Influenza Virus Reservoirs and Intermediate Hosts: Dogs, Horses, and New Possibilities for Influenza Virus Exposure of Humans

The inventory of mammalian hosts that are known to sustain transmission of IAVs has increased in the past 15 years due to the emergence of two strains of the virus in dogs: the H3N8 equine influenza virus (EIV) that transferred to dogs in the United States in about 1999 and the avian virus-like H3N2 strain that transferred to dogs in Asia around 2005 or 2006. Both of these canine influenza viruses (CIVs) have continuously circulated in the dog population since they emerged, creating many opportunities for exposure in humans and other species.

Horses have been natural hosts for sustained influenza transmission on at least three occasions, with an outbreak of an unknown equine influenza virus (EIV) subtype occurring in 1872 (20), an H7N7 EIV that emerged around 1956 (21), and an H3N8 EIV that was recognized around 1963 when there was an outbreak of respiratory disease in Florida that occurred soon after the importation of horses from Argentina (22). The H3N8 EIV appears to have emerged from an avian influenza virus that spread to horses, and there was some reassortment of that virus with the H7N7 viruses in the 1960s and 1970s that eventually died out during the 1970s (23). Limited transmission of H3N8 EIV to dogs, including in the United Kingdom (2002) (24) and Australia (2007) (25), has been observed on a number of independent occasions. However, despite these spillover events, only one EIV transfer that occurred around 1999 in the United States resulted in the establishment of H3N8 canine influenza virus in dogs in Florida, which involved the transmission of a single EIV strain from horses. This outbreak was first described in 2004 as the cause of disease among racing greyhounds, and the virus was transferred within a year to racing tracks in many parts of the United States (26) and, subsequently, to other dog populations and breeds where it has been maintained ever since (27, 28). Notably, H3N8 CIV appears to have been primarily maintained since about 2008 in animal shelters in a few large cities where dogs live in dense populations and where susceptible animals are introduced at high rates (27, 29, 30). Because infected dogs are frequently transferred from these shelters, the virus is often passed to other dog populations where it causes outbreaks but where it is seemingly unable to sustain long-term transmission, likely because of overly heterogeneous contact networks (27, 29). In particular, although there are an estimated 80 million susceptible household dogs in the United States, they appear to lack the degree of contact necessary for continuous transmission of influenza virus such that chains of transmissions quickly die out (29). As a result, CIV infections in small shelters or in the household dog populations die out within days or weeks (27, 29). To date, no reassortment by the H3N8 CIV with any other human or animal influenza viruses has been detected.

It is therefore clear that horses acted as an intermediate host for the emergence of the canine H3N8 strain, although only after about 37 years of continuous circulation of the virus, suggesting that there are major adaptive or epidemiological barriers to CIV emergence. Since its emergence, the H3N8 CIV lineage has diverged from the EIVs that have continued to circulate in horses, with distinct amino acid substitutions in each gene segment, some of which may have been selected for canine adaptation (30, 31). For example, certain amino acid substitutions in the sialic acid binding site modify the replication of the virus in the respiratory tract of the dog by altering the binding to the sulfated glycans and likely by other changes to sialic acid binding (32). Those EIV strains that circulated at the time of emergence of CIV appear to readily infect dogs after experimental challenge or under conditions in which dogs are housed with EIV-infected horses (33, 34) and can infect canine tracheal explant cultures (35), so the amount of adaptation required to infect dogs with these viruses is likely small. Notably, there is currently no evidence for transfer of a CIV back to horses in nature, and CIV replicates inefficiently when used to challenge horses or horse tracheal cells (36–38). Besides the infection in dogs, the H3N8 influenza virus has also been reported to infect pigs in China (39) and to infect cats in experimental studies (40).

The H3N2 CIV most likely arose around 2005 or 2006 in China but was first recognized in South Korea around the same time, and the virus or specific antibodies have since been detected in many parts of China and Korea, as well as Thailand (41–44). The epidemiology of the H3N2 CIV is still poorly understood, but some reports indicate moderate seroprevalence in some household dog populations, and it has been transferred over long distances both within and between the three countries in which has been detected (45). Populations of farmed dogs may play a role in the maintenance of the virus (46). H3N2 CIV is an avian virus-like strain, and all isolates share a single common ancestor, but the circumstances of the transfer are unknown, and its gene segments are related to both Eurasian and American avian influenza virus lineages (42, 47). H3N2 CIV appears to have a relatively broad host range, and the virus naturally infects cats and spreads among them under some circumstances (48). Under experimental conditions, H3N2 CIV has also been shown to infect ferrets, albeit with limited transmission (49, 50).

