2005). Such substitution was also observed in many of the human H5N1 isolates, where 23 out of 43 analyzed isolates had this substitution. Interestingly, Lys was also observed in H5N1 isolates from tigers in Thailand, but not in raccoon dog isolates from China which maintained avian-like Glu at this position (Maines et al., Reference Maines, Lu, Erb, Edwards, Guarner, Greer, Nguyen, Szretter, Chen, Thawatsupha, Chittaganpitch, Waicharoen, Nguyen, Nguyen, Nguyen, Kim, Hoang, Kang, Phuong, Lim, Zaki, Donis, Cox, Katz and Tumpey2005; Amonsin et al., Reference Amonsin, Payungporn, Theamboonlers, Thanawongnuwech, Suradhat, Pariyothorn, Tantilertcharoen, Damrongwantanapokin, Buranathai, Chaisingh, Songserm and Poovorawan2006; Naffakh et al., Reference Naffakh, Tomoiu, Rameix-Welti and van der Werf2008; Le et al., Reference Le, Sakai-Tagawa, Ozawa, Ito and Kawaoka2009; Qi et al., Reference Qi, Li, Rider, Fan, Gu, Xu, Yang, Lu, Wang and Liu2009). Moreover, during the H7N7 virus outbreak in the Netherlands in 2003, the same Glu627Lys substitution was found in the virus isolated from humans with fatal pneumonia, but not with cases of conjunctivitis in the same outbreak (Fouchier et al., Reference Fouchier, Schneeberger, Rozendaal, Broekman, Kemink, Munster, Kuiken, Rimmelzwaan, Schutten, Van Doornum, Koch, Bosman, Koopmans and Osterhaus2004). On the other hand, swine and equine isolates of avian-like lineages retained Glu at this position, suggesting less selective pressure for this residue in both species (Shinya et al., Reference Shinya, Watanabe, Ito, Kasai and Kawaoka2007). The 2009 pandemic H1N1, which is thought to have originated from the swine influenza virus, contains avian-like residue at this position (Chen and Shih, Reference Chen and Shih2009). In support of these findings, Glu627Lys substitution in PB2 gene of swine-origin H3N2 TR virus did not affect its replication in primary epithelial cells generated from pig and human tracheas (unpublished observations).

When PB2 from a duck strain (A/Mallard/78) was introduced to a human strain (A/Los Angeles/78), the reassortant had restricted growth in MDCK cells and squirrel monkeys. Revertants from the reassorted strain that were able to replicate in both systems showed a single substitution (Glu627Lys) that was responsible for the extended host range phenotype (Subbarao et al., Reference Subbarao, London and Murphy1993). Using the ferret model to study influenza transmission, Lys at position 627 of PB2 protein was necessary and sufficient for respiratory droplet transmission, but not required for direct contact transmission, of 1918 H1N1 virus amongst ferrets (Van Hoeven et al., Reference Van Hoeven, Pappas, Belser, Maines, Zeng, Garcia-Sastre, Sasisekharan, Katz and Tumpey2009). Studying the pathogenesis of H5N1 (A/Vietnam/1203/04) and H7N7 (A/Netherlands/219/03) in a mouse model, it was shown that the PB2 gene alone, with or without other genes, contributed to virulence of the above two strains in mice. In both cases, substitution of Glu with that of Lys determined the outcome of infection in mice. However, the presence of either Glu or Lys at PB2-627 does not strictly correlate with the severity of human infection with H5N1 (Gao et al., Reference Gao, Watanabe, Ito, Goto, Wells, McGregor, Cooley and Kawaoka1999).

In addition to its role in virulence and host range adaptation, PB2-627 plays a role in tissue tropism of the virus. Lys at position 627 of the PB2 protein enabled H5N1 virus to replicate in nasal turbinates and lungs of infected mice, whereas viruses with Glu at this position replicated poorly in nasal turbinates.

Human viruses with Lys at PB2-627 replicate well at 33°C, while avian viruses with the same substitution replicate poorly at the above temperature. Taking into account that avian viruses replicate in the intestinal tract of birds with a body temperature of 41°C, cold sensitivity of the polymerase proteins might limit the growth of avian viruses in the upper respiratory tract of humans and thus limit human-to-human transmission (Naffakh et al., Reference Naffakh, Tomoiu, Rameix-Welti and van der Werf2008).

