Influenza viruses

Abstract

Avian influenza viruses are the precursors of human influenza A viruses. They may be transmitted directly from avian reservoirs, or infect other mammalian species before subsequent transmission to their human host. So far, avian influenza viruses have caused sporadic—yet increasingly more frequently recognized—cases of infection in humans. They have to adapt to and circulate efficiently in human populations, before they may trigger a worldwide human influenza outbreak or pandemic. Cross-species transmission of avian influenza viruses from their reservoir hosts—wild waterbirds—to terrestrial poultry and to humans is based on different modes of transmission and results in distinctive pathogenetic manifestations, which are reviewed in this paper.

Keywords: :

• avian
• cross-species transmission
• human
• influenza
• pathogenesis
• zoonosis

Introduction

Avian influenza viruses are the precursors of human influenza A viruses.^1,2 They may be transmitted directly from their avian reservoirs, or infect other mammalian species before subsequent transmission to human hosts. In humans, these zoonotic viruses may cause sporadic cases of infection which upon further adaptation and/or reassortment with other influenza A viruses occasionally lead to human influenza pandemics. The latter are eventually followed by recurring epidemics caused by further human-adapted variants, which bear high morbidity and mortality burdens. It should be noted that the cumulative burden of disease caused by the annually recurring influenza epidemics in the last century may roughly have equalled that of the influenza pandemics of the same period.^3,4

Waterbirds of the orders Anseriformes (mainly geese, ducks and swans) and Charadriiformes (mainly gulls and waders) are the natural host reservoirs of avian influenza A viruses.^1,5 Influenza A viruses are classified into subtypes that are defined by the hemagglutinin (HA) and neuraminidase (NA) surface glycoproteins. A large diversity of subtypes circulates in wild waterbirds. To date, 16 subtypes of HA protein and 9 subtypes of NA protein have been described and virtually all possible combinations of these two proteins have been identified on influenza viruses of these birds. Avian influenza viruses may be transmitted from wild waterbirds to poultry, in which a smaller diversity of subtypes is found.⁶ Occasionally, some strains have been transmitted from wild or domestic birds to mammals.⁷ Besides in humans, adaptation and circulation of influenza viruses independently of avian reservoirs have occurred in swine, horses, and domestic dogs. In these animals specific lineages of a limited number of subtypes have established.

The pathogenesis of influenza virus infection in wild birds, poultry, and mammalian hosts, including humans, differs. In wild birds, avian influenza viruses typically cause very mild or sub-clinical intestinal tract infection, and are transmitted via the fecal-oral route.^1,8 In poultry, they generally cause mild or sub-clinical respiratory tract infection,^9,10 and for this reason are referred to as low pathogenic avian influenza viruses (LPAIV). Upon transmission into terrestrial poultry such as chickens and turkeys, LPAIV of the H5 and H7 subtypes may evolve into highly pathogenic avian influenza viruses (HPAIV) and usually cause lethal systemic infection in these species. LPAIV and HPAIV may be transmitted via fecal-oral and respiratory routes in poultry. In mammals, including humans, LPAIV and adapted variants generally cause respiratory disease of variable severity, and are transmitted via the respiratory route. In humans, LPAIV infections have resulted from cross-species transmission of LPAIV H9N2, H7N2, H7N3, and H7N7 (Table 1). HPAIV are rarely transmitted from poultry to other species, yet have caused, over the past decade, ocular, respiratory or systemic infection in humans and other animal hosts. In humans, HPAIV infections have resulted from cross-species transmission of HPAIV H5N1, H7N3 and H7N7 (Table 1).^11-16 Atypical disease manifestations, such as gastro-intestinal or neurological symptoms, have been reported following infection with HPAIV H5N1, while ocular infection was reported for HPAIV of the H7 subtype. In most of these cases, HPAIV are transmitted from birds to humans via the respiratory or ocular route, although some strains of HPAIV H5N1 may use both oral and respiratory routes of infection in mammals and possibly also in humans. So far, no or little human-to-human transmission of LPAIV or HPAIV has been reported.

Table 1. Reported cases of avian influenza virus infection determined by viral isolation

Host of origin		Year, Place	Patho-type	Virus subtype	Associated activity	Ref.
Wild birds	Swans, Cygnus spp	2005, Azerbaijan	HP	H5N1	De-feathering Meat processing	34
Poultry	Chickens and domestic ducks	1997–present, Asia and Africa	HP	H5N1	Caregiving Butchering De-feathering Meat processing Possibly consumption Unknown	15
	Chickens	2007, UK	LP	H7N2	Visiting bird market	106
	Chickens	2006, UK	LP	H7N3	Caregiving	106
	Chickens	2004, Canada	HP	H7N3	Caregiving Managing outbreak	106
	Chickens	2003, Netherlands	HP	H7N7	Caregiving Managing outbreak	106
	Chickens Turkeys	2002–2003, Italy	LP	H7N3	Managing outbreak	106
	Turkeys	2002, USA	LP	H7N2	Managing outbreak	106
	Domestic ducks	1996, UK	LP	H7N7	Caregiving	106
	Unknown	2007, Hong Kong	LP	H9N2	Unknown	55
	Unknown	2003, USA	LP	H7N2	Unknown	106
	Unknown	2003, Hong Kong	LP	H9N2	Unknown	55
	Unknown	1999, Hong Kong	LP	H9N2	Unknown	55
	Unknown	1998, China	LP	H9N2	Unknown	55

