This study focused on the evolutionary dynamics of the H1N2 subtype isolated from Italian swine, beginning in 1998 and continuing into 2012. To perform a phylogenetic and molecular analysis of the HA and NA genes of 53 Italian H1N2 strains, we compared their gene sequences with the Influenza Virus Resource NCBI sequence collection of swine influenza viruses and influenza viruses of human and avian origin. Our analyses revealed a clear difference between Italian strains of group A, which were closely related to the EU H1N2 SIVs, and group B, which had a different HA-NA combination. This difference was characterized by an HA that derived from an H1N2 strain that was isolated in Italy in 1998 that had two aa deletions within the receptor binding site of the HA protein (Residues 146 and 147, which are equivalent to 133 and 133A in H3 numbering), and an NA gene that was closely related to the human H3N2 viruses of 1997. To our knowledge, a single aa deletion at position 147 has been observed in 4/285 (1.6%) of H1 European SIVs sequences available in GenBank, whereas the deletion of two aa was only found in Italian strains. In addition, we identified three different groups of strains that arose from reassortment in group C, which are characterized by an avian-like H1 that is combined with the N2 of one of the following clades: EUH1N2SIVs, EU H3N2SIVs or Human H3N2SIVs. Interestingly, strains with an N2 protein that is closely related to EU H1N2SIVs apparently have not circulated in Italy since 2003, because all the strains isolated after 2003 contained NA proteins that were closely related to human H3N2 or EU H3N2SIVs.
The time-scaled phylogeny of the HA protein revealed that Italian strains could be segregated into four different groups. The majority of them was in groups A and B, and had a human-like H1 antigen, which probably originated in 1986. Group B however, formed a sub-group that was characterized by a two aa deletion in the HA protein. In contrast to A and B, the third group of 11 isolates (Group C) was characterized by an avian-like H1, which probably originated in 1996. Group D had only one strain that had an H1 characteristic of the 2009 pandemic strain. The analysis of the dated NA phylogeny suggests that the observed Italian swine strains arose from a series of reassortment events. In particular, whereas the oldest strains in group A have an NA protein that originated from human H3N2 viruses in the late 1970s, the isolates in the more recent group B, excluding one strain isolated in 1998, are characterized by the two aa deletion in the H1 protein, and a different human N2. This N2 probably resulted from reassortment with a more recent H3N2 virus, which was circulating among humans and took place in 2000 or later. The analysis of group C isolates was more complex because the NA protein of five of these strains was derived from EU swine H3N2, for one it was derived from EU swine H1N2, and for the other five, similar to the group B isolates, the N2 was derived from human H3N2. Furthermore, the five strains characterized by the avian-like H1 and swine H3N2 NA protein segregated into two groups, suggesting that at least two other reassortment events took place in 2003 and 2006. Finally for group D, the virus has the H1(2009)pdm and an NA protein derived from swine H3N2, which originated in 2003.
To better characterize the effect of the two aa deletion found in the Group B H1 gene, the 3D-MMD of the HA protein was predicted. The region of HA that contains the receptor binding residues is located at the membrane –distal tip of each monomer of the HA trimer. This binding site is flanked by three elements. The 220 and 130 loops contain amino acids that interact with sialic acid or internal sugars of the glycan chain. The 190 helix forms the membrane-distal region of the site and includes residues that have the potential to contact the sialic acid or internal glycans on the receptor . The two deleted amino acids are at position 133 and 133A (H3 numbering), which is located in the 130 loop and therefore, could affect virus interactions with cell surface receptors. In order to better investigate the effects of these interesting deletions on the receptor interactions further studies are required.
