G1-like M and PB2 genes are preferentially incorporated into H7N9 progeny virions during genetic reassortment

Background Genotype S H9N2 viruses have become predominant in poultry in China since 2010. These viruses frequently donate their whole internal gene segments to other emerging influenza A subtypes such as the novel H7N9, H5N6, and H10N8 viruses. We recently reported that the PB2 and M genes of the genotype S H9N2 virus, which are derived from the G1-like virus, enhance the fitness of H5Nx and H7N9 avian influenza viruses in chickens and mice. However, whether the G1-like PB2 and M genes are preferentially incorporated into progeny virions during virus reassortment remains unclear; whether the G1-like PB2 and M genes from different subtypes are differentially incorporated into new virion progeny remains unknown. Results We conducted a reassortment experiment with the use of a H7N9 virus as the backbone and found that G1-like M/PB2 genes were preferentially incorporated in progeny virions over F/98-like M/PB2 genes. Importantly, the preference varied among G1-like M/PB2 genes of different subtypes. When competing with F/98-like M/PB2 genes during reassortment, both the M and PB2 genes from the H7N9 virus GD15 showed an advantage, whereas only the PB2 gene from the H9N2 virus CZ73 and the M gene from the H9N2 virus AH320 displayed the advantage. Conclusion Our findings highlight the preferential and variable advantages of H9N2-derived G1-like M and PB2 genes in incorporating them into H7N9 progeny virions over SH14-derived F/98-like M/PB2 genes. Supplementary Information The online version contains supplementary material available at 10.1186/s12917-021-02786-0.


Background
Widespread reassortment of H9N2 viruses in China has created various subtypes that can be phylogenetically grouped into the A-W genotypes [1]. More than one genotype may circulate simultaneously in one region. Some genotypes became dominant over a long period of time [2]. For instance, H9N2 viruses that harbour three polymerase genes and the NP gene from the F/98-like virus plus the remaining four genes from the BJ/94-like virus form a group of viruses designated as F/98-like or H genotype ( Fig. 1) [2]. F/98-like (genotype H) viruses were first identified in 1998, became dominant in 2000, and persisted for several years thereafter [2,3]. Since 2007, the genotype S viruses, which were generated through the reassortment of the F/98-like viruses (genotype H) by substituting their M and PB2 genes with those of the G1-like virus, have emerged ( Fig. 1) and gradually dominated in chicken flocks after 2010 [2][3][4][5]. Moreover, genotype S H9N2 viruses often donate some or all the six internal genes to other emerging influenza A viruses in China [2,6] such as the novel H7N9, H10N8, H7N7, and H5N6 viruses [2,[7][8][9]. That is, the H9N2 and H7N9 viruses currently circulating in China both possess G1-like M and PB2 genes [10].
The S genotype, which differs from F/98-like viruses (genotype H) only in their M and PB2 genes [11,12], has not been replaced by a new genotype since 2010 [2]. As S genotype viruses carry the genetic backbone of F/ 98-like viruses (genotype H) plus the M and PB2 genes of the G1-like viruses, it is presumed that the G1-like M and PB2 genes confer better viral fitness over F/98-like counterparts.
Indeed, Pu et al. reported that H9N2 viruses with the G1-like M gene replicate faster in primary chicken embryonic fibroblasts and chickens than do the H genotype viruses. Furthermore, the H9N2 virus with the G1-like M gene exhibit an early surge of viral mRNA and genomic RNA production, suggesting of increased fitness of the virus [13]. Consistently, Hao et al. found that H5Nx and H7N9 viruses harbouring the G1-like PB2 and M genes display better viral fitness than those with F/98like PB2 or M genes and have high virulence and replication capacity in chickens and mice [11,12].
Our present study aims to determine if G1-like M and PB2 genes hold a competitive advantage during genetic reassortment, whether they play a role in maintaining the stability of "gene cassette" in H7N9 viruses. Several representative H7N9 and H9N2 viruses were chosen to test the relative copy number of G1-like M/PB2 genes and F/98-like M/PB2 genes in reassortant viruses on the H7N9 genetic background. The TaqMan-MGB quantitative realtime PCR (qRT-PCR) approaches for accurately quantifying the heteroplasmy level of G1-like M/PB2 and F/98-like M/PB2 were introduced in our study. The MGB probes had higer melting temperature. Therefore, they are much shorter than the traditional TaqMan probes, which makes MGB probes more sensitive to single base mismatches [14,15]. Multiple studies have demonstrated that TaqMan-MGB qRT-PCR is an accurate technique with high specificity, sensitivity and remarkable reproducibility and is quite attractive for use in SNP (single nucleotide polymorphism) detection and allelic discrimination [16,17].
Our results suggests that the G1-like M and PB2 genes are more likely to be incorporated into the novel H7N9 viruses than that of SH14 virus derived F/98-like M/PB2 genes; G1-like PB2/M genes derived from different virus strains display variable competitive advantages during virus reassortment.

