H9N2 viruses have undergone extensive reassortments to generate multiple reassortants and genotypes [1–7, 9, 13, 14]. In the present study, extensive sequence data was used to characterize the evolutionary pattern by genetic and pathogenic diversity of the H9N2 influenza viruses in China. Phylogenetic analysis indicated that both Ck/Bei-like and F98-like viruses co-circulated in China from 1998 to 2010 (Additional file 1, Figure.S1). The internal gene segments of most viruses isolated after 2008 were derived from the F98-like group, indicating that F98-like has been the dominant lineage in recent years. Phylogenetic analysis also indicated multiple sub-lineages in each gene segments were well-separated and evolved independently, which indicated that each gene segment of the H9N2 viral genome displayed a high evolutionary rate.
Multiple reassortants and an increasing number of H9N2 influenza virus genotypes have been continuously identified throughout China in recent years, although most of the viruses were transients and could not be established in avian hosts [1, 4, 5]. In the present study, all of the representative strains isolated from diseased chicken in northern China were reassortants exception Ck/SD/WF/98 (which was a pure Ck/Bei-like strain) (Table 1 and Additional file 1, Figure.S1). One novel stable and predominant lineage (genotype B65) was identified in our study. This triple-reassortant lineage possessed a G1-like M gene segment, a PB2 gene with an unknown origin, and other gene segments from F98-like viruses (Table 1 and Additional file 1, Figure.S1). Together with our 7 isolates, 13 H9N2 viruses have so far been detected in this genotype which were mainly isolated from chickens in northern and southeastern China from 2007 to 2010 (Additional file 1, Figure.S1E). Interestingly, A/swine/Yangzhou/1/2008 (isolated from swine) also belonged to this new genotype (Additional file 1, Figure.S1E), indicating that genotype B65 virus has been established in chickens in different regions of China and has spread to mammals. Other new identified genotypes were composed of one or two strains (Table 1, Figure 1). However, further sequence data collection is needed to confirm if they are transient reassortants or have been established in avian.
H9N2 influenza viruses can also generate novel reassortment or genotype viruses that carry gene segments with wide evolutionary distances. It has been reported that H9N2 strain A/swine/Korea/S190/2004 was a reassortant between a Eurasian lineage and a North American lineage viruses . In China, Ck/HLJ/35/00 has only been reported as H9N2 virus that possesses HA [Genbank: DQ064366] and NP [Genbank: DQ064447] genes from an early North American lineage (represented by Ty/WI/1/66) [1, 2]. Interestingly however, the Ck/HLJ/u/98 isolate analyzed in our study possessed each of eight gene segments with the highest homology to those of Ck/HLJ/35/00 (Additional file 1, Figure.S1). These findings suggested that the two strains might have the same origin, and this genotype was not transient rather circulated in China for several years.
The novel H9N2 reassortants or genotypes with wide evolutionary distances not only generated by hybridization among different lineages of H9N2 viruses, but also by reassortment between H9N2 and H5N1 viruses. The first reported evidence of hybridization between H9N2 and H5N1 was the case of human infected by highly pathogenic avian influenza (HPAI) H5N1 virus in 1997, and the Qa/HK/G1/97(H9N2)-like viruses were hypothesized to have been involved in the generation of the H5N1 virus . Previously isolated H9N2 influenza viruses that reassorted with H5N1 viruses could also be found elsewhere: those possessing the H5N1-like PB1 gene segments were from southeastern China , and H9N2 viruses with NS genes originating from the HPAI H5N1 lineage were isolated from Pakistan . However, the pathogenicity of these reassortants was not clear [14, 16]. Noticeably, in our study, both the PA and NP genes of Ck/SD/LY-1/08 (genotype 63) shared high homology (98%-99%) with the H5N1 viruses circulating around the year 2005, such as A/chicken/China/1/02 (H5N1) [GenBank: DQ023146], A/swine/Shandong/2/03 (H5N1) [GenBank: AY646426, AY700213], and A/environment/Qinghai/1/2008 (H5N1) [GenBank: FJ455823, FJ455825]. Interestingly, one H5N1 reassortment virus, A/plateau pika/Qinghai/04/2007 (H5N1), was closely related to Ck/SD/LY-1/08 in its PB2 (98.4% homology) [GenBank: FJ390058], PA (98.5% homology) [GenBank: FJ390060], and NP (99.6% homology) [GenBank: FJ390062] gene segments (Additional file 1, Figure.S1E, G and H). It also contained the Ck/Bei-like M and NS genes, as well as a Y439-like PB1 gene. It was demonstrated that multiple reassortments have occurred between H9N2 and H5N1 subtypes since the early outbreaks, and they continued exchanging internal gene segments and generating novel viruses.
H9N2 viruses of different genotypes and reassortment patterns could have huge differences in their pathogenicity and transmission in mammals (BALB/c mice and swine) under experimental conditions [2, 7, 9]. In the present study, the majority of the recently circulated genotype B65 and B55 viruses, which include the PB2 gene segment from an unknown origin and/or the M gene segment from the G1-like lineage, displayed low infective ability in mice (Table 3 and Figure 2). In addition, Ck/SD/LY-1/08 (genotype B63) was able to replicate well in mice lungs with high virus titer but caused mild clinical symptoms (Table 3 and Figure 2). Natural selection suggests that a less virulent strain is more likely to co-exist with the host population because mobile, living hosts will transmit the strains most effectively . Thus, whether the changes in the host adaptation and replication ability of Ck/SD/LY-1/08 are associated with the H5N1 donor of the PA and NP gene segments must be explored further. Importantly, two early strains in our study (Ck/HLJ/u/98 in genotype G2 and Ck/SD/WF/98 in genotype B0) caused the death of infected mice without prior adaptation (Figure 3), though a previous study of several H9N2 viruses belonging to the same genotypes reported non-lethality in mice . The similar discordant infectivity in mice with same genotype viruses was also observed in this study, Ck/SD/BD/10, Ck/SD/03/10 and Ck/SD/02/09 were all in the genotype B65, and the former two viruses nearly couldn't replicate in mice lungs, but the third strain were able to replicate well in mice lungs without prior adaptation. It was reported that the mutations of E627K and D701N in PB2 gene were the key factors for a virus to acquire the ability to adapt to increase virulence in a new mammalian host . Molecular analysis demonstrated that all of the viruses in our study possessed the conservative residues Glu627 and Asp701 in PB2 gene. Therefore, it is probable that the virulence of H9N2 influenza viruses is not fully relevant to whole-gene homology, and some amino-acid mutations or deletions may dramatically alter the virulence of influenza viruses.
In light of the persistent evolution of H9N2 viruses with high evolution rate, it is highly necessary to monitor the evolution and evaluate the virulence of novel avian H9N2 viruses to mammal animals, which may discover and prohibit the potential threat to human by novel viruses, and provide the key clues for preventing new pandemic influenza.