LP, low pathogenic; HP, highly pathogenic

The observed differences in pathogenesis illustrate that wild waterbirds, poultry and mammals, such as humans, offer host environments that are exploited differently by avian influenza viruses. Understanding the pathogenesis of avian influenza virus infections upon cross-species transmission from birds to humans is essential for pandemic preparedness. It represents the initial step potentially heralding further adaptation and efficient circulation of these viruses in human populations, at the origin of human influenza pandemics. Comparing the pathogenesis of avian influenza virus infection in avian and human hosts upon cross-species transmission is the subject of this review.

Routes of Entry and Transmission

Avian influenza viruses are transmitted in wild waterbirds via the fecal-oral route.^1,8 Aquatic habitats and associated waterbird ecology favor (1) wild waterbirds as preferred reservoir hosts of LPAIV, and (2) the fecal-oral route of transmission as the basis for spread and maintenance of LPAIV in these avian species. The ecology of waterbirds drives, at least partly, the observed high prevalence of LPAIV infection in birds of the orders Anseriformes and Charadriiformes.⁵ Feeding and social behavior as well as population sizes and densities of most waterbird species likely promote contact between individual birds and bird species and therefore exposure to and transmission of avian influenza viruses.^17,18 Although LPAIV may transiently be present in the respiratory tract, they are excreted into the environment mainly from the digestive tract of infected waterbirds. Large quantities of virus may be shed from the cloaca, estimated in one study at more than 10⁸ EID₅₀ per gram of duck faeces, for a total of 10¹⁰ EID₅₀ per duck per day.¹⁹ Aquatic habitats may facilitate fecal-oral transmission of these viruses among waterbirds. LPAIV can persist several months in environmental reservoirs, such as surface water of lakes.²⁰ Indirect transmission of LPAIV from aquatic environmental reservoirs is increasingly believed to be essential for maintaining LPAIV locally and from year to year.^21-23 Waterbirds feeding on surface water may therefore be particularly prone to LPAIV exposure and infection.

Avian influenza viruses are transmitted from wild birds to poultry most likely via the fecal-oral route, following shared use of aquatic habitats or shared sources of drinking water.⁶ Domestic ducks appear to be infected with a large diversity of LPAIV.²⁴ It is probable that, in these species, LPAIV circulate independently of other wild birds, through the fecal-oral route of transmission. The route of transmission of avian influenza viruses in terrestrial poultry is more difficult to assess. However, the ecology of terrestrial birds, e.g., associated with drier environments, may favor respiratory transmission of LPAIV and HPAIV via respiratory droplets and aerosols. Furthermore, pharyngeal shedding is typically equal to, or higher than cloacal shedding in these species.⁹ Experimental transmission of a strain of LPAIV H9N2 occurred via contact, fecal exposure and aerosols in quails.²⁵ In another study, LPAIV H6N2 transmitted via fecal exposure and aerosols in chickens.²⁶

The ability of LPAIV from wild birds to transmit to and among poultry hosts depends on the subtype and strain of the virus, as well as on the poultry species, and requires a degree of adaptation of the virus to their novel hosts after introduction.²⁴ In general, LPAIV isolated from poultry replicate to higher levels in poultry than LPAIV isolated from wild birds.²⁷ Similarly, a HPAIV strain isolated from turkeys was shown to infect and transmit only inefficiently in chickens.²⁸ Likewise, LPAIV and HPAIV circulating in poultry rarely spill back into wild waterbird populations. HPAIV H5N1 that re-emerged in 2002 in South-East Asia are exceptions as these viruses have infected a wide range of wild bird species.²⁹ The potential routes of transmission of these viruses from poultry to wild birds include inhalation of infectious droplets or aerosols, ingestion of contaminated water, or ingestion of infected carcasses. The latter is the most probable route of infection for predatory and scavenging bird species, such as buzzards (Buteo spp) and crows (Corvus spp), found infected with HPAIV H5N1. In wild birds, HPAIV H5N1 is shed mainly from the pharynx and is hardly detected in the cloaca. Transmission of these viruses among wild birds has resulted in self-limiting epidemics, yet it remains unknown whether HPAIV H5N1 can be maintained in wild bird populations. Transmission of HPAIV H5N1 among wild birds may occur via the respiratory route, or via the oral route following ingestion of contaminated water or infected carcasses. The potential difference in the route of transmission of HPAIV H5N1 in wild birds compared with that of LPAIV may well have important consequences for the ability of these viruses to be maintained in wild bird populations.

