1. F/98 serial passaging with and without selection pressure of vaccine antibodies
At present, the major prevention and control measure for H9N2 subtype avian influenza virus relies on mainly the whole-virus inactivated vaccine [4]. The inactivated vaccine is able to induce humoral immunity in chickens [18]. However, H9N2 subtype avian influenza virus has spread among the chickens of China, and the H9N2 subtype avian influenza virus can still be isolated from chicken flocks with high vaccine antibodies. One of the reasons is possibly the effect of the selection pressure of vaccine antibodies. To explore the effect of homologous selection pressure of vaccine antibodies on the evolution of H9N2 subtype avian influenza virus, this study adopted 4-week-old SPF chickens for use in an animal model with selection pressure of vaccine antibodies, and these naive chickens were called nonimmunized chickens; an F/98 strain oil emulsion vaccine was used to immunize 4-week-old SPF chickens, and the ones whose serum contained a 40~320 HI titer of F/98 strain virus were regarded as animals with selection pressure of vaccine antibodies, namely, the immunized chickens (Figure 1). In the 1-10 passaging generations, the HI titer in the serum of immunized chickens was controlled within 40 (Table 2). In the 11-20 passaging generations, the HI titer in the sera of immunized chickens was controlled within 320 (Table 2). In the first passage generation, 106 EID50 of the F/98 strain virus was used to infect three immunized and nonimmunized chickens by nasal drop, eye drop and intratracheal injection. On the 3rd day after infection, the chickens were sacrificed, and trachea and lung tissues were collected and added into PBS according to a mass/volume ratio of tissue to PBS of 1:10. Under aseptic conditions, the tissues were ground completely, with the supernatant being collected. Afterward, the liquid was divided into three parts. One part was used to vaccinate a 10-day-old SPF chick embryo to test the passaged virus from the trachea and lung tissues. The other part was reserved in a -80℃ freezer for purification of quasispecies viruses; then, 100 μL of supernatant from the trachea and lung tissues was collected and mixed as the seed virus to infect the next generation of immunized or nonimmunized chickens. After passaging the virus for 20 generations in the immunized or nonimmunized chickens, purification of quasispecies viruses was carried out for the supernatant from trachea and lung tissues of immunized and nonimmunized chickens from the 1-5, tenth, fifteenth and twentieth generations by plaque isolation. Viruses obtained from plaques were propagated in 10-day-old SPF chick embryos, and the allantoic fluid was tested with 1% red blood cells. The HA-positive allantoic fluid was stored in a -80℃ freezer. Consequently, 390 progeny strains of virus were isolated (Table 3). One percent red blood cells were used to test the liquid, and the results show that all of the HA titers of purified quasispecies viruses were within 6-10 log2.
2. Antigen variation of H9N2 subtype avian influenza virus accelerated by selection pressure of vaccine antibodies
To explore the effect of the selection pressure of vaccine antibodies on F/98 strain viral antigenicity, the study first used F/98 to test the HI titer of generations 1-5 of purified viruses from the trachea and lung tissues of the immunized and nonimmunized chickens. The results show that the HI titer of F/98 was 6144. The average HI titers of quasispecies viruses of generations 1-5 isolated from trachea tissues with selection pressure of vaccine antibodies were 1952, 1312, 1152, 832 and 624, respectively. The proportion of progeny viruses whose HI titer decreased by more than 4 times in generations 1-5 was 40% (the first generation), 80% (the second generation), 100% (the third generation), 100% (the fourth generation) and 100% (the fifth generation); the average HI titers of quasispecies viruses of generations 1-5 without selection pressure of vaccine antibodies isolated from trachea were 4864, 3840, 3328, 2688 and 2112, respectively. The proportion of progeny viruses whose HI titer decreased by more than 4 times in generations 1-5 was 0% (the first generation), 0% (the second generation), 0% (the third generation), 10% (the fourth generation) and 30% (the fifth generation). The results showed that, in the trachea, when the F/98 strain with selection pressure of vaccine antibodies was passaged for the first generation, there were antigen variant strains, and when it was passaged for the third generation, variation took place in the antigenicity for all the quasispecies viruses. Without selection pressure of vaccine antibodies, only 30% of the quasispecies viruses after the F/98 strain was passaged to the fifth generation had undergone variation (Figure 2).
