Influenza A viruses infect a wide variety of animal species including humans, horses, pigs, dogs, sea mammals and birds. All 16 haemagglutinin (HA) and nine neuraminidase (NA) subtypes of influenza A virus have been isolated from aquatic birds (Alexander, 2000; Fouchier et al., 2005; Webster et al., 1992), which are thought to be the primary reservoirs for all subtypes of influenza A viruses from which novel viruses can emerge and infect other animal species (Webster et al., 1992). Influenza A virus, a negative-strand RNA virus in the family Orthomyxoviridae, contains eight RNA segments encoding ten or 11 proteins. The segmented nature of the influenza viral genome provides opportunity for reassortment when two (or more) different influenza viruses infect the same cell or host. Although only three subtypes (H1N1, H3N2 and H1N2) of influenza A viruses are consistently isolated from pigs worldwide, pigs are known to be susceptible to infection with many subtypes of influenza A viruses (Kida et al., 1994). Because naturally occurring reassortant viruses derived from different host species have been recovered from pigs, they have been considered to be a ‘mixing vessel’, supporting potential influenza virus reassortment (Scholtissek, 1994). Involvement of pigs in interspecies transmission of influenza A viruses between birds and humans is at least partially due to their respiratory tract lining epithelial cells having receptors for both avian and human influenza viruses (Ito et al., 1998). The recent isolation of a unique H2N3 virus from pigs provided direct evidence that swine are able to act as a natural host from which novel influenza viruses may emerge (Ma et al., 2007).

The first swine influenza virus (SIV) isolate belonging to the H1N1 subtype in the USA was identified in 1930 (Shope, 1931); subsequently, a similar virus was isolated from humans (Smith et al., 1933). This H1N1 swine virus and closely related viruses are designated classical H1N1 (cH1N1) viruses and currently still circulate in swine populations worldwide. With the exception of one isolation of human H3N2 virus from pigs in Colorado in 1977 (Karasin et al., 2000b), only the cH1N1 virus was isolated from US swine prior to 1998 (Olsen, 2002). In August 1998, a novel subtype H3N2 virus (double-reassortant H3N2 virus) was isolated from pigs in North Carolina that contained HA, NA and polymerase basic 1 (PB1) genes from human influenza virus origin and the remaining five genes from the cH1N1 SIV origin (Zhou et al., 1999). Subsequent H3N2 isolates from other states were triple-reassortant viruses containing HA, NA and PB1 genes from human influenza viruses, matrix (M), non-structural (NS) and nucleoprotein (NP) genes from classical swine viruses, and polymerase acidic (PA) and PB2 genes from avian viruses (Webby et al., 2000; Zhou et al., 1999). Subsequent reassortments occurred between the H3N2 and the cH1N1 subtypes producing H1N2 (Karasin et al., 2000a), reassortant H1N1 (rH1N1) (Webby et al., 2004) and H3N1 viruses (Lekcharoensuk et al., 2006; Ma et al., 2006). At present, the H3N2, rH1N1 and H1N2 viruses have become endemic, and co-circulate in most major swine-producing regions of both the USA and Canada (Choi et al., 2002; Richt et al., 2003; Webby et al., 2004). More recently, the introduction of human-like H1 viruses that are genetically and antigenically distinct from the classical swine H1 lineage was identified in pigs in both Canada and the USA (Karasin et al., 2006), and these viruses have spread among US swine herds (Vincent et al., 2009). All the successful reassortant SIVs that have become endemic in US swine herds to date contain similar internal genes representing the PA and PB2 genes of avian origin, NS, NP and M genes of classical swine origin, and the PB1 gene of human origin; this has been referred to as the triple-reassortant internal gene (TRIG) cassette (Ma et al., 2009; Vincent et al., 2008). This consistent finding suggests that the TRIG cassette can accept multiple HA and NA types and may contribute to a selective advantage over other influenza viruses infecting swine. The 2009 human pandemic H1N1 virus is a reassortant between North American triple-reassortant and Eurasian SIVs and has also been found in pigs in over 20 countries including the USA. Interestingly, this human virus has a modified TRIG, which contains a new M gene of the Eurasian swine lineage (Garten et al., 2009; Smith et al., 2009). To date, no studies have been conducted on the role of the TRIG cassette in the production of current endemic SIVs in North American swine herds and how this TRIG cassette affects virus replication and adaptation in pigs. In this study, we used the triple-reassortant prototype, A/Swine/Texas/4199-2/98 (H3N2) (Tx/98), and A/Swine/Iowa/15/1930 (cH1N1) (IA/30) SIVs to co-infect pigs that were then exposed to a contact group of pigs, which was subsequently exposed to a second group of pigs, to investigate reassortment between these viruses and to characterize which constellation of viral genes is critical for virus replication and efficient maintenance and transmission in pigs.