Avian influenza viruses are typically transmitted from wild or domestic birds to humans via the respiratory route, presumably following inhalation of infectious fomites, droplets or aerosols, or possibly following self-inoculation of the upper respiratory tract.³⁰ Such routes of transmission also characterize influenza viruses adapted to circulate in human populations (i.e., pandemic and seasonal influenza viruses), yet at a much more efficient level.³¹ Occupational exposure to poultry greatly increases the risk of infection with avian influenza viruses.³² Most human cases of infection with avian influenza viruses are sporadic and are associated with close contact with poultry or their products. At-risk activities include caregiving, slaughtering, butchering and preparation of meat for consumption (Table 1). In Egypt, mostly women care for poultry and carry out the slaughtering, de-feathering and preparation of meat for consumption, and were shown to present a higher risk of infection with HPAIV H5N1 than men.³³ Likewise, close contact with and de-feathering of infected wild swans (Cygnus spp) was considered to be the most probable source of exposure to HPAIV H5N1 that resulted in two clusters of human infection and six human deaths in Azerbaijan.³⁴ Inhalation of influenza virus particles or self-inoculation of the upper respiratory tract during such at-risk activities likely resulted in infection. However, given the relatively low frequency of human infection with avian influenza viruses despite intense and widespread contact between poultry and humans worldwide, such route of transmission remains greatly inefficient, for poorly understood reasons. Bathing or swimming in contaminated waters has been incriminated as other possible routes of exposure and infection with HPAIV H5N1 in South-east Asia.³⁰ Direct intranasal or possibly conjunctival inoculation while swimming in contaminated water or inhalation of water may represent respiratory routes of transmission associated with these activities. In addition to respiratory transmission of avian influenza viruses from birds to humans, other atypical routes have been suggested. Infection via consumption of raw duck blood or poorly-cooked poultry products has been suspected in some cases, supporting a potential digestive route of infection by HPAIV H5N1. A digestive route of infection is unusual for influenza viruses in humans and other mammals, yet was demonstrated to occur for HPAIV H5N1 in ferrets, mice, hamsters and cats.^35-37,107 Similarly to respiratory transmission of avian influenza viruses, such a route of transmission apparently remains highly inefficient, given its relatively low frequency. Lastly, self-inoculation of the conjunctiva appears to be an important and atypical route of bird-to-human transmission of avian influenza viruses of the H7 subtype, which tends to present preferential tropism for ocular tissues in humans.³⁰

Human-to-human transmission of avian influenza viruses is rare, and typically results from intensive and close physical contacts with the index case, and is not sustained into additional chains of transmission.³⁰ This supports the inefficient ability of these viruses to further use the respiratory route of transmission from human to human. The reasons behind the limited transmissibility of avian influenza viruses are poorly understood yet discriminate between viruses causing sporadic cases of infection, and those potentially triggering a worldwide pandemic, four of which have occurred since the beginning of last century.^2,38,39 The most severe influenza pandemic (“Spanish flu”) started in 1918 and was caused by a virus of the H1N1 subtype; it resulted in approximately 50 million deaths worldwide and infection of about one-third of the global population. In 1957, the second influenza pandemic to be recorded (“Asian flu”) was caused by a virus of the H2N2 subtype. About a decade later, a virus of the H3N2 subtype emerged and resulted in the pandemic of 1968 (“Hongkong flu”). Each led to 1–4 million deaths. Lastly, the recent pandemic of 2009 (“Mexican flu” or “Swine flu”) was caused by a virus of the H1N1 subtype, resulting in infection of 20–30% of the global population.⁴⁰ These pandemic viruses were at least partly of avian origin. Possibly with the exception of the 1918 H1N1 virus,^41,42 reassortment of avian influenza viruses with mammalian influenza viruses—of human origin in the cases of the 1957 and 1968 viruses,² and of swine origin in the case of the 2009 H1N1 virus³⁹—appears to be a common factor at the onset of past three pandemics. Although avian influenza viruses can support infection of humans upon cross-species transmission, adaptation to mammalian hosts thus may be required for sustained human-to-human transmission. Avian influenza virus tissue tropism is among the factors that may limit avian influenza virus transmissibility from birds to humans, and subsequently among humans, and may represent a key target for such adaptation to take place (see below).

Tissue Tropism

In waterbirds, LPAIV preferentially infect epithelial cells of the intestinal tract, particularly in the distal regions of the small intestine, as well as epithelial cells of the bursa of Fabricius.^43-46 The HA protein of LPAIV is cleaved by extracellular trypsin-like proteases present in the intestinal tract of birds; this contributes to the localized nature of the infection.^47-49 Following ingestion, LPAIV must pass through and survive the acidic environment of the proventriculus of waterbirds. Uncleaved HA proteins of LPAIV appear more resistant to acidic pH than those of mammalian influenza viruses,¹ and ingestion of large amount of water may partly neutralize the acidic pH of the waterbird stomach, which together probably allow infectious LPAIV to reach their target site of replication in the intestine.