Another remarkable residue to support the adaptation of avian influenza virus in humans is amino acid 375 of PB1. While Asn exists at this position in most avian species, most human viruses have Ser at this position. This substitution was the only conserved mutation observed in the PB1 gene (avian origin) in viruses from three pandemics in the last century (Kawaoka et al., Reference Kawaoka, Krauss and Webster1989; Taubenberger et al., Reference Taubenberger, Reid, Lourens, Wang, Jin and Fanning2005). Additionally, a Gly375Ser modification was observed upon introduction of swine viruses to human (Kawaoka et al., Reference Kawaoka, Krauss and Webster1989). Nevertheless, PB1-375 cannot be classified as a genetic signature, where some human H3N2 viruses have Asn at this position, and Ser-375 is found in a considerable number of avian isolates (Taubenberger et al., Reference Taubenberger, Reid, Lourens, Wang, Jin and Fanning2005).

In addition to those specific amino acid residues mentioned above, PB2-701, PB2-714, PA-97, PA-615_, PA-624 and NP-319 have been associated with high replication and pathogenicity of a H7N7 (A/Seal/Massachussets/1/80) and many H5N1 isolates in mice and humans. The role played by these residues in host-range determinacy is more significant in combination with Glu627lys substitution (Naffakh et al., Reference Naffakh, Tomoiu, Rameix-Welti and van der Werf2008). In a recent study by Steel et al. (Reference Steel, Lowen, Mubareka and Palese2009), it was shown that PB2-701Asn could compensate for the reduced transmission of two human viruses (H3N2 and H5N1) harboring avian-like PB2-627(Glu) in the guinea pig model. In support of this study, Gao et al. (Reference Gao, Zhang, Shinya, Deng, Jiang, Li, Guan, Tian, Li, Shi, Liu, Zeng, Bu, Xia, Kawaoka and Chen2009) showed in a more recent publication that Asn at position 701 in the PB2 protein was a prerequisite but not sufficient for duck strain transmission in guinea pigs. Another mutation (Thr160Ala) in the HA molecule was also required to sustain the avian virus transmission in a guinea pig model. This mutation in the HA protein, which results in loss of glycosylation at 158–160 position, was responsible for HA binding to sialylated glycans and was critical for H5N1 virus transmission in guinea pigs (Gao et al., Reference Gao, Zhang, Shinya, Deng, Jiang, Li, Guan, Tian, Li, Shi, Liu, Zeng, Bu, Xia, Kawaoka and Chen2009).

Although no specific role has been assigned to PA protein in the RNP complex, several studies have demonstrated its potential role in influenza virus adaptation to a new host. Serial passage of an LPAI H5N1 strain in mice resulted in non-conserved mutations in six viral genes. Of these, Thr to Iso substitution at position 97 of the PA protein played a key role in determining the enhanced virulence of the adapted strain in mice. Such mutation was also shown to enhance polymerase activity, and thus virus replication, in mammalian but not avian cells (Song et al., Reference Song, Pascua, Lee, Baek, Lee, Kim, Kim, Webby, Webster and Choi2009).

The molecular mechanisms involved in the above actions are still not fully understood; however, interactions amongst viral and host proteins play a critical role in viral adaptation and pathogenesis in the different host systems. For example, the crystal structure of the C-terminus domain of PB2 protein complexed with human importin α5 showed how the last 20 residues unfold to permit binding to the importin factor. The domain contains three surface residues (including Asp701) implicated in adaptation from avian to mammalian hosts (Tarendeau et al., Reference Tarendeau, Boudet, Guilligay, Mas, Bougault, Boulo, Baudin, Ruigrok, Daigle, Ellenberg, Cusack, Simorre and Hart2007). In a more recent study, adaptive mutations in PB2 (Asp701Asn) and in NP (Asn319Lys) enhanced the binding of these proteins to importin α1 in mammalian cells. This paralleled increased transport of PB2 and NP to the nucleus of the mammalian cells (Gabriel et al., Reference Gabriel, Herwig and Klenk2008). Tarendeau et al. (Reference Tarendeau, Crepin, Guilligay, Ruigrok, Cusack and Hart2008) demonstrated that Lys627Glu mutated domain did not show any structural difference, but a change in charge disrupted a striking basic patch on the domain. RNPs derived from avian viruses showed low levels of polymerase activity in human cells as compared to their counterparts of human lineage viruses. Such observations were associated with reduced binding of NP to PB1–PB2–PA complex and to the PB2 alone. Both deficiencies were restored upon introducing Lys to PB2-627 residue, suggesting a pivotal role of PB2 in molecular interactions of viral NP and other cellular proteins (Labadie et al., Reference Labadie, Dos Santos Afonso, Rameix-Welti, van der Werf and Naffakh2007).