Since the first triple-reassortant H3N2 SIV was isolated from US swine populations in 1998, all triple-reassortant SIVs including H1N1, H1N2 and H3N2 subtypes circulating in North American swine herds have contained a similar TRIG cassette (Vincent et al., 2008). In this reassortment and transmission study, four different subtypes of influenza virus (H3N2, H1N2, H3N1 and H1N1) with different genetic combinations were found in the lungs of pigs co-infected with the triple-reassortant H3N2 Tx/98 and cH1N1 rgIA/30 viruses, providing further evidence that pigs can be a ‘mixing vessel’ for different subtypes of influenza viruses. In the reassortant viruses from lungs of co-infected pigs, most of the internal genes (PB1, PB2, PA, NP and NS) were from the triple-reassortant Tx/98 virus whilst the surface genes (HA and NA) were from the cH1N1 rgIA/30 virus; the M genes were derived almost equally from both initial viruses. Most viruses isolated from the lungs of co-infected pigs had the cH1N1 surface proteins with the original TRIG cassette, confirming that this original TRIG cassette can accept and support different combinations of HA and NA genes, as in the field. Noticeably, some viruses from the lungs of co-infected pigs also contained the modified TRIG cassette with the IA/30 M gene, suggesting that the modified TRIG cassette is also able to accept different HA and NA types. The appearance of the human 2009 pandemic swine-origin H1N1 virus further confirmed this point, as this virus also contains a modified TRIG cassette with the M gene from the Eurasian SIV (Garten et al., 2009). The significance of the TRIG cassette for virus replication and transmission as well as adaptation between human and swine hosts needs to be investigated in future studies.

The polymerase subunits PB2, PB1 and PA are central to the replication cycle of influenza virus and are required for viral RNA replication and transcription. Viral-like reporter replication (the functional ribonucleoproteins constituted by co-transfecting an expressing viral-like reporter RNA plasmid with four plasmids expressing PB1, PB2, PA and NP into COS-1 cells) was more efficient when PB2 and NP were both derived from the same avian or human virus or when PB1 was derived from an avian virus (Naffakh et al., 2000). However, current circulating SIVs in North America contain the avian PA and PB2, the human PB1 and the swine NP genes, similar to the pandemic H1N1 virus, and the role of the novel polymerase complex in current SIVs is still not completely understood. In this study, two double-reassortant (human/swine) H1N1 and H3N1 viruses without the avian PA and PB2 polymerase genes were isolated from the lungs of co-infected pigs. However, these viruses were not successfully transmitted from co-infected pigs to contact animals. This result coincides with the finding that the double-reassortant H3N2 virus was initially isolated in one US herd, carrying an HA gene with identical residues in critical receptor-binding regions similar to subsequently isolated triple-reassortant H3N2 viruses, and was subsequently displaced by the triple-reassortant H3N2 SIVs (Vincent et al., 2008; Webby et al., 2004). Currently, the triple-reassortant H3N2 viruses are endemic in US swine herds. This suggests that the introduction of avian PA and PB2 genes is one of the critical factors for the triple-reassortant SIVs becoming well established in pigs (Webby et al., 2000). In this study, more than 70 % of isolates from the lungs of co-infected pigs contained this novel polymerase complex (avian PA and PB2 and human PB1 genes). This novel polymerase complex is also found in currently circulating triple-reassortant SIVs, indicating its importance in virus replication and adaptation; this might explain why the triple-reassortant virus containing the novel polymerase complex successfully spread in the pig population. Previous studies have shown that the current swine viruses in North America appear to have an increased rate of antigenic drift and reassortment and have the ability to evade established herd immunity (Richt et al., 2003; Vincent et al., 2006) due to acquisition of the avian PA and PB2 genes and the human PB1 gene. The exact role of each novel polymerase gene remains unknown and needs to be investigated.