HPAIV rarely infect wild waterbirds. However, since 2002, HPAIV H5N1 have spilled back from poultry into a wide range of wild bird species. The HA protein of HPAIV presents a multi-basic cleavage site, which can be cleaved by intracellular, subtilisin-like proteases, present in a wide range of cell types in avian and mammalian species.⁴⁹ In contrast to LPAIV infection, cells along the intestinal epithelium are not or rarely reported infected by HPAIV H5N1 in wild birds, and the respiratory epithelium appears to be the initial site of HPAIV H5N1 replication in these species.²⁹ Subsequent viraemia may lead to secondary infection of other organs. However, the degree of dissemination differs greatly among species, from localized respiratory tract infection to dissemination to multiple organs. In many wild bird species, tropism for the central nervous system is among the most frequent tropisms beyond the respiratory tract. The virus also may spread to other organs, typically to pancreas, liver, adrenal gland, kidney, heart and skeletal muscle, where it was found to replicate in experimentally infected wild birds. In most naturally and experimentally infected wild bird species, parenchymal cells are the main targets for HPAIV H5N1 replication.

In poultry, LPAIV preferentially infect epithelial cells of the respiratory tract, including along the trachea, bronchi, alveoli and air sacs.^10,50 Extra-cellular proteases required for the cleavage of LPAIV HA proteins are present in the respiratory tract of birds.⁴⁹ Infection of renal tubular epithelial cells and intestinal mucosal epithelial cells was also reported following intravenous inoculation of LPAIV in chickens.⁵¹

HPAIV have a wide tissue tropism in poultry. Initial infection occurs in respiratory epithelial cells, including in the nasal cavity, and rapidly extends into the underlying submucosa, with the presence of influenza virus antigen detected in capillary endothelial cells.^9,10 HPAIV massively disseminate via the circulatory system, resulting in widespread infection of endothelial cells in many organs. Endothelial infection is a hallmark of HPAIV pathogenesis in terrestrial poultry. It rarely occurs in wild birds infected with HPAIV H5N1, but was reported in mute swans (Cygnus olor), whooper swans (C. cygnus) and tufted ducks (Aythya fuligula).^52-54 Infection of parenchymal cells, mainly in the pancreas, brain, heart, kidney and skeletal muscle, may occur, essentially in poultry surviving the acute phase of infection.²⁹

In humans, LPAIV and HPAIV preferentially infect respiratory epithelial cells, in particular along the deeper regions of the respiratory tract.^55,56 Extra-cellular proteases activating cleavage of LPAIV are present in the respiratory tract of mammals, including humans.⁴⁹ LPAIV and HPAIV may also infect cells beyond the respiratory tract. LPAIV and HPAIV of the H7 subtype can infect the human eye, possibly epithelial cells along the conjunctiva and cornea.^55,57 Detailed studies on the distribution of HPAIV H5N1 RNA or viral antigen in post-mortem tissues of two fatal human cases revealed infection of ciliated and non-ciliated epithelial cells in the trachea, and more abundantly of alveolar epithelial cells (type II pneumocytes) in the deeper lungs as well as infection of neurons in the brain, T cells in the hilar lymph node, Kuppfer cells in the liver, and mucosal epithelial cells in the small intestine.⁵⁸ Whether the latter can be infected following ingestion of HPAIV H5N1 is unknown. Such a route of infection has been demonstrated in ferrets, mice, hamsters and cats^35-37,107 and implies that these viruses can survive the acidic environment of the stomach in these species.

The tissue tropism of influenza viruses is at least in part determined by the receptor binding affinity of the HA protein. Avian influenza viruses have a preferred receptor binding affinity for cellular glycans harboring sialic acids with α2,3 linkage to galactose.^59,60 These cellular receptors are abundant in the intestinal and respiratory tracts of wild and domestic birds, but also in other tissues, such as the heart, kidney, brain and endothelium in Pekin ducks and chickens.^47,48,61 In humans, sialic acids with α2,3 linkage to galactose were shown to be expressed along the respiratory tract, on rare epithelial cells of the nasal mucosa and pharynx, focally on tracheal, bronchial and bronchiolar epithelial cells, more abundantly on alveolar epithelial cells (type II pneumocytes), as well as on acinar cells of the submucosal glands and in secreted mucins along the airways, as determined by use of lectin histochemistry.^62-64 Beyond the respiratory tract, sialic acids with α2,3 linkage to galactose were detected on Kuppfer cells in the liver, on neurons in the brain and in the wall of the intestine, on endothelial cells in the heart and kidney, and on ocular and lachrymal duct epithelial cells.^63,65 Binding studies of avian influenza viruses to human tissues by use of virus histochemistry further demonstrated virus attachment mainly to bronchiolar cuboidal epithelial cells, type II pneumocytes and alveolar macrophages in the lower respiratory tract, to acinar cells of the submucosal glands and mucus in the trachea and bronchi, and to corneal and conjunctival epithelial cells in the eye.^57,66-68