The average HI titers of quasispecies viruses of generations 1-5 isolated from lungs with selection pressure of vaccine antibodies were 2176, 1168, 704, 592 and 320, respectively, which were 2.8, 5.3, 8.7, 10.4 and 19.2 times lower than the HI titer of the F/98 strain, respectively. The proportion of progeny viruses whose HI titer decreased by more than 4 times in generations 1-5 was 60% (the first generation), 100% (the second generation), 100% (the third generation), 100% (the fourth generation) and 100% (the fifth generation). The average HI titers of quasispecies viruses of generations 1-5 isolated from lungs without selection pressure of vaccine antibodies were 4608, 3840, 3328, 2816 and 2048, respectively, which were 1.3, 1.6, 8.7, 1.8 and 3 times lower than the HI titer of the F/98 strain, respectively. The proportion of progeny viruses whose HI titer decreased by more than 4 times in generations 1-5 was 0% (the first generation), 0% (the second generation), 0% (the third generation), 0% (the fourth generation) and 40% (the fifth generation). The results showed that, in the lung, when the F/98 strain with selection pressure of vaccine antibodies was passaged for the first generation, there were antigen variant strains, and when it was passaged for the second generation, variation took place in the antigenicity for all the quasispecies viruses. Without selection pressure of vaccine antibodies, only 40% of the quasispecies viruses after the F/98 strain was passaged to the fifth generation had undergone variation (Figure 2). All the above results showed that the selection pressure of vaccine antibodies was effective in obviously promoting the antigen variation of H9N2 subtype avian influenza virus.
3 Analysis of variation in the sequence of the HA gene for virus serially passaged with and without selection pressure of vaccine antibodies
3.1 Selection pressure of vaccine antibodies regulated HA gene mutation of H9N2 subtype avian influenza virus
HA is the major surface protein of avian influenza virus as well as the major antigen protein of the virus. On the HA protein, antigen determinants and receptor-binding sites of the virus are distributed. After infection with avian influenza virus or vaccination with avian influenza virus, specific antibodies for HA protein are generated in the body. In the process of the interaction between influenza and the host, the virus will escape from antibodies by amino acid variation on HA. After passaging the F/98 strain in the chickens for generations with and without selection pressure of vaccine antibodies, the study showed that selection pressure of vaccine antibodies can obviously promote the variation of F/98. To further study the evolution of the HA antigen protein of the F/98 strain when it escapes from the selection pressure of vaccine antibodies, the isolated viruses were sequenced. Comparing to HA gene sequence of F/98, there were 32 nucleotide mutations in the HA sequence of the virus isolated from the trachea with selection pressure of vaccine antibodies, including 19 meaningful mutations, accounting for 43.75% of the total amount; it was found out that there were 32 nucleotide mutations in HA of the virus isolated from the lung with selection pressure of vaccine antibodies, including 14 meaningful mutations, accounting for 57.58% of the total amount; it was found out that there were 32 nucleotide mutations in HA of the virus isolated from the trachea without selection pressure of vaccine antibodies, including 32 meaningful mutations, accounting for 53.13% of the total amount; it was found out that there were 26 nucleotide mutations in HA of the virus isolated from the lung without selection pressure of vaccine antibodies, including 20 meaningful mutations, accounting for 76.92% of the total amount; and it was found out that the meaningful mutation rate of the virus isolated from the lung was higher than that of the virus isolated from the trachea ($, & and P<0.05, seen in Table 4). These results show that the meaningful mutation rate with selection pressure of vaccine antibodies was positively lower than that without selection pressure of vaccine antibodies (seen in Table 4, P<0.05) All of the results implied that selection pressure of vaccine antibodies had the effect of limiting the mutation of the HA genes of the F/98 strain in H9N2 subtype avian influenza virus; moreover, the lung had larger effect on the HA genes of the F/98 strain of H9N2 subtype avian influenza virus than the trachea.