The cH1N1 IA/30 virus generated by reverse genetics (Weingartl et al., 2009) has been shown to have a pathogenicity in pigs similar to the wild-type IA/30 virus (Lekcharoensuk et al., 2005; Vincent et al., 2008). However, no parental IA/30 virus generated by reverse genetics was found in BALF samples and nasal swabs from co-infected pigs, indicating that the parental triple-reassortant H3N2 viruses found in the lungs of the co-infected animals replicated more efficiently than the parental IA/30 virus. These results confirmed our previous findings that the IA/30 virus does not shed efficiently via the noses of pigs when compared with other H1N1 SIVs (Vincent et al., 2008). The rH1N1 virus levels were fourfold higher in the lungs of co-infected pigs than the other three subtypes (H3N2, H1N2 and H3N1). Eighteen H1N1 isolates contained the TRIG or modified TRIG cassette; however, these triple-reassortant H1N1 viruses were not isolated from the nasal swab samples of primary co-infected pigs and from contact animals, and were not successfully transmitted to contact pigs, indicating that the triple-reassortant H1N1 virus did not transmit efficiently among pigs in the presence of other viruses. Although other subtypes of virus containing the same TRIG cassette as the triple-reassortant Tx/98 virus were present in the lungs, only the parental Tx/98 H3N2 virus was transmitted from primary co-infected pigs to two groups of sentinel animals and became established in the contact pigs. The seroconversion data of contact pigs on day 14 p.c. to the Tx/98 virus [haemagglutination titre (HI) titre of 640] vs the IA/30 virus (HI titre of <20) support the assumption that the Tx/98 virus had a tremendous advantage when replicating in pigs (data not shown). These results indicated that the H3N2 Tx/98 virus has an optimal genetic constellation that contributes to efficient replication and successful transmission among pigs when compared with the IA/30 and other reassortant viruses. If currently circulating cH1N1 virus other than IA/30 had been used in this study, the results may have been different, and not only the parental Tx/98 H3N2 virus but also other viruses may have been transmitted to the sentinel animals. Although the triple-reassortant H1N1, H1N2 and H3N1 subtypes have been isolated from swine herds in the field, in this particular study, the similar subtype viruses containing the TRIG cassette of the Tx/98 virus were not able to transmit efficiently and be maintained in the small groups of pigs in the presence of the H3N2 Tx/98 virus. This finding suggests that a prerequisite for an influenza virus to establish and maintain itself in pigs is the right combination of HA and NA genes as well as the TRIG cassette, which may be reflected in the field situation with North American SIVs. The current pandemic H1N1 virus may be another example – a virus that contains the modified TRIG cassette, the NA from Eurasian swine virus and the HA from the North American triple-reassortant swine virus (Garten et al., 2009). This virus has infected humans and has been transmitted effectively from humans to pigs and to other species (WHO, 2009), and has caused over 15 000 human deaths worldwide as of February 2010 (WHO, 2010). The virulence and transmission capacity of this virus suggest that the genetic constellation of the pandemic H1N1 virus is optimal for its effective replication, transmission and adaptation to different hosts. In this study, only the Tx/98 virus was successfully transmitted and established in pigs, even though other subtypes of virus containing a similar TRIG cassette were produced simultaneously. These results indicate that reassortment will occur when pigs are infected with different viruses, but only the virus with an optimal gene constellation will have the opportunity to establish itself in herds under similar selective pressure. Therefore, the old dogma of the pig as a ‘mixing vessel’ needs to be readdressed. In a scenario where efficient reassortment occurs in the lung, only the fittest virus will be maintained in a pig population.