The roles of other internal proteins in host-range specificity are limited compared to those mentioned above. M protein of avian viruses was shown to restrict the replication of avian H2N2 and H3N8 in squirrel monkeys and pigs, respectively. Sequence analysis showed at least 10 sites with amino acids specific for avian and human influenza strains, but which of these contribute to host-range restriction is not known (Buckler-White et al., Reference Buckler-White, Naeve and Murphy1986; Murphy et al., Reference Murphy, Buckler-White, London and Snyder1989; Kida et al., Reference Kida, Ito, Yasuda, Shimizu, Itakura, Shortridge, Kawaoka and Webster1994). In a 2008 study at the CDC, it was shown that the replacement of M protein in an avian virus (H5N1) with that of a human virus (H3N2), not only diminished virus replication in mice, but also altered the tropism of the virus in the infected mice (Chen et al., Reference Chen, Davis, Zhou, Cox and Donis2008). This might indicate a possible role of M protein in functional compatibility of surface glycoproteins (HA and NA) in influenza A viruses.

Determinants of pathogenicity have been identified for NS1 and PB1-F2 proteins, as an interferon antagonist and inducer of apoptosis, respectively. How both proteins contribute to host restriction is not fully understood (Neumann and Kawaoka, Reference Neumann and Kawaoka2006; Conenello et al., Reference Conenello, Zamarin, Perrone, Tumpey and Palese2007). It is speculated that differences in NS sequences might influence their activities in different host systems. While most influenza viruses are sensitive to interferon (IFN) and tumour necrosis factor (TNF)-α, most HPAI H5N1 isolates from human and birds expressed resistance to both cytokines in porcine lung cells. This was associated with Glu at residue 92 of the NS1 protein (Seo et al., Reference Seo, Hoffmann and Webster2002).

Host receptors

Influenza replication in a specific host cell starts at the interaction of virus HA protein with the host cell receptors. Specific receptors for influenza viruses are still not well characterized; however, SA have been identified as receptor determinants for influenza viruses. SA are a family of nine-carbon acid sugars that occupy terminal positions on oligosaccharide chains of glycolipids and glycoproteins, forming exclusively α-glycosidic linkage (Matrosovich et al., Reference Matrosovich, Klenk and Kawaoka2006b). More than 40 elements have been identified in this family which include two main species: N-acetylneuraminic acid (Neu5Ac) and N-glycolylneuraminic acid (Neu5Gc). The Neu5Ac is the most common form and the biosynthetic precursor of other SA. Neu5Gc is found in all mammals except humans. Generally, SA are bound in the form of α2-3- or α2-6-linkage to the D-galactose (Gal), N-acetyl-D-galactosamine (GalNAc) and N-acetyl-D-glucosamine (GlcNAc) of the penultimate sugar (Matrosovich et al., Reference Matrosovich, Klenk and Kawaoka2006b).

The distribution of these receptors varies between species and in the different organs of the same species. The use of linkage-specific lectins, like Maackia amurensis (MAA, specific for α2-3-linkage) and Sambucus nigra (SNA, specific for α2-6-linkage), has enabled the study of distribution of SA receptors in different species. Unfortunately, different studies have shown conflicting results due to the use of different lectins manufactured by different companies. Nevertheless, interesting findings have been reported in the last few years regarding the distribution of SA in different species. In a recent study (Kuchipudi et al., Reference Kuchipudi, Nelli, White, Bain, Chang and Dunham2009), Pekin ducks showed differential expression of receptors in different organs. α2-3-linked SA was exclusively expressed in the intestine, supporting earlier findings. Although tracheal epithelial cells were shown to express the α2-6-linkage of SA, their ratio was very low compared to that of α2-3-linkage (1:20). Interestingly, other organs (vascular endothelium and tubular cells in the kidney, endocardium and alveolar cells) were shown to express substantial amounts of both receptors. Like in ducks, chickens expressed mainly α2-3-linked SA on their intestinal wall; however, α2-6-linked SA were dominant in the tracheal epithelial cells in a ratio of 10:1 to the α2-3-linked form of SA. They also expressed both receptors in various amounts in different organs, with a dominance of α2-6-linkage in the kidney vascular epithelial cells. In turkeys, both types of receptors are expressed on tracheal cells in almost equal amounts, whereas the α2-3-linkage is dominant on turkey intestinal cells (Pillai et al., Reference Pillai, Suarez, Pantin-Jackwood and Lee2009). Examining quail as an intermediate host for influenza zoonotic transmission, Wan and Perez (Reference Wan and Perez2006) reported the abundance of α2-6 as well as α2-3-linked SA receptors in both the trachea and intestine.