Results
The sensitivity and specificity of duplex real-time RT-qPCR assay We first evaluated the sensitivity of duplex RT-qPCR by using ten-fold serially diluted plasmid mixtures as the templates in the amplification reaction. As shown in Supplementary Fig. S1A & B and Fig. S2A & B, each gene could be readily amplified with approximately 10 copies of templates when crossing point was less than 35(cp < 35). The standard curves revealed excellent correlation coefficient and amplification efficacy (   1 Schematic representation of the genetic constellations of the different viruses. SH14 virus were genotype H H9N2 virus; CZ73 and AH320 viruses were genotype S H9N2 viruses; and GD15 were H7N9 virus whose internal genes all came from genotype S H9N2 viruses specificity of duplex RT-qPCR used in this study, the G1-like M/PB2 plasmids from GD15, CZ73, AH320 viruses and the F/98-like M/PB2 plasmids from genotype H H9N2 SH14 virus were used as templates for the amplification reactions. As shown in Supplementary Fig.  S4, the M/PB2 genes from GD15, CZ73 and AH320 could not be detected with SH14-M probe974 or SH14-PB2 probe713 ; whereas M/PB2 genes of SH14 virus could not be detected by using GD15-M probe974 , M probe975-17 , GD15-PB2 probe713 or PB2 probe974RC . G1-like M and PB2 genes from GD15 (H7N9) virus hold a competitive advantage during reassortment To determine whether the G1-like M segment was preferentially incorporated into reassorted virus progeny when it was in competition with F/98-like M gene, 293 T cells were co-transfected with eight plasmids of the GD15 virus plus the ninth one encoding f98 SH14-M gene (500 ng /plasmid) (Fig. 2a) [18]. After incubation for 72 h, the conditioned media of transfected cells, which contained the reassorted progeny virions, were used to inoculate into embryonated chicken eggs. Quantitative RT-PCR (qRT-PCR) analysis of allantoic fluids revealed that the copy number of the g1 GD15-M gene was significantly greater than that of the f98 SH14-M gene (Fig.  2b). However, when just 250 ng of the g1 GD15-M plasmid was repeated for the above mentioned 9plasmid transfection, approximate level of gene copies was displayed between g1 GD15-M and f98 SH14-M (Fig. 2c).
We next determined whether the G1-like PB2 gene also exhibited competitive advantage during genetic reassortment. Co-transfection experiments with the nine-plasmid system were similarly carried out as above. As shown in Fig. 2d, the copy number of the g1 GD15-PB2 gene was significantly higher than that of the   GD15-PB2 and f98 SH14-PB2 genes in virus population expanded from the contransfection. 293 T cells were cotransfected with nine plasmids, 500 ng (E) and 250 ng (F) of g1 GD15-PB2 plasmid was used respectively. Supernatants were collected and inoculated into eggs. qRT-PCR was performed to determine the numbers of g1 GD15-M and f98 SH14-M genes in allantoic fluid. Data are represented as mean ± SD (N = 8). The statistically significant differences were analyzed by ANOVA (*P < 0.05; **P < 0.01; ***P < 0.001) gene from GD15 virus was more prominent than that of the G1-like M gene.