Despite its recent emergence, reassortants between H3N2 CIV and other influenza viruses, including a CIV carrying the M segment of human influenza virus H1N1, have been described (51). Another CIV has been suggested to harbor the nucleoprotein (NP) segment of an avian influenza virus related to H5N1 viruses (46), although that is not a clear reassortant (P. R. Murcia, unpublished results). Notably, an H3N1 influenza virus carrying the H3 of CIV and the other gene segments from the H1N1 pandemic human influenza virus has also been reported (52). This should be a cause of concern, as previous influenza pandemics have originated when viruses harboring a novel HA were introduced into the human population.

There are only a small number of well-documented examples of mammalian intermediate hosts allowing virus transfer onward to new hosts or acting as “mixing vessels” facilitating reassortment leading to new outbreaks. In addition to transmission of the human H1N1 pandemic virus and the likely role of humans in the emergence of the classical H1N1 virus in swine in 1918, horses clearly played a role in the emergence of the H3N8 virus in dogs. The CIV/human virus reassortant reported for the H3N2 CIV means that we should be alert to the possible emergence of intact or reassortant CIVs in humans. Such a threat results from the extensive and sustained contacts between humans and their pet dogs, which would result in significant human exposure to CIV if the viruses were to spread widely among household dogs. It is also possible that additional reassortments between canine and human viruses could occur in dogs or humans or possibly other hosts such as cats, potentially creating a new human influenza virus for which there may be little or no natural immunity. Clearly, these risks may vary depending on the virus, with H3N2 CIV seemingly posing a greater risk that H3N8 CIV (see below). Importantly, to date there have been no documented transmissions of H3N8 EIV or CIV to humans despite the close interactions between humans and horses since the first appearance of EIV in the early 1960s or between humans and dogs since 2000 (53, 54). However, direct experimental inoculation of human volunteers with EIV in the 1960s showed that infection did occur but only at low levels (55). These results suggest that there are high intrinsic barriers to the establishment of H3N8 in humans, such that onward transmission in two mammalian hosts (horses and dogs) does not guarantee successful infection of another (humans).

In contrast, the observation of H3N2 CIV-human virus reassortants in dogs shows that coinfection of one of the hosts (likely dogs) is possible and hence a cause for concern. In addition, H3N8 EIV has been isolated from pigs in Asia (39) and from bactrian camels in Mongolia (56), and swine are susceptible to infection with many avian virus-like and human virus strains (10, 11). However, the ability of canine influenza viruses to infect pigs (and the ability of pig influenza viruses to infect dogs) has not been formally tested.

Despite the emergence and spread of the CIVs in both the United States and Asia, there may still be opportunities to control or even eradicate the viruses while they are in their reservoir hosts. Although the household dog population in the United States is about a quarter the size of the human population, the transmission of influenza virus among dogs appears to be much less efficient than that of influenza viruses among humans (27, 29). After 14 years of continuous dog-to-dog transmission, H3N8 CIV is still circulating in a sustained fashion only among dogs in a small number of large animal shelters, and the transmission rate is sufficiently low that the virus might be eradicated from those shelters by reducing the rate of transmission or by reducing the efficiency of infection (29). The spread of the H3N2 CIV in Asia may likewise also be controlled by timely interventions, particularly if the household dog population is similarly inefficient in allowing virus transmission, so that the virus is being maintained in other dog-rearing or -housing facilities.

The emergence of canine influenza viruses, like the emergence of any influenza virus lineage in a novel mammalian host population, may constitute a direct or indirect pandemic risk. However, adaptation to mammals should not be seen as a direct conduit to emergence in humans. Clearly, there are important host barriers that have yet to be understood or breached. While the lack of transmission of EIV or CIV to humans suggests that long-standing infection of a mammal does not necessarily indicate a threat to humans, a better understanding of the ecological, evolutionary, and molecular mechanisms of influenza emergence is essential to accurately determine which viruses pose a risk to human health.