Background
The influenza virus is a common cause of respiratory infection all over the world. The influenza A virus can infect not only humans but also avian, swine, and equine species. The virus has a negative single-stranded RNA with eight gene segments, namely PB2, PB1, PA, HA, NP, NA, M, and NS. The subtype of influenza A virus is determined by the antigenicity of two surface glycoproteins, hemaglutinin (HA) and neuraminidase (NA). The subtypes currently circulating in the human population are H1N1 and H3N2. Influenza A viruses cause epidemics and pandemics by antigenic drift and antigenic shift, respectively [
1]. Antigenic drift is an accumulation of point mutations leading minor and gradual antigenic changes. Antigenic shift involves major antigenic changes by introduction of new HA and/or NA subtype into human population.
All known HA and NA subtypes are maintained in avian species, and all mammalian influenza A viruses are thought to be derived from the avian influenza A virus pool [
1]. In avian species, influenza A viruses are in an evolutionary stasis [
1]. In contrast, all gene segments of mammalian viruses continue to accumulate amino acid substitutions [
1]. Today, the emergence of an influenza pandemic is of great global concern. If a novel influenza A subtype acquires the ability to spread between humans efficiently, it could cause the next pandemic [
1]. This ability is acquired by reassortment between human and non-human influenza A viruses or by the accumulation of mutations in the non-human influenza virus. It is necessary to understand the evolutionary processes of influenza A viruses in various hosts so that we have better knowledge about the emergence of this pandemic virus. We conducted the present study to investigate the evolution of the M gene among different species. Although there are numerous studies on the evolution of the HA gene [
2‐
7], only a few studies on the evolution of the M gene have been conducted [
8].
The M gene is intriguing because it encodes both matrix and membrane proteins, and has multiple functions. The M gene (1027 bps) encodes two proteins, namely M1 (at nucleotide position 26 to 784) and M2 (at nucleotide position 26 to 51 and 740 to 1007) [
9]. M1 is a matrix protein that lies just beneath the viral envelope in the form of dimers and interacts with viral ribonucleoprotein (vRNP) complex, forming a bridge between the inner core components and the membrane proteins [
10‐
13]. vRNPs harbor the determinants for host range [
1,
14,
15]. M1 contacts with both viral RNA and NP, promoting the formation of RNP complexes and causing the dissociation of RNP from the nuclear matrix [
16‐
21]. M1 plays a vital role in assembly by recruiting the viral components to the site of assembly and essential role in the budding process including formation of viral particles [
22,
23]. M2 is a membrane protein which is inserted into the viral envelope and projects from the surface of the virus as tetramers [
24,
25]. The M2 protein comprises 97 amino acids – 24 in the extracellular domain, 19 in the transmembrane domain, and 54 in the cytoplasmic domain. Extracellular domain of M2 is recognized by hosts' immune system [
26‐
28]. Transmembrane domain of M2 has ion channel activity, which involved in uncoating process of the virus in cell [
29]. Amantadine inhibits virus replication by blocking the acid-activated ion channel. The cytoplasmic domain of M2 interacts with M1 and is required for genome packaging and formation of virus particles [
30‐
36].
The molecular mechanism of how the host range of influenza A viruses is determined is still not fully understood. The M gene may be involved in determining host tropism. Besides, novel vaccines targeting M1 or M2 proteins to confer cross-subtype protection have been shown to be promising [
37‐
43]. Therefore, understanding of evolution of the M gene is of great importance and practical relevance.