Variable advantage of G1-like M/PB2 genes from different strains during reassortment
We next investigated if the competitive advantage of the M and PB2 genes from G1-like viruses in reassortment was strain-specific. The G1-like M and PB2 genes from two additional S genotype H9N2 viruses, A/Chicken/ Jiangsu/CZ73/2014 (CZ73) and A/Chicken/Anhui/ AH320/2016 (AH320), were used in co-transfection experiments as described above. As shown in Fig. 3a, b, the copy number of the g1 CZ73-PB2 gene was significantly higher than that of the f98 SH14-PB2 gene in the rescued viruses (Fig. 3c). However, there was no significant difference in the copy number of the M gene in reassortant viruses rescued from co-transfection with the g1 CZ73-M and f98 SH14-M genes (Fig. 3d). We then investigated the effect of the internal gene cassette of the CZ73 virus on virus reassortment. Co-transfections with six plasmids encoding the internal genes of CZ73 virus plus the plasmid encoding the f98 SH14-PB2 or f98 SH14-M (500 ng/ plasmid) (Fig. 3e, f) revealed that the copy number of g1 CZ73-PB2 genes was significantly higher than that of f98 SH14-PB2 gene (Fig. 3g), although g1 CZ73-M genes did not exhibit any competitive advantages (Fig. 3h).
We then conducted a similar co-transfection experiment by using the M or PB2 gene of another S genotype H9N2 virus, AH320 (Fig. 3i, k). qPCR analysis revealed that the copy number of the g1 AH320-M gene in the rescued viruses was significantly higher than that of the f98 SH14-M gene (Fig. 3j). The copy number of the g1 AH320-PB2 gene in rescued viruses was higher than that of the f98 SH14-PB2 gene. However, this was not statistically significant (Fig. 3l).
Lack of competitive advantage for the g1 GD15-M gene incorporation into the H7N9 background over f98 SH14-M gene during co-infection We next determined if the competitive advantage of G1like M and PB2 genes during reassortment was in part Data are represented as mean ± SD (N = 6). The statistically significant differences were analyzed by ANOVA (*P < 0.05; **P < 0.01; ***P < 0.001) due to faster replication of newly reassorted viruses. We generated four recombinant viruses (WT-GD15, GD15-M SH14 , GD15-PB2 SH14 and GD15-M SH14 PB2 SH14 ) ( Fig. 4a-d). As shown in Fig. 4e, the titers of WT-GD15 virus at 48 h post infection (h.p.i) were significantly higher than three recombinant viruses, which gave similar virus titers in the conditioned media. Quantitative RT-PCR analysing the M gene revealed almost equal g1 GD15-M and f98 SH14-M vRNA levels at 48 hpi (Table 1). We then co-infected MDCK cells with GD15-M SH14 and WT-GD15 viruses, each with 0.01 MOI (Fig. 5a). Again, quantitative RT-PCR revealed that the levels of the g1 GD15-M and f98 SH14-M genes in the conditioned media were not significantly different (Fig.  5b). To further investigate whether PB2 genes affected the competitive advantage of M genes at the virus level, we analysed the portion of g1 GD15-M gene in progeny virions by co-infecting MDCK cells with recombinant GD15-PB2 SH14 and GD15-M SH14 PB2 SH14 viruses (Fig.  5c). However, the copy number of g1 GD15-M genes in progeny virions did not demonstrate significantly advantage over the f98 SH14-M gene (Fig. 5d).
Preference for the g1 GD15-PB2 gene incorporation into the H7N9 background over f98 SH14-PB2 gene during coinfection MDCK cells were co-infected with GD15-PB2 SH14 and WT-GD15 viruses (0.01 MOI each) (Fig. 6a). qRT-PCR analysis revealed that the levels of the G1-like PB2 gene in the progeny viruses was significantly higher than that of the F/98-like PB2 gene (Fig. 6b). Given that these two viruses replicate differentially, 0.005 moi WT-GD15 virus was used in co-infection experiments. We found that at this dosage the two PB2 genes replicated at a comparable rate at 48 hpi (Table 1). Besides, another conserved gene-NP segment were detected to verify whether the changes in the levels of M and PB2 are true. The results showed that at this dosage the number of NP gene copies in progeny viruses were similar among the reassortant viruses, too (Table 1). In the coinfection experiment the number of WT-GD15 virus used was only half of that GD15-PB2 SH14 , the g1 GD15-PB2 gene still demonstrated significant advantages (Fig. 6 C).
Since genotype S H9N2 strains are generated through the replacement of the M and PB2 genes of F/98-like viruses with those from the G1-like viruses, we wondered if the M or PB2 genes would influence each other's preference. MDCK cells were coinfected with GD15-M SH14 and GD15-M SH14 PB2 SH14 viruses (Fig. 6 D). The copy number of the g1 GD15-PB2 gene were significantly higher than that of f98 SH14-PB2 genes in the progeny viruses (Fig. 6e). However, the g1 GD15-PB2 to f98 SH14-PB2 ratio in this co-infection experiment was similar to that co-infected with g1 GD15-PB2 SH14 and WT-GD15 viruses (date not shown), suggesting that the competitive advantage of g1 GD15-PB2 to f98 SH14-PB2 during reassortment is not influenced by the M gene.