Avian influenza virus tissue tropism may limit the ability of these viruses to transmit from birds to humans, and subsequently among humans, via the respiratory, fecal-oral or ocular routes of transmission. First, due to the localization, mostly in the deeper regions of the respiratory tract of humans, target cells may not easily be reached by avian influenza viruses transmitted via the respiratory route of transmission. Only the smallest droplets deposit in the deeper regions of the human respiratory tract,⁶⁹ and the extent to which avian influenza virus particles can be transported and released in association with such droplets, and attain the deeper regions of the human respiratory tract remains unknown. Furthermore, the muco-ciliary escalator may trap and remove influenza virus particles depositing along the airways toward the upper regions of the respiratory tract,⁷⁰ preventing them from reaching their target cells. Second, the environmental conditions within the intestinal tract of humans may limit the ability of these viruses to cause infection following ingestion. The acidic pH of the human stomach may inactivate HPAIV H5N1, due to the sensitivity of cleaved HA proteins.¹ Furthermore, avian influenza virus receptors are rare along human intestinal tract, and were detected below the intestinal epithelial layer, i.e., on neurons of the submucosal and myenteric plexi.⁵⁸ Such location may limit the accessibility of cellular receptors to avian influenza viruses present in the lumen of the intestinal tract, and may require high infectious doses for infection to take place, possibly upon ingestion of large quantities of virus or in association with animal tissues. Third, the mechanical and innate defenses associated with the human eye likely require invasive insults to allow avian influenza virus infection of the ocular epithelia. Furthermore, the localized nature of ocular infection may impede further transmission among humans, limiting possible adaptation to human hosts.

Influenza viruses that are successfully transmitted among humans typically attach to different cellular receptors than avian influenza viruses, resulting in different patterns of tissue tropism. Human influenza viruses preferentially attach to sialic acids with α2,6 linkage to galactose.⁵⁹ These receptors are mostly expressed on ciliated epithelial cells along the upper regions of the respiratory tract.^66,71,72 It has been suggested that α2,6 receptor binding affinity and replication to high viral titers in the upper regions of the respiratory tract are necessary for efficient transmission of influenza viruses in human populations. Pandemic influenza viruses had either dual α2,3 and α2,6 receptor binding affinity or α2,6 receptor binding affinity, and show tropism for the upper regions of the respiratory tract.^2,73 Although the H1 proteins of 1918 and 2009 pandemic viruses were of mammalian virus origin,^2,41 the H2 and H3 proteins of 1957 and 1968 pandemic viruses were of avian virus origin.² Avian H2 and H3 proteins have strong α2,3 receptor binding affinity, indicating that the switch to α2,6 receptor binding affinity coincided with the emergence of these pandemic viruses. Interestingly however, some strains of the 1957 H2N2 pandemic virus had conserved α2,3 receptor binding affinity.⁷⁴ Whether such receptor binding affinity characterized the initial isolates or was acquired during passage in ovo in the laboratory is not known. One of these strains did not transmit efficiently via aerosols in ferrets.⁷⁵ On the other hand, some strains of LPAIV H9N2 and LPAIV and HPAIV of the H7 subtype that have crossed from avian to human hosts presented α2,6 receptor binding affinity, yet were apparently unable to transmit among humans.^76-79 Likewise, recombinant HPAIV H5N1 viruses displaying α2,6 receptor binding affinity replicated in the upper regions of the respiratory tract, yet were unable to transmit via aerosols in the ferret model.⁸⁰ Although α2,6 receptor binding affinity and replication to high viral titers in the upper regions of the respiratory tract may be necessary for efficient human-to-human transmission of influenza viruses, these characteristics are most likely not sufficient.⁸⁰ In particular, the conditions and viral characteristics for initial inefficient transmissibility, potentially preceding acquisition of α2,6 receptor binding affinity, are elusive, and call for further research. Likewise, the effect of pathogenicity, host responses and immunity on the ability of avian influenza viruses to adapt and spread efficiently in human populations is unclear, yet may not be negligible.

Pathology and Host Responses

In wild waterbirds, LPAIV infection typically causes no or very mild intestinal lesions or clinical signs.^1,43-46 Similarly, no or very mild lesions in the bursa of Fabricius were detected in naturally infected mallards and experimentally infected ducks. LPAIV are cytolytic viruses that result in destruction of infected cells. Because intestinal epithelial cells have a relatively rapid turnover, it is possible that destruction of infected intestinal epithelial cells coincides with shedding of these cells into the intestinal lumen, reducing the pathological impact of the infection while favoring viral excretion.⁴³ In addition to the absence or very mild nature of the lesions, the host response to infection is remarkably limited in wild waterbirds, with no detection of inflammatory cells around foci of infection,⁴³ and only transient low-levels antibody titers following infection.⁴⁶ It has been suggested that the apparent confinement of LPAIV infection to surface epithelia and the absence of lesions may limit activation of systemic immune responses.⁴¹ This may contribute to prolonged infection and cloacal shedding for up to 3–4 weeks in these birds.¹⁹ Most waterbird species harboring LPAIV are migratory birds. Sustained infection of pathogens by migratory birds is likely facilitated when these bear low morbidity costs,⁸¹ thus reducing their impact of the birds’ health and ability to migrate. Selection of LPAIV in wild waterbirds may thus be directed toward low or avirulence⁵ and the intestinal tract may represent the optimal location for productive yet sub-clinical infection.