Humans, on the other hand, are known to express α2-6-linkages on tracheal epithelial cells. Recent findings (Nicholls et al., Reference Nicholls, Chan, Chan, Wong, Cheung, Kwong, Wong, Chui, Poon, Tsao, Guan and Peiris2007) have shown that α2-3-linked SA are also expressed on alveolar pneumocytes, which could explain the susceptibility of humans to H5N1 HPAI viruses, where such viruses replicated ex vivo in human lungs (Nelli et al., Reference Nelli, Kuchipudi, White, Perez, Dunham and Chang2010; Nicholls et al., Reference Nicholls, Chan, Chan, Wong, Cheung, Kwong, Wong, Chui, Poon, Tsao, Guan and Peiris2007; Chan et al., Reference Chan, Chan, Yu, Ho, Chui, Lo, Yuen, Guan, Nicholls and Peiris2009). Moreover, humans express α2-3-linked SA receptors on their ocular and lacrimal duct epithelial cells, which could explain the disease of conjunctivitis in many patients infected with avian influenza viruses, specifically H7 and H9 subtypes (Olofsson et al., Reference Olofsson, Kumlin, Dimock and Arnberg2005).

Pigs express substantial amounts of both α2-3-Gal- and α2-6-Gal-linked SA receptors on their tracheal epithelial cells (Nelli et al., Reference Nelli, Kuchipudi, White, Perez, Dunham and Chang2010). In addition, pigs have been hypothesized to serve as a ‘mixing vessel’ where both avian and human viruses can co-infect epithelial cells, resulting in the generation of reassortant virus. Though pigs express both types of receptors, avian-like swine viruses showed a shift in receptor binding specificity over time. Until 1984, viruses isolated from pigs in Europe recognized both types of receptors, where those isolated after 1985 were strictly binding Neu5Ac(α2-6)Gal-linked receptors. Ser142Leu mutation was shown to contribute to the loss of binding to Neu5Ac(α2-3)Gal-linked receptors (Ito et al., Reference Ito, Couceiro, Kelm, Baum, Krauss, Castrucci, Donatelli, Kida, Paulson, Webster and Kawaoka1998). In a further study, Glu190Asp and Gly225Glu were shown to be likely possible for initial changes in receptor-binding specificity (Matrosovich et al., Reference Matrosovich, Tuzikov, Bovin, Gambaryan, Klimov, Castrucci, Donatelli and Kawaoka2000). Although still not proven experimentally, consistent results from two reports suggest that swine H1 and H3 viruses bind Neu5Gc not Neu5Ac receptor types (Matrosovich et al., Reference Matrosovich, Klenk and Kawaoka2006b).

Equine (H3 and H7 subtype) and seal (H7 subtype) influenza viruses are more like avian viruses than mammalian viruses. H3 equine viruses have stricter receptor specificity compared to their counterparts from duck origin. They only agglutinate erythrocytes with Neu5Ac(α2-3)Gal-terminated receptors, whereas some duck viruses can agglutinate Neu5Ac(α2-6)Gal-containing erythrocytes. This was supported by the finding that horses express solely α2-3-linked SA receptors on their tracheal wall (Suzuki et al., Reference Suzuki, Ito, Suzuki, Holland, Chambers, Kiso, Ishida and Kawaoka2000). Influenza viruses isolated from seals are mostly of avian origin, but some of these display a lower binding affinity to Neu5Ac(α2-3)Gal-containing receptors without increase in binding affinity towards Neu5Ac(α2-6)Gal (Matrosovich et al., Reference Matrosovich, Klenk and Kawaoka2006b). Dogs are also susceptible to influenza infection of avian (H5) and equine (H3) origin. α2-3-linked SA were shown to be present in dogs' trachea (Maas et al., Reference Maas, Tacken, Ruuls, Koch, van Rooij and Stockhofe-Zurwieden2007), which supports the susceptibility of dogs to both avian and equine influenza viruses.