Discussion
If two homologous gene segments are available in a cell, they will compete with other for incorporation into progeny viruses [19]. We performed co-transfection and coinfection experiments with the M or PB2 gene derived from the G1-like and F/98-like genotypes. During reassortment, the M or PB2 gene from two genotypes would compete with other for their incorporation into progeny virions. Our co-transfection experiments showed that the copy number of g1 GD15-M/ g1 GD15-PB2 genes were higher in progeny virions than that of f98 SH14-M/ PB2 gene. These observations suggest that there is a biased genetic reassortment between G1-like M/PB2 and F/98-like M/PB2 genes. The advantage of the g1 GD15-PB2 but not g1 GD15-M gene in incorporating into progeny virions was conformed in our co-infection experiment. The discrepancy in the results obtained from co-transfection and co-infection experiments are likely due to the differences in the materials and methods used in the study. In addition, the interactions between viruses are more complex than plasmids. Consistent with our observations, Essere and Kawaoka reported that co-transfection and co-infection result in different reassortant genotypes [18,20,21]. These authors postulated that some unknown cellular factors may affect genetic reassortment [18].
The advantage of the G1-like M/PB2 genes over that of the F98-like M/PB2 genes were further investigated by using two additional genotype S H9N2 viruses, CZ73 and AH320. We found that the incorporation advantage of the G1-like M/PB2 gene was variable among strains. The findings that only the PB2 gene from the H9N2 virus CZ73 and the M gene from the H9N2 virus AH320 had an advantage suggest that the competitive advantage is not equal for the M and PB2 genes from one virus. We speculate that the internal genes of a novel H7N9 virus are not necessarily from one H9N2 virus but rather from a super recombination of genes from different H9N2 viruses.
The segmented genome of influenza viruses allows for the reassortment of gene segments between viruses when they co-infect same cells [22,23], resulting in multiple genotypes of influenza viruses [23][24][25]. Nevertheless, the substitution of the G1-like M and PB2 genes  has reduced the genetic diversity of H9N2 viruses [8,13]; S genotype H9N2 viruses have been predominant in chickens since 2010 [2][3][4]. We speculate that the uniqueness of the internal gene cassette of the S genotype makes it possible to reach a more ideal equilibrium. Our study has demonstrated an advantage of G1like M and PB2 genes over F/98-like M and PB2 genes. It is not clear why the internal genes of the S genotype H9N2 virus remain stable as a cassette and has stayed prevalent in H9N2 and H7N9 viruses for many years. The genotype S H9N2 viruses provide their internal genes to various emerging viruses, especially H7N9 viruses [1,5,10,13,26]. Epidemiological evidence suggests that nearly all human and avian H7N9 isolates possess internal genes that originated from H9N2 viruses [8,27,28]. It is likely that the six internal genes of genotype S H9N2 viruses have reached a stable and optimal combination, ensuring the donation of the internal gene cassette to emerging viruses. On the other hand, we also observed gene segments that originated from H9N2 viruses in other viruses, such as H5N2 and H7N7 viruses but not as a whole cassette [4]. The uniqueness of internal genes of H9N2 viruses, especially those of genotype S, warrants further exploration.