In contrast to LPAIV, HPAIV H5N1 infection in wild birds may cause severe necrosis and inflammation in multiple organs of susceptible species, including pancreas, lungs, air sacs, brain, liver, heart and adrenal glands.²⁹ In addition, whooper swans and mute swans may have hemorrhages in multiple organs, in accordance with the endotheliotropism of HPAIV H5N1 in these species.^52-54 Virus shedding occurs mostly from the pharynx and lasts from 3 to 11 d in wild waterbirds following experimental infection.²⁹ Although HPAIV H5N1 have caused self-limiting epidemics in wild waterbirds, it remains unknown whether they can be maintained in wild waterbird populations. High mortality may deplete susceptible species populations and limit their ability to maintain these viruses; however, several species were shown to present only mild or sub-clinical infection,⁸² and may represent candidate species for maintenance of HPAIV H5N1. Interestingly, maintenance of HPAIV H5N1 in wild birds may occur via a different mode of transmission than that of LPAIV.

In poultry, LPAIV infection typically causes sub-clinical infection or mild respiratory lesions and disease.¹⁰ Lesions of tracheitis, bronchitis and broncho-interstitial pneumonia are characterized by foci of inflammation associated with loss of cilia and submucosal lymphoid hyperplasia, from the trachea to the lungs and air sacs. Intravenous inoculation of LPAIV results in additional lesions in the reproductive and urinary tracts, including atrophy and inflammation of the ovaries and oviducts, renal necrosis and inflammation (nephritis), and bursal necrosis. Clinical signs associated with natural LPAIV infection include dysfunction in the respiratory, reproductive or urinary systems, resulting in mild morbidity and drop in egg production.

In contrast, HPAIV infection causes disseminated lesions in poultry, associated with dysfunction of the cardiovascular system principally, and of the nervous system in some cases. In chickens, common lesions include edema or necrosis of comb and wattle, edema of the head and legs, subcutaneous hemorrhage of legs, lungs that fill with fluid and blood, and small hemorrhages in internal organs. Microscopically, these lesions are characterized by inflammation, necrosis, hemorrhage, edema or a combination of these. Lesions are located in multiple organs, including the comb and wattle, pancreas, brain, heart, kidney, comb and wattle, and lymphoid organs. These lesions are associated with high morbidity and mortality in poultry. Death occurs rapidly, typically within 2 d, and is sudden often even without visible clinical signs, or in association with edema and hemorrhage in the head and legs. The emergence and maintenance of HPAIV in poultry, despite their typical high virulence in these species, may stem from the high densities characterizing poultry populations of industrialized farms, and from movements and trade of infected birds between premises.^83,84

In humans, LPAIV infection typically causes mild respiratory disease similar to influenza-like illness, as well as conjunctivitis in the case of viruses of the H7 subtype, which resolves within 1–2 weeks.^55,56 Clinical symptoms associated with LPAIV H9N2 infection include fever and coughing, sore throat, decreased appetite, abdominal pain and vomiting. Clinical symptoms associated with LPAIV or HPAIV of the H7 subtype generally include conjunctivitis or in rare cases intraepithelial keratitis, as well as respiratory and general clinical symptoms associated with mild influenza-like illness. However, during the HPAIV H7N7 outbreak in the Netherlands in 2003, one patient developed severe pneumonia and died of acute respiratory distress syndrome.¹¹ The lungs presented with severe diffuse alveolar damage, characterized by flooding of the alveolar lumina with serosanguineous fluid mixed with fibrin and neutrophils.¹¹ No viral antigens were detected in association with these lesions, which may be due to the protracted nature of the lesions and late course of the disease in this individual. Similar lesions and disease are observed upon human infection with HPAIV H5N1, although dissemination of the virus and occurrence of lesions in organs beyond the respiratory tract are also reported. The lungs are centrally involved upon infection with HPAIV H5N1 and present with severe lesions associated with the exudative and proliferative phases of diffuse alveolar damage. These are characterized by lesions of necrosis and inflammation resulting in broncho-interstitial pneumonia, with loss of respiratory epithelium, interstitial infiltration of lymphocytes and neutrophils, and predominance of macrophages in alveolar lumina; as well as by regenerative lesions, such as type II pneumocyte hyperplasia.^85,86 Extra-respiratory lesions include necrosis and inflammation in the brain, liver and lymphoid organs, in association with the presence of HPAIV H5N1 antigen in parenchymal cells in these organs.^55,56 The severity of respiratory lesions in humans may be associated with dysfunction of the immune responses, leading to high levels of pro-inflammatory cytokines and immunopathology in the lungs. Several genetic markers of pathogenicity also characterize HPAIV H5N1.⁵⁵ Clinically, HPAIV H5N1 infection leads to severe acute respiratory distress syndrome and multi-organ dysfunction including liver and renal failure. Atypical manifestations include gastro-intestinal and neurological clinical symptoms. The case fatality rate approaches 60% and median time from onset to death is 9 to 10 d. The severe lesions and associated clinical signs following HPAIV H5N1 infection may limit virus excretion due to mechanical obstruction of the smaller airways, may limit contact between infected and naive individuals due to high morbidity rates, and may reduce the length of the infectious period due to high mortality rates, impeding transmission and maintenance of these viruses among humans. This does not explain however why other less pathogenic avian influenza viruses, that have crossed from avian to human hosts, in particular LPAIV H9N2, have so far not adapted to spread in human populations.