Discussion
The phylogenetic tree showed that the M gene of influenza A viruses has evolved independently in each host. It revealed host-specific lineages, which were compatible with other reports. In previous reports, Av1, Av2, Sw1, Sw2, and CE were named as Eurasian (Old World) avian, North American (New World) avian, classic (old) swine, European (avian-like) swine, and recent (avian-like) canine lineages, respectively [
1,
8,
46,
47]. Since the emergence of the Russian Flu, both H1N1 and H3N2 have been co-circulating in human populations and undergoing different evolutionary processes, which have resulted in two distinct human influenza lineages, Hu1 and Hu2 (Figure
1A, B, and
1F). Although reassortment of human influenza A viruses between the same subtype (intratypic recombination) has occurred frequently [
48‐
51], we found only a few strains that seemed to be generated by reassortment between H1N1 and H3N2 human influenza, including H1N2 strains. These strains were not maintained in human populations. When the H3N2 virus with the M gene in Hu1 acquires the M gene from H1N1 in Hu2, such a virus might not replicate and/or transmit effectively. On the other hand, M genes of avian influenza are frequently shifted between subtypes as shown in Figure
1A and
1B. This suggests that reassortment between subtypes (intertypic recombination) is common in avian influenza. This result is compatible with the study by Dugan et al., which showed a high rate of gene reassortment among avian influenza A viruses [
52]. It is still unclear why the M gene of avian influenza is interchangeable among subtypes, while the M gene of human influenza is not. Further experiments in vitro are necessary to answer this question.
After Spanish Flu, the same M gene has been maintained in human influenza, even after two pandemics (Asian Flu and Hong Kong Flu) that were thought to have been generated by reassortment between avian and human influenza A viruses [
1] (Figure
1A and
1C). In the phylogenetic tree (Figure
1A), Spanish Flu is located at the root of a human lineage and close to a swine lineage; there is a greater distance between Spanish Flu and the avian influenza A viruses identified around 1918. This result supports the hypothesis that an ancestral virus of Spanish Flu had entered the mammalian population before 1918 [
53,
54]. It remains to be seen whether this M gene will be retained after further pandemics. It was shown that the M gene of recent human influenza cannot incorporate the HA segment of avian influenza in vitro [
55].
There have been several sporadic infections with viruses from non-human lineages to humans, including the recent H5N1 infections in humans. However, these viruses were not maintained, and therefore, they disappeared from the human population without efficient transmission from human to human. In addition, it is implied that swine can be a "mixing vessel" in which human and avian viruses are reassorted to generate a human pandemic strain [
1,
56]. However, infections of strains with avian or human M genes in swine were also rare, and most of these viruses were not maintained in the swine population, except for the Sw2 lineage, in which viruses with the avian lineage M gene became established in the swine population.
Our phylogenetic analysis showed that viruses were clustered in host-specific lineages. This suggests that the M gene may be host specific and viruses with an M gene from other hosts are difficult to replicate. It is possible that the M gene determines the host range through the interaction between M1 and vRNPs [
13,
14,
57]. An M gene that can match with host-specific vRNPs may be needed to replicate and transmit in a certain host. In addition, many studies have shown the interaction between M1 protein and host proteins, such as RACK1, MAPK, and core histone [
13,
58‐
60]. The M gene may be directly and/or indirectly linked to host tropism of the virus.
The evolutionary rate of the M gene was low in avian viruses compared to human and swine viruses (Figure
2 and Table
2). This result is rational because birds are considered to be a natural host for the influenza A virus [
1]. The avian influenza A virus may have already been adapted to the host and not subject to pressure to induce further amino acid changes. This is also supported by the result showing that ω of the M gene was the lowest in avian influenza (Figure
3). Additional amino acid changes might be required in mammalian hosts to allow the viruses to adapt to these relatively new hosts. This stronger selective pressure on human and swine influenza may make human and swine influenza evolve more rapidly than avian influenza (Figures
2 and
3).
Interestingly, evolutionary rates were significantly different between lineages of the same host (Table
2). The evolutionary rates of Hu2 and Sw2 were faster than Hu1 and Sw1, respectively. The evolution of the M gene might not only be controlled by host species. One possible explanation is that strains in a lineage that appeared more recently such as Hu2 or Sw2, have to evolve more rapidly in order to be adapted better to the host than strains in other pre-existing lineages (Hu1 or Sw1), which have already adapted to some extent. Social factors at the time when new lineages appeared such as the growth of the population and globalization may also facilitate a faster evolution. This may be the reason why the evolutionary rates of Hu2 and Sw2 are higher than those of Hu1 and Sw1, respectively (Figure
2). However, reason of difference between evolutionary rates of Av1 and Av2 is unclear.