3.2 A special mutation hotspot generated by the H9N2 subtype avian influenza virus with selection pressure of vaccine antibodies
A spot with a higher mutation rate is called a mutation hotspot. To analyze the differences in the effects of the situations with and without selection pressure of vaccine antibodies on HA of F/98, we defined a special mutation hotspot of the HA after passaging for generations with and without selection pressure of vaccine antibodies by 5 amino acids in a group. After analyzing and comparing with F/98, we found that there were 5 mutation hotspots on HA of the viruses isolated from the trachea with selection pressure (Figure 3), namely, amino acids 131-135, 166-170, 196-200, 201-205 and 231-235. Among them, the 131-135 mutation hotspot contained K131R in 17 strains; the 166-170 mutation hotspot contained A168T in 3 strains; the 196-200 mutation hotspot contained A198V in 56 strains; the 201-205 mutation hotspot contained N201D in 3 strains; and the 231-235 mutation hotspot contained Q234L in 36 strains. We found that there were 4 mutation hotspots in HA of the viruses isolated from the trachea without selection pressure, namely, amino acids 196-200, 221-225, 231-235 and 281-285. Among them, the 196-200 mutation hotspot contained A198V in 55 strains; the 221-225 mutation hotspot contained M224K in 24 strains; the 231-235 mutation hotspot contained Q234L in 83 strains; the 281-285 mutation hotspot contained L281F in 9 strains; and the 281-285 mutation hotspot contained N285D in 4 strains. The 131-135, 166-170 and 201-205 mutation hotspots were specific to the tracheas with selection pressure; the 221-225 and 281-285 mutation hotspots were specific to the tracheas without selection pressure; and the 196-200 and 231-235 common mutation hotspots were common. The results showed that in the trachea, the selection pressure of vaccine antibodies can generate mutations in the F/98 strain.
In the lung, we found that there were 3 mutation hotspots in HA of the viruses isolated with selection pressure of vaccine antibodies, namely, amino acids 131-135, 196-200 and 231-235. The 131-135 mutation hotspot contained K131R in 18 strains; the 196-200 mutation hotspot contained A198V in 62 strains; and the 231-235 mutation hotspot contained Q234L in 32 strains. There were 4 mutation hotspots in HA of the viruses isolated without selection pressure of vaccine antibodies, namely, amino acids 111-115, 196-200, 221-225 and 231-235. Among them, the 111-115 mutation hotspot contained E114K in 10 strains; the 196-200 mutation hotspot contained A198V in 66 strains; the 221-225 mutation hotspot contained M224K in 16 strains; and the 221-235 mutation hotspot contained Q234L in 25 strains. The number of HA mutation hotspots of the viruses isolated from the lungs with selection pressure of vaccine antibodies was 1 less than that of the viruses isolated from lungs without selection pressure, but the 131-135 mutation hotspot was specific to HA of the viruses from the lungs with selection pressure of vaccine antibodies. The 111-115 and 221-225 mutation hotspots were specific to HA of the viruses from the lungs without selection pressure of vaccine antibodies, while all the HA genes of the viruses isolated from the lungs with and without selection pressure of vaccine antibodies had the 196-200 and 231-235 mutation hotspots. In summary, F/98 generated 3 specific hotspots affected by the selection pressure of vaccine antibodies, namely, amino acids 131-135, 166-170 and 201-205; F/98 also generated 3 specific hotspots without being affected by the selection pressure of vaccine antibodies, namely, amino acids 111-115, 221-225 and 281-285; and F/98 generated the 196-200 and 231-2352 hotspots with and without the selection pressure of vaccine antibodies. These results showed that the HA protein of H9N2 has its own specific evolution and mutation strategies when escaping from the selection pressure of vaccine antibodies of the host.
3.3 The mutation position and distribution features in the natural environment of the HA gene of H9N2 subtype avian influenza virus with selection pressure of vaccine antibodies
The previous results show that the HA gene of the F/98 strain virus with selection pressure of vaccine antibodies generated specific mutation hotspots in some areas. To understand the features of these specific mutations, we marked the areas for HA amino acid sequence mutation hotspots of quasispecies viruses serially passaged from generation to generation with and without selection pressure of vaccine antibodies on the 3D structural model of the HA protein and then analyzed the ratio of them observed in natural isolated strains. We found that the HA gene mutations K131R, A168T, A198T and N201D of quasispecies viruses serially passaged from generation to generation with selection pressure of vaccine antibodies were all located in the head of HA protein; the mutations HA gene E114K, A198T, M224K and L234Q of quasispecies viruses serially passaged from generation to generation without selection pressure of vaccine antibodies were located in the head of HA protein, while L281F and N285D were located in the neck of the HA protein (Figure 4).