Immunity

Avian and mammalian hosts mount innate and adaptive immune responses upon infection with influenza viruses. Innate immune responses are contemporary to the acute infection, and in mammals, include production of pro-inflammatory cytokines (e.g., tumor necrosis factor TNF-α and type I interferons IFN-α/β) by both infected and dendritic cells. These cytokines prevent virus replication by inducing uninfected cells to enter into a refractory state, and attract natural killer and antigen-presenting cells to the site of infection.⁸⁷ The innate immune responses of avian species following influenza virus infection are not well understood, but likely involve type I interferons IFN-1, which are similar to mammalian IFN-α/β.⁸⁸ Cellular and humoral adaptive immune responses involve T-helper lymphocytes, immunoglobulin-producing B-lymphocytes and cytotoxic T-lymphocytes in mammals. These appear several days after infection and contribute to influenza virus clearance and to the development of immune memory. While humoral immunity has been described in avian species upon influenza virus infection, little is known on cellular immune responses in these species.

Wild waterbirds infected with LPAIV produce transient low levels of antibodies, which have been claimed not to be virus neutralizing, possibly due to deficiencies associated with the structure of their antibodies.⁸⁸ IgA antibodies were detected in the bile of ducks infected with different LPAIV,⁸⁹ and are probably also present on other mucosal surfaces, including along the intestinal epithelium, based on the expression pattern of IgA genes.⁹⁰ Although short-lived,⁸⁹ these antibodies may confer protection against subsequent LPAIV infection with viruses of the same subtype.⁸⁸ Immunity to LPAIV protected domestic ducks and mallards against disease and greatly reduced virus shedding upon challenge with the same subtype,^46,91 and to a lesser degree upon challenge with HPAIV H5N1.⁹² Heterosubtypic immunity may be conferred by cellular immune responses in wild waterbirds, and requires further research. However, the circulation of a wide diversity of LPAIV in wild waterbirds and high rates of viral reassortments may indicate that the levels of heterosubtypic immunity in these birds are limited.⁹³

Poultry infected with LPAIV mount humoral immune responses targeted against a variety of influenza virus proteins, including neutralizing antibodies against the HA and NA proteins.⁸⁸ These neutralizing antibodies correlate with protection against challenge with viruses of the same subtype, whether of low or of highly pathogenic pathotype. CD8+ lymphocytes may reduce viral shedding of LPAIV but their effect is less clear on HPAIV.⁸⁸ Mucosal IgA antibodies and cellular immunity likely play essential roles in protecting poultry against re-infection, including with avian influenza viruses of different subtypes, and call for further research. In particular, infection of the respiratory tract and not of the intestinal tract may lead to the development of stronger immune responses and immune memory than in wild waterbirds. Pre-existing humoral immunity in poultry may drive evolutionary changes in avian influenza viruses in these species, contributing to antigenic drift.^84,94 Interplay between adaptive changes of avian influenza viruses to poultry hosts, escape from herd immunity and possibly some level of heterosubtypic immunity may contribute to the emergence of diverse lineages within a small number of introduced subtypes and to their maintenance in these species.^93,95

In humans, innate and adaptive immune responses following infection with avian influenza viruses contribute to clearance of and recovery from infection.⁸⁷ They have also been associated with severity of the disease. Elevated levels of cytokines, including IFN-α/β and TNF-α in fatal cases of HPAIV H5N1 infection in humans, have been associated with high viral loads.⁹⁶ The NS1 protein of HPAIV H5N1 confers resistance against the antiviral effects of IFN-α, IFN-γ and TNF-α in vitro.⁹⁷ Viral replication resulting in high HPAIV H5N1 titers and high levels of pro-inflammatory cytokines in the lower respiratory tract may induce sufficient damage to the alveoli for viral entry into the bloodstream, potentially favoring systemic viral spread and infection of extra-respiratory organs. Because of the nature of the target tissues for replication—the deeper regions of the respiratory tract—avian influenza virus infection may induce severe damage as a result of both the destruction of infected cells, and host immune responses to infection, such as edema and infiltration of inflammatory and immune cells.⁵⁶

In humans, pre-existing protective immunity is conferred by immunoglobulins lining the respiratory tract (mostly long-lived IgG) and by memory B- and T-lymphocytes, induced in response to past influenza virus infections.^98-100 Neutralizing IgG are typically highly strain-specific and target the antigenic sites of the HA protein. They drive the antigenic drift of influenza viruses circulating in human populations, contributing to the emergence and maintenance of new variants within a single subtype.¹⁰¹ Antibodies directed against the NA protein have also been shown to confer protection, but to what extent NA specific antibody response drives a similar mechanism of antigenic drift has less well been studied.¹⁰⁸ Neutralizing antibodies are little cross-reactive and fail to recognize antigenically distinct HA proteins, such as those harbored by avian influenza viruses. Such escape from pre-existing humoral immunity likely favors their cross-species transmission. Past pandemics have resulted from the introduction of influenza viruses against which the human population had no or little pre-existing humoral immunity, leading to antigenic shifts.² However, antibodies against more conserved regions of the HA protein or of other viral proteins may confer protection against antigenically more distant influenza viruses.¹⁰² Likewise, the other arm of the immune system, and in particular memory T-lymphocytes, such as CD8+ cytotoxic T-lymphocytes, may not only confer subtypic, but also heterosubtypic immunity. These cells contribute to increased and more rapid clearance upon infection with influenza viruses, including those of different subtypes.¹⁰³ Heterosubtypic immunity may confer protection against avian influenza viruses upon cross-species transmission. The effect of pre-existing immunity and immunological memory based on both cellular and humoral arms of the immune system targeting conserved viral features, on cross-species transmission of avian influenza viruses is not well understood.^103,104 Yet it may have a significant impact on the outcome of infection and further adaptation of avian influenza viruses in humans following cross-species transmission.