The actual virus binding to its receptor is greatly affected by many factors that control the chemistry of binding between molecules. Human viruses bind to Neu5Ac(α2-6)Gal(β1-4)GlcNAc (6′SLN) more strongly than to Neu5Ac(α2-6)Gal(β1-4)Glc (6′SL). In contrast, avian viruses bind to both forms, 3′SL and 3′SLN (Mochalova et al., Reference Mochalova, Gambaryan, Romanova, Tuzikov, Chinarev, Katinger, Katinger, Egorov and Bovin2003). Avian viruses bind strongly to Neu5Ac rather than the Neu5Gc (Ito et al., Reference Ito, Suzuki, Suzuki, Takada, Horimoto, Wells, Kida, Otsuki, Kiso, Ishida and Kawaoka2000). Moreover, duck and shore bird viruses bind with greater affinity to Neu5Ac(α2-3)Gal(β1-3)GlcNAc containing receptors than Neu5Ac(α2-3)Gal(β1-4)GlcNAc. Chicken viruses bind more strongly to gangliosides with longer sugar chains compared to duck viruses which bind strongly to those with short sugar chains.

In summary, influenza virus binding to its receptor(s) on the host cell is a complex interaction including several factors on both sides of the reaction. There are currently more questions than answers to explain the specificity of binding of influenza viruses to host cell receptors. Further investigations is needed on the biological, chemical and physical aspects of the virus binding, to understand this process better.

Host immune system

Immunity against influenza virus in a specific host is determined by many factors including the presence of natural barriers that prevent virus replication, vaccination or pre-infection of the host with homologous and/or heterologous virus and genetic makeup of the host.

The respiratory tract secretions in Chimpanzees contain mucins that can bind specifically to influenza viruses and prevent the virus entry to the airway epithelial cells. This could explain the reasons for relative resistance of Chimpanzees to experimental respiratory exposure to human influenza viruses (Kuiken et al., Reference Kuiken, Holmes, McCauley, Rimmelzwaan, Williams and Grenfell2006).

In birds, influenza virus infection can directly affect the immune response, but the insights about the roles of the specific innate and adaptive immune response mediators, such as the Mx protein, interferons and cytokines remain limited (Suarez and Schultz-Cherry, Reference Suarez and Schultz-Cherry2000). Wild ducks are the principle reservoir of influenza A viruses that can be transmitted to domestic poultry and mammals including humans (Kim et al., Reference Kim, Negovetich, Forrest and Webster2009). Further evidence revealed that the intestinal tract of ducks (crypts of Lieberkuhn in the large intestine) is the major replication site of influenza viruses (Kida et al., Reference Kida, Yanagawa and Matsuoka1980). The species diversity of ducks may also play an important role in the pathogenicity of influenza viruses (Kim et al., Reference Kim, Negovetich, Forrest and Webster2009). It is noteworthy that a lack of adequate specific secondary immune responses in ducks following re-infection, may be responsible for ducks being susceptible to all influenza subtypes, and thus serve as a potential reservoir (Kida et al., Reference Kida, Yanagawa and Matsuoka1980). In one exhaustive study on influenza viral pathogenesis and immune responses in 2-week-old chickens, turkeys, quails and ducks infected with eight influenza viruses of H5 subtype (by intramuscular, intranasal and contact routes), six of the viruses were highly pathogenic for chickens and turkeys, but only two viruses produced consistent serological responses in ducks (Alexander et al., Reference Alexander, Parsons and Manvell1986). This study confirmed the contribution of host factors within and between bird species in pathogenicity as well as transmission of influenza viruses.

Because influenza is an acute infection, cell-mediated immune responses through anti-influenza specific cytotoxic lymphocytes often fail to provide adequate protection, suggesting important roles for the innate and humoral immune responses in effective viral clearance. Most of the available information regarding immune responses to influenza viruses has been based on experiments in inbred mice, which many times are not relevant in other species.