Conclusions
Our study has demonstrated that G1-like PB2/M genes had competitive advantages over SH14 virusderived F/98-like PB2/M gene during reassortment. However, their competitive advantage varied among different strains. The competitive advantage of the PB2 gene was more prominent than that of the M gene. Our results suggest that the preferential incorporation of H9N2-derived G1-like M and PB2 genes into progeny virions of H7N9 influenza viruses may help maintain the stability of the internal gene cassette in the novel H7N9 viruses.

Plasmid construct
The 8-plasmid reverse genetic systems of GD15(H7N9) virus, CZ/73(H9N2) virus and AH320(H9N2) virus were constructed in the present study, while the 8 plasmids for SH14(H9N2) viruswas constructed as previously report [29]. All constructs were verified by sequencing and preserved by our laboratory.

Nine plasmid co-transfection
Six plasmids containing PB1, PA, NP, NS, HA, and NA genes of GD15 (H7N9) virus, plus PB2/M plasmid from GD15/CZ73/AH320 virus and PB2/M plasmid from SH14 (F98-like H9N2) virus were cotransfected into 293 T and MDCK cells to examine the competition between G1-like and F/98-like PB2. The supernatants were collected after 72 h for inoculation in 9-10 day-old egg. One egg was used for each sample. 36 h post inoculation the allantoic fluid was collected and stored in − 70°C.cotransfection experiment was run in triplicate wells and repeated at least twice for each sample.

Reverse genetics
Reassortant viruses were generated by reverse genetics as previously described [30]. Plasmids were transfected into 293 T and MDCK cells using Polyfect transfection reagent (Qiagen) according to the manufacturer's instructions. After 72 h, supernatants were harvested and each of the newly generated viruses was plaque purified. The purified viruses were amplified on SPF embryonated chicken eggs to generate the viral stock used in the study.

Viral growth kinetics
The growth properties of reassortant viruses were assessed as follows. Triplicate wells containing MDCK cells were infected with the indicated viruses at a multiplicity of infection (MOI) of 0.01, supplemented with Opti-MEM (catalog no. 31985-070; Gibco) and incubated at 37°C. Supernatants were collected from each well at 12, 24, 48 and 72 h post infection (h.p.i.) and were stored at − 70°C. Viral titres were subsequently determined as the 50% tissue culture infection dose (TCID50) per 0.1 ml in MDCK cells using the method of Reed and Muench [31].

Coinfection of MDCK cells with Reassortant viruses
MDCK cells were coinfected with viruses at 0.01 or 0.005 MOIs. After incubation at 37°C for 1 h, the virus inoculum was removed, and the cells were washed three times with phosphate-buffered saline (PBS), following by incubation at 37°C in Opti-MEM. After 48 h, supernatants were collected from each well for subsequent analysis. Every experiment was run in triplicate wells and repeated at least twice.