Conclusions

Cross-species transmission of avian influenza viruses has resulted in the circulation of a wide diversity of viruses in a wide range of avian and mammalian hosts. However, adaptation resulting in circulation of influenza viruses independently of wild waterbird reservoirs are rare events, given the relatively low number of adapted strains and subtypes in other avian and mammalian hosts. So far, human infection with avian influenza viruses has resulted from sporadic cross-species transmission of the viruses from birds to humans. In the past decade, reports on such zoonotic transmissions have increased dramatically, which is most likely due to dramatic growth and industrialization of poultry farming.¹⁰⁵ It should however be noted that also among health authorities and the public at large it is increasingly appreciated that events of zoonotic transmission of avian influenza viruses may be the prelude of a devastating influenza pandemic. Yet for this to happen avian influenza viruses that have crossed from birds to humans have to adapt in order to circulate efficiently in the human population. Remarkably, cross-species transmission of avian influenza viruses to different host species is based on different modes of transmission, and leads to different disease manifestations, based on different pathogenetic pathways (Table 2). For example, while on the one hand waterbird ecology and tissue morphology (e.g., distribution of cellular receptors) may favor fecal-oral transmission of avian influenza viruses with a preferential tropism for the intestinal tract, on the other hand, the ecology and tissue morphology of terrestrial birds, humans and other mammals may favor respiratory transmission of viruses with a preferential tropism for the respiratory tract. Differences in organ tropism and pathogenesis lead to different patterns of lesion distribution and severity. Together, differences in pathogenesis and associated lesions may impact on the ability of avian influenza viruses to further spread among different host species, largely depending on population structure and ecology. Therefore, the migratory behavior of wild waterbirds may favor influenza viruses that cause minimal intestinal lesions, while current practices in poultry production likely allow the emergence and maintenance of highly pathogenic strains in terrestrial poultry. Lastly, different immune responses are likely triggered upon infection in wild birds, poultry and mammals, including humans, which will further affect virus evolution and maintenance. While intestinal infection of wild birds may elicit a limited immune response, allowing co-circulation of and co-infection with diverse influenza virus subtypes, respiratory tract infection of mammals results in a vigorous and effective immune response. Although the impact of humoral immunity against the HA protein of influenza viruses in humans has been studied in significant detail, little is known about the impact of humoral and cellular, subtype and heterosubtype specific immunity on cross-species transmission and subsequent adaptation of avian influenza viruses. Integrated interdisciplinary research efforts are therefore needed to unravel the complex interaction of virus, host, ecology and environment determined factors, that drive interspecies transmission and adaptation toward sustained circulation of avian influenza viruses in mammalian species, since this may lead to the next influenza pandemic in humans with potentially devastating consequences.

Table 2. Summary of the pathogenesis characteristics of avian influenza virus infection in wild waterbirds, terrestrial poultry and humans

	Route of entry and transmission	Tissue tropism	Pathology and host responses	Immunity
Wild waterbirds	Fecal-oral (LPAIV) Oral and/or respiratory (HPAIV H5N1)	Intestinal tract (LPAIV) Respiratory tract, brain and other organs except the intestinal tract (HPAIV H5N1)	No or very mild degenerative lesions (LPAIV) From mild to severe lesions of necrosis and inflammation (HPAIV H5N1)	Low and transient humoral immunity Cellular immunity?
Terrestrial poultry	Fecal-oral and respiratory	Respiratory tract and possibly reproductive and urinary tracts (LPAIV) Respiratory tract, endothelium and other organs (HPAIV)	Mild lesions of necrosis and inflammation (LPAIV) Severe lesions of necrosis, inflammation, hemorrhage, edema (HPAIV)	Protective homosubtypic humoral immunity Cellular immunity?
Humans	Respiratory Ocular (H7) Potentially oral (HPAIV H5N1)	Respiratory tract Respiratory tract and other organs (HPAIV H5N1)	Mild to severe lesions of necrosis and inflammation	Protective humoral and cellular homosubtypic and heterosubtypic immunity

LPAIV, low pathogenic avian influenza viruses; HPAIV, highly pathogenic avian influenza viruses; H7, avian influenza viruses of the H7 subtype

Abbreviations:
LPAIV	=	low pathogenic avian influenza viruses
HPAIV	=	highly pathogenic avian influenza viruses
HA	=	hemagglutinin
NA	=	neuraminidase