To analyze the distribution of these mutation hotspots in naturally isolated strains, we analyzed the residues 114, 131, 168, 198, 201, 224, 234, 281 and 285 of the HA genes of the H9 subtype avian influenza virus released in GenBank from 1994-2015. It was found that the major amino acid at the positions 114, 131 and 281 on the HA genes of natural isolated strains was E, R, and L, respectively (accounting for >99.5%), while the positions 168, 198, 201, 224, 234, and 285 had multiple amino acids existing in the HA gene of the H9 subtype avian influenza viruses isolated over many years. These mutations also existed in the 131-135 and 166-170 (166-170) hotspots, which were specific to the HA gene from viruses serially passaged from generation to generation with selection pressure of vaccine antibodies. The HA genes of viruses that were serial passaged from generation to generation with and without selection pressure of vaccine antibodies all had the 196-200 and 231-235 (198, 224, 234) mutation hotspots, and the mutation hotspot of the HA genes of viruses that were serially passaged from generation to generation with selection pressure of vaccine antibodies was amino acids 281-285 (281 and 285) (Figure 5). Accordingly, mutation of the positions 131, 168 and 201 of natural isolated strains was related to the immunization status of the host, while mutation of the positions 224, 281 and 285 was related to the continuous replication and dissemination of the avian influenza virus in the host, and mutation of the positions 198 and 234 may be related to both factors.
4. The specific amino acid variation of the H9N2 subtype avian influenza virus F/98 strain with selection pressure of vaccine antibodies in chickens
After serial passaging from generation to generation for H9N2 subtype avian influenza virus F/98, in addition to mutation of the HA gene, other genes had also undergone corresponding changes. After serial passaging, the quasispecies viruses from the tracheas and lungs of the twentieth-generation chickens were isolated from plaques. Then, the NA, PB2, PB1, PA, NP, M and NS genes of the progeny viruses were sequenced.
4.1 Ten amino acids lost in NA of the F/98 strain of the H9N2 subtype avian influenza virus with selection pressure of vaccine antibodies
NA is endowed with sialidase activity, which cleaves host cell sialic acid and generates new virus particles. NA is related to virus germination and release [19]. Compared with those in F/98, in the 44 strains of the twentieth generation isolated from trachea and lung with selection pressure of vaccine antibodies, ten amino acids (67-76) of NA were lost, while the virus without selection pressure had not undergone mutation and deletion (Table 5). The selection pressure of vaccine antibodies led to the NA gene losing amino acids 67-76, and their functions were further studied.
4.2 Analysis of polymerase gene mutation of the twentieth-generation quasispecies viruses serially passaged from generation to generation with and without selection pressure of vaccine antibodies
Since the polymerase of avian influenza virus lacks proofreading function, it leads to the avian influenza virus constantly mutating, which occurs on not only the surface genes of avian influenza virus but also the internal genes. To study the effect of the selection pressure of vaccine antibodies on F/98 internal genes, the internal genes of the twentieth-generation quasispecies viruses (78 strains) were sequenced.
In the 20 strains isolated from the tracheas of chickens with selection pressure, PB2 of 1 strain (5%) had a D256N mutation, PB2 of 13 strains (65%) had an R318K mutation and PB2 of 20 strain (100%) had an R355K mutation; in the 24 strains isolated from the tracheas of chickens with selection pressure, PB2 of 19 strains (79.1%) had an R318K mutation, PB2 of 24 strains (100%) had an R355K mutation PB2 of 1 strain (5.2%) had a K526R mutation, and PB2 of 2 strains (10.5%) had an A622V mutation. In the 19 strains of viruses isolated from the tracheas of chickens without selection pressure, all the PB2 proteins of the 19 strains (100%) had an R318K mutation; in the 15 strains isolated from the tracheas of chickens without selection pressure, PB2 of 1 strain (6.6%) had an I63M mutation and PB2 of the 15 strains (100%) had an R318K mutation (Table 5). After analyzing the gene data of wild-type strains in the NCBI avian influenza virus database, the position 318 of PB2 in the viruses isolated from the natural environment is an R in the majority and a K in 1%, while the position 355 of PB2 in the viruses isolated from the natural environment is an R in the majority and a K in 6.25% (Figure 6). The results showed that all the PB2 genes of the viruses isolated from tracheas and lungs of the twentieth-generation chickens with selection pressure had R355K mutations, and 72% of the PB2 genes from isolated strains had R318K mutations. The PB2 gene of 100% of isolated strains from the tracheas and lungs of the twentieth-generation chickens without selection pressure had R318K mutations, but there was no R355K mutation, which signified that the R355K mutation in the PB2 genes was the result of the selection pressure of vaccine antibodies on F/98. The R318K mutation in the PB2 gene appeared with and without selection pressure of vaccine antibodies, although the PB2 gene mutation rate of the isolated strains after serial passaging was lower than that without selection pressure of vaccine antibodies.