In a recent study, infection of several laboratory inbred mice with H1N1 (PR8) and H7N7 (SC35M) viruses resulted in different weight loss kinetics and survival rate after infection (Srivastava et al., Reference Srivastava, Blazejewska, Hessmann, Bruder, Geffers, Mauel, Gruber and Schughart2009). Additionally, the inbred mouse strains (C57BL/6J, DBA/2J, FVB/NJ, CBA/J, BALB/cByJ, (C57BL/6J×DBA/2J) F1, DBA/2NHsd and A/JOlaHsD (A/J)) showed distinct susceptibility (LD₅₀) to infection and expressed differential virus replication titers and cytokine production upon infection. The mechanisms for such variations have not been identified and similar studies should be done in other species. It is worth mentioning that two of the mice strains were deficient for Mx1, a major effector of interferon response to inhibit virus replication, and the other strains carried mutated Mx1. Whether these variations alone or together with other elements explain the above findings requires further investigation (Srivastava et al., Reference Srivastava, Blazejewska, Hessmann, Bruder, Geffers, Mauel, Gruber and Schughart2009).

Mx proteins are GTPases that are expressed in the nucleus and cytoplasm of rodent cells, but mostly in the cytoplasm of other vertebrate species. The expression of Mx protein in mice was associated with a high degree of resistance to influenza infections, in vivo and in vitro (Haller et al., Reference Haller, Staeheli and Kochs2007). The antiviral activity of Mx proteins is reported in many species, including mammals, birds and fishes. They are polymorphic in most species, and thus, they have different modes of action in different species. Their mechanism of action is not fully understood; however, it is suggested that Mx proteins bind to essential viral components (mostly NP) during virus replication and block their function (Haller et al., Reference Haller, Staeheli and Kochs2007). In an earlier study, the expression of chicken and duck Mx genes in avian and mammalian cell lines (including mouse 3T3 cells) demonstrated no antiviral activity for several viruses including influenza. The Mx of chicken in that study was cloned from the White Leghorn breed in Germany (Bernasconi et al., Reference Bernasconi, Schultz and Staeheli1995; Ko et al., Reference Ko, Jin, Asano, Takada, Ninomiya, Kida, Hokiyama, Ohara, Tsuzuki, Nishibori, Mizutani and Watanabe2002). In a later study, Ko et al. (Reference Ko, Jin, Asano, Takada, Ninomiya, Kida, Hokiyama, Ohara, Tsuzuki, Nishibori, Mizutani and Watanabe2002) demonstrated the existence of a highly polymorphic Mx gene in different chicken breeds. Some Mx proteins expressed resistance to influenza and other viruses, and such activity was associated with residue 631 of the Mx protein. On the other hand, based on the above findings, duck Mx protein lacks antiviral activity. A comparison study like the one performed in chickens should be done on ducks before drawing any conclusions.

The duck immune system is quite different from that of chickens and other poultry species. Ducks have been reported to produce poor antibody responses and to express low HI antibody activity to natural and experimental infection with avian influenza viruses. The inability of ducks to produce the hemagglutinating antibody is also associated with other deficiencies in duck antibodies, including precipitation, complement activation and opsonization. The presence of only two constant regions in the heavy chain of the smaller IgY form in ducks, probably eliminates the functions associated with Fc region of the antibody molecule, including hemagglutination (Suarez and Schultz-Cherry, Reference Suarez and Schultz-Cherry2000). In one recent study, duck and chicken peripheral blood mononuclear cells expressed distinct levels of cytokines upon infection with H11N9 LPAI (Adams et al., Reference Adams, Xing, Li and Cardona2009). How such a difference in host response between the two species is related to the subclinical and persistent infections observed in ducks is something to be further investigated.

Influenza viruses from wild birds (including ducks) show a lower rate of evolution compared to those from poultry and mammals (Suarez, Reference Suarez2000). This might be explained by the difference in immune pressure (lesser in ducks compared to chickens and humans) exerted on the replicating viruses in different hosts. When the virus jumps from one species to another (example from ducks to chickens) it becomes more prone to mutations, and thus, antigenic drift might result in more transmissible and pathogenic viruses. On the other hand, the pre-existing immunity in the host due to prior immunization or natural infection might protect the host against clinical signs from LPAI or HPAI virus infection, but not necessarily against viral shedding. In such situations, there is a probability of generation of many mutants in the shed material (quasi-species), with potentially different behavioral characteristics than the parental strain.