The selective pressure is stronger in M2 than in M1 (Figure
3) and more sites under positive selection were identified in M2 than in M1 (Table
3 and Figure
7). Among them, most of the sites (7 out of 10) under positive selection in M2 are located in the extracellular domain (Table
3 and Figure
7). Infection of influenza A virus induces the host's immune response to M2, especially to the extracellular domain [
26‐
28]. It has been shown that antibodies recognizing the extracellular domain including the sites under positive selection confer protective immunity [
37‐
39]. The host's immune response may make stronger selective pressure on M2 than that on M1. However, of course, selective pressure is much higher in the HA segment, the major antigenic component, than in the M2 gene [
61], and this M2 gene is thus more conserved than the HA gene [
42].
M1 is thought to play a vital role in the assembly and budding process [
12,
22,
23]. Even minor mutations in M1 may cause a critical deficiency in virus replication. This could also explain why M1 is under strong negative pressure and why the selective pressure on M1 is smaller than that on M2 (Figure
3). Nevertheless, the selective pressure on M1 of the human influenza was stronger than that of the avian influenza (Figure
3). M1 of human influenza should be under stronger selective pressure than that of avian influenza to be better adapted.
Position 219 in M1 is under positive selection in human influenza. It was also reported that this site was positively selected using a different method of calculation [
62]. However, this site is under negative selection in avian influenza (Figure
7). M1 is recognized by cytotoxic T cells [
40,
63,
64] and the C-terminal of M1 determines antigenicity [
65,
66]. The site, located at the edge of structure (Figure
5), is part of the T-cell and MHC epitope. M1 may also be under selective pressure from the host's immune response, although this is weaker than M2. Besides, the C-terminal of M1 is important for binding to vRNPs [
16]. This site might play an important role in the interaction with vRNPs, being associated with host range. Therefore, it is under positive selection only in the human and not in avian influenza virus.
Positions 115 and 121 in M1, which are under significant negative selection in both human and avian influenza, had different consensus amino acids between these two hosts (Figure
7). These results indicate that these sites may be important for host tropism and are therefore under negative selection. In addition, position 137 also has different consensus amino acids between the hosts, though this site is not under significant negative selection in human influenza (the site is under negative selection in avian influenza). The two domains in M1 have been reported to affect the disposition of viral RNA. One domain resides in a palindromic stretch of basic amino acids (position 101 to 105) [
17,
18] and the other domain is located at position 148 to 162 containing a zinc finger motif [
19,
20]. The three sites (positions 115, 121, and 137) are located between these two domains. These sites might affect the disposition of viral RNA and be involved in the determination of host range.
Position 27, which is a site in the transmembrane domain, is positively selected in M2. This site is associated with amantadine resistance [
67]. The selective pressure on the site may be due to drug pressure. However, we could not show any positive pressure on position 31, which is associated with the recent spread of amantadine resistance [
68]. Details on drug pressure and possible mechanism for recent surge of amantadine-resistant strains will be described in another manuscript (in preparation).
The cytoplasmic domain of M2 is important for interaction with M1, genome packaging, and formation of virus particles [
33‐
36]. Two sites are under positive selection in the cytoplasmic domain of M2 (positions 57 and 89, Table
3). In particular, position 57 showed different consensus amino acids between human and avian influenza (Figure
7). These results indicate that the amino acids in these sites have frequently changed, and these sites are likely to be involved in several functions of M2. The M2 cytoplasmic tail (position 45 to 69) has been shown to be a binding domain for M1 [
35]. Position 82 to 89 is important for infectious virus production [
35]. Another study showed that vRNP packaging is mediated by amino acids at position 70 to 89 of the M2 gene [
69]. The M2 gene must, therefore, have evolved with several functions.
In conclusion, the M gene of the influenza A virus has evolved with different selective pressures on M1 and M2 among different hosts. We found potentially important sites that may be related to host tropism and immune responses. These sites may be important for evolutionary processes in different hosts and host adaptation. However, Dunham et al. concluded that it is difficult to predict what specific genetic changes are needed for mammalian adaptation by comparing evolution of avian and swine influenza A viruses [
47]. Further studies to clarify the specific role of each site identified in the present study are needed.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
YF carried out all analyses and drafted the manuscript. AS, TK, and HO participated in the design of the study and helped to draft the manuscript. All authors have read and approved the final manuscript.