Compared with the amino acid sequence of PB1 of F/98, there were no mutations detected for the viruses isolated from the lungs with and without selection pressure of vaccine antibodies. In PB1 of 20 strains of quasispecies viruses isolated from the trachea in the twentieth generation after serial passaging with selection pressure of vaccine antibodies, there was only an I397V mutation in PB1 of 1 strain; in PB1 of 19 strains of quasispecies viruses isolated from the tracheas in the twentieth generation after serial passaging without selection pressure, there was only an I517V mutation in PB1 of 1 strain virus (Table 5). The strains with I397V and I517V mutations in the PB1 genes were not advantageous, and their biological significance was not manifested in this study.
In terms of PA genes, compared with the amino acid sequence of the PA gene of the F/98 strain virus, there were no mutations detected for the 20 strains isolated from the trachea in the twentieth generation after serial passaging with selection pressure. For the 24 strains isolated from the lung in the twentieth generation after serial passaging with selection pressure, 1 strain had an F46L mutation, 1 strain had a V127I mutation, 1 strain had an L132I mutation and 1 strain had an E319V mutation. For the 19 strains isolated from the trachea in the twentieth generation after serial passaging without selection pressure, 1 strain had an R185K mutation, 1 strain had an S190F mutation, 1 strain had an F646S mutation and 1 strain had a C693G mutation. For the 15 strains isolated from the lungs in the twentieth generation after serial passaging without selection pressure, 1 strain had a V554I mutation (Table 5). These mutations detected from the PB1 and PA proteins independently existed in one strain of virus. Since avian influenza virus undergoes error-prone replication, we were not sure whether these mutations were related to immunization, so their significance remains to be further studied.
4.3 Analysis of NP protein, M protein and NS protein mutations of the twentieth generation quasispecies viruses after serial passaging with and without selection pressure
The NP protein of avian influenza virus is able to combine with the RNA of the virus genome and form a ribonucleoprotein (RNP) complex with RNA polymerase PB2, PB1, and PA to assist the virus in realizing transcription and replication [20]. The smallest gene section of influenza A virus is NS, which codes the two proteins NS1 and NS2 [21-24]. The NS1 protein is related to the toxicity of influenza A virus, which is engaged in replication of influenza virus and regulation of viral protein synthesis and affects the morphological structure of virus particles, as well as prohibits the host immune response and apoptosis of cells [25].