Theoretically, reassortment events between human and swine influenza viruses might frequently occur but fail to persist in the pig population . Indeed, new virus strains with different antigenic characteristics may be at an advantage or disadvantage compared to well-adapted, established viruses already circulating in pigs. Those at an advantage could persist in pigs and, following adaptation, could be associated with clinical disease . The successful transmission of Influenza A virus (IAV) depends on a specific gene constellation  and a better balanced HA-NA gene combination . HA and NA proteins both recognize sialic acid but with conflicting activities, and a balance of HA and NA protein activity is essential to ensure efficient viral replication . While a virus is adapting to a new host, possibly by improving its transmission efficiency, a functional balance between HA binding and NA enzymatic activity may occur . The emergence and persistence of group B strains suggest that a particular gene constellation and an optimum balance between HA and NA activity contributes to its efficient replication and successful transmission among pigs. Another interesting group that was formed by strains of group C with an avian-like H1 has been increasing in the last few years. Indeed out of eleven C isolates, eight were isolated in the last three years and 50% of the H1N2 strains isolated in 2011–2012 also contained this hemagglutinin variant. These data suggest that, together with the previously established group B, there is now a persistence of group C strains and both B and C have probably replaced the EU H1N2 SIVs in our country, because they have not been isolated since 2003.
Positive selection for the full length HA and NA sequences of the H3N2 and H1N1 viruses was previously reported [49–52]. However, there was a lack of detailed description of these analyses in swine influenza viruses because the two swine lineages were not differentiated. Although Li et al.  distinguished between them, the European SIVs were analysed as a unique dataset and differences between the HA protein of viruses circulating in Europe (EU H1 avian origin SIVs and EU H1 human origin SIVs) were not considered. In our study, five different datasets were taken into account: for HA - European avian-like H1 and human-like H1 and for NA – three datasets that corresponded to different clusters in the NA tree. These were the N2 of EU H1N1 SIVs, recent Italian SIVs and recent human H3N2 IVs. Swine influenza virus is considered to be under weak selection pressure by the host’s immune system  and this was confirmed by the lower ω values we observed for the different datasets (Table 2). Selection pressure analyses of Italian strains, except for Italian B and C strains with the NA gene related to human H3N2, were performed together with the EU SIVs because of the observed close genetic relationship. The NA gene of the remaining Italian strains is derived from the recent human H3N2 strains; it is found only in Italy and therefore was analyzed separately. Interestingly, for this dataset the global ω value was lower than that of human viruses and similar to the EU SIVs according to a host-specific evolution of influenza virus genes .
Because of the short average life span of pigs, swine influenza virus evolution may be determined to only a limited extent by immune pressure, which is the driving force of antigenic drift of influenza viruses in humans . Human influenza viruses require frequent antigenic changes of HA to ensure that a sufficiently large pool of immunologically susceptible hosts is available. The situation for swine influenza viruses is different due to a continuous renewal of the susceptible pig population since the major part of pigs are killed at the age of 6 months (up to 8 months in Italy) consequently limiting the increase of immune pressure. Only adult sows used for pig breeding have a long life, experiencing more than one influenza season and could create some degree of immune pressure. Influenza vaccination of pigs is applied in Europe using inactivated, bivalent vaccines, that are used mainly in gilts and sows; however this vaccination is used in a low number of breeding farms (below 30%) in Italy (G Alborali, unpublished observations). Considering the limited vaccination in Italy and the continual supply of non-immune animals, we could hypothesize that vaccination does not play an important role in the evolution of Italian swine influenza viruses.
Site-by-site analysis revealed that for both HA types, several sites were under positive selection. Some sites were located in the receptor binding site and some of them formed part of the Sa, Sb and Ca antigenic sites. For NA, the N2 of human origin was subject to the strongest positive selection. Four of the six sites identified were located in important domains (phylogenetically important regions, NA head and antibody binding domains). These results demonstrate that although the selection pressure on SIVs is weak, selection has influenced the evolution of the virus, leading to amino acid substitutions at several antigenic sites.
Continuous monitoring of the genetic content of circulating IAV in order to detect new reassortment events, and studies that define the processes involved in viral reassortment are essential if we are to understand how pandemic IAV arise. Understanding IAV evolution and adaptation to various hosts will also provide information on their ability to cross host barriers and develop into pandemic strains.