For NP, compared with the F/98 sequence, there were no mutations detected for viruses isolated from the tracheas (20 strains) and lungs (24 strains) in the twentieth generation after serial passaging with selection pressure. For the 19 strains isolated from the trachea in the twentieth generation after serial passaging without selection pressure, all 19 strains had a V186I mutation, and 9 strains had an L466I mutation. For the 15 strains isolated from the trachea in the twentieth generation after serial passaging without selection pressure, all 15 strains had a V186I mutation, and 14 strains had an L466I mutation, which shows that the V186I and L466I mutations were related to the mutation-prone nature of the virus. Similarly, this result also indicates that the selection pressure of vaccine antibodies limits the NP gene to generate the V186I and L466I mutations (Table 5). For the M gene, compared with the F/98 sequence, for the 20 strains isolated from the trachea in the twentieth generation after serial passaging with selection pressure, the M1 gene of 1 strain had an L28P mutation, and the M1 gene of 2 strains of the 24 strains isolated from the lung had an R304G mutation. For the 19 strains isolated from the trachea in the twentieth generation after serial passaging without selection pressure, the M1 gene of 2 strains had an F62L mutation, and the M2 gene of 1 strain had an R256H mutation; for the 15 strains isolated from the lung, the M2 gene of 1 strain had an A209T mutation (Table 5). For the NS gene, compared with the F/98 sequence, there were no mutations detected in the strains isolated from the trachea; for the 24 strains isolated from the lung, the NS gene of 1 strain had a P85L mutation. For the 19 strains isolated from the trachea in the twentieth generation after serial passaging without selection pressure, the NS gene of 1 strain had an L33Q mutation, and the NS1 gene of 19 strains had an L77I mutation; for the 15 strains isolated from the lung, the NS gene of 15 strains had an L77I mutation, and the NS gene of 1 strain had a V84M mutation (Table 5). After analyzing the gene data for wild-type strains in the NCBI influenza database, it was found that the V186I and L466I mutations of NP genes do not exist in wild type, while the L77I mutation in the NS gene accounted for only 0.92% (Figure 6). The V186I and L466I mutations of the NP gene and the L77I mutation of the NS gene occur in only the viruses passaging from generation to generation without the selection pressure of vaccine antibodies, which indicates that the selection pressure of vaccine antibodies was less effective for NS and NP than for other viral proteins.
5 Selection pressure of vaccine antibodies regulated the infection of the F/98 strain virus
To explore the effect of the selection pressure of vaccine antibodies on the infection of F/98, the study selected progeny viruses to detect the EID50 and ELD50, according to the mutations of each gene. For the viruses with and without selection pressure of vaccine antibodies, we selected 5 strains each, and the mutations in each strain were as follows (seen in Table 6): V1: HA (K131R, A198V, Q234L), NA (67-76 amino acid deletion), PB2 (R318K, R355K); V2: HA (K131R, A168T, A198T, L234Q), NA (67-76 amino acid deletion), PB2 (R318K); V3: HA (K131R, N201D, A198T, L234Q), NA (67-76 amino acid deletion); V4: HA (K131R, A198T, L234Q), NA (67-76 amino acid deletion), PB2 (R318K, R355K); V5: HA (K131R, A198T, L234Q), NA (67-76 amino acid deletion); N1: HA (A198V, M224K, Q234L, L281F, N285D), PB2 (R318K), NP (V186I, L466I), NS1 (L77I), M1 (F62L); N2: HA (A198V, M224K, Q234L, N285D), PB2 (R318K), NP (V186I, L466I), NS1 (L77I), M1 (F62L); N3: HA (A198V, M224K, Q234L, L281F, N285D), PB2 (R318K), NP (V186I, L466I), NS1 (L77I), M1 (F62L); N4: HA (E114K, A198V, M224K, Q234L, L281F, N285D), PB2 (R318K), NP (V186I, L466I), NS1 (L77I), M1 (F62L); and N5: HA (A198V, M224K, Q234L, L281F, N285D), PB2 (R318K), NP (V186I, L466I), NS1 (L77I), M1 (F62L). The EID50 values of progeny viruses V1-V5 with selection pressure of vaccine antibodies were within 6.9-7.2; in the testing process of ELD50, within 120 hours after infection, there were no cases of infectious embryo death; the EID50 values of the progeny viruses N1-N5 without selection pressure of vaccine antibodies were within 7.2-7.5 (shown in Figure 7). Compared with that of F/98, there was no obvious change in the EID50 of the viruses with selection pressure of vaccine antibodies (increased by 1.9 times), while the EID50 of the viruses without selection pressure of vaccine antibodies was 794 times more than that of F/98 and 397 times more than that of the viruses with selection pressure of vaccine antibodies, so their infection ability rose obviously (P<0.01). Compared with that of F/98, the ELD50 of V20 with selection pressure of vaccine antibodies was 0, and it was not fatal; its reduced ELD50 was obvious (P<0.01), while the EID50 of N20 was 6.3 times greater than that of F/98, and the ELD50 was obviously increased (P<0.01). These results show that the selection pressure of vaccine antibodies can regulate the infectious ability of H9N2 and the fatality of the chick embryo.