Polymorphism frequencies at positions 271, 590-591, 627, and 701 in PB2, according to virus lineage. On the left is the neighbor-joining tree for PB2, with collapsed tips and branches colored by major lineage. A total of 2,259 PB2 sequences were analyzed (see Materials and Methods). Also shown are pie charts for each lineage, depicting the PB2 polymorphisms (across all hosts) at amino acids 271, 590-591, 627, and 701. The avian-like amino acids are colored blue, and the main alternative is colored red. Other amino acids are colored white. Details of the percentages of the different amino acids found at each PB2 site can be found in Table S2 in the supplemental material. 

Contexts in source publication

Context 1

... and pig cells. G1 polymerase was more active than either Bav or 50-92 polymerase in human and pig cells, with an activity approaching that of Eng195. Expression of the firefly luciferase reporter gene by the polymerase derived from the HPAI H5N1 Ty05 virus was stronger than expression driven by the G1 or Eng195 polymerase in human or pig cells. In avian cells, the differences between the influenza virus polymerase activities were not dramatic, and the pattern was different. For example, the H9N2 G1 polymerase was the most active, followed by the Eng195 and Vic polymerases. The difference between the 50-92 and Vic polymerases was only 3.3-fold in avian DF-1 cells, whereas 50- and 98-fold differences were observed in swine NPTr cells and human 293T cells, respectively. Taken together, these data imply that a host range barrier at the level of polymerase activity for typical avian influenza viruses exists in pig cells, just as it does in human cells. virus polymerase activity in pig cells. In order to gain some in- sights into the nature of the genetic changes that mediate efficient replication of a typical avian-origin influenza virus polymerase in swine, we first analyzed the polymorphisms at known host range- determining residues of the PB2 component of influenza virus polymerases that are endemic in different hosts, including pigs. The polymorphism frequencies in PB2 at positions 271, 590-591, 627, and 701 were estimated from a sample of complete genome sequences downloaded from the NCBI influenza virus resource (110). A total of 2,259 PB2 sequences were analyzed (see Tables S1 and S2 in the supplemental material for details). The amino acids at these positions in the strains used in the present study are shown in Table 1. The most studied of these PB2 polymorphisms is undoubtedly the PB2 E627K mutation. The only swine lineage of influenza virus that contains PB2 627K is the classical swine lineage that was reintroduced in 1918 into pigs from humans rather than evolving naturally in pigs via an avian source (Fig. 3). In contrast, more than 30 years after introduction of the Eurasian avian-like H1N1 lineage from birds directly to pigs, the PB2 E627K mutation has not been naturally selected. Similarly, 627E is found in PB2 sequences from swine influenza viruses that have the TRIG cassette, in which PB2 and PA originated from a North American avian- origin virus 14 years ago (70, 71). This suggests that there is not a strong selection pressure for 627K to emerge in pigs, in contrast to other mammalian species. To test whether this was because 627K does not affect polymerase function in pig cells, the E627K mutation was engineered by site-directed mutagenesis in PB2 genes from the typical host-restricted avian virus polymerases, the Bav and 50-92 polymerases. Activities of polymerase complexes containing WT PB2 or PB2 with a mutation at position 627 were compared in human, swine, and avian cells (Fig. 4A, B, and C). To demonstrate that the host range mutations did not affect expression of the PB2 mutant proteins, the mutations were additionally engineered into an epitope-tagged 50-92 PB2 gene and expressed individually in each cell type. Western blot analysis using antibody to the C-terminal epitope tag (Flag tag) showed that none of the host range mutations affected the accumulation of PB2 protein in any of the cell types (see Fig. S1 in the supplemental material). This result is in line with several other publications that also show that mutations that affect polymerase activity, such as those at PB2 position 627, do not alter the expression or accumulation of the protein (35, 37, 69, 111). As previously reported (34, 35, 37–40), the introduction of 627K into avian-origin viral polymerases dramatically increased expression of the reporter gene in human cells (Fig. 4A). Interestingly, 627K also significantly increased avian virus polymerase activity in pig cells ( P Ͻ 0.5 by unpaired Student’s t test) (Fig. 4B). Compared to the respective WT polymerases, increases of about 12-fold and 10-fold were observed for Bav E627K and 50-92 E627K polymerases, respectively, meaning that they approached the level of activity of the Eng195 polymerase constellation in pig cells (Fig. 2B). The activity of the H9N2 G1 polymerase was also increased even further in pig cells (see Fig. S2 in the supplemental material) and human cells (data not shown) when the E627K mutation was introduced. Conversely, a reduction of activity in both cell types was observed when the human-adapted Vic PB2 or Ty05 PB2 was mutated (K627E) (see Fig. S2). Similar to the E627K mutation, the D701N adaptive mutation has been reported to emerge when avian influenza viruses are passaged in animals such as mice (36, 41, 59, 112, 113). The D701N mutation in PB2 is associated with enhanced polymerase activity in human cells (35, 36, 113–115). During the natural evolution of influenza virus, this mutation has been selected in the Eurasian swine lineage only (Fig. 3). Indeed, 701N is present in the PB2 proteins of early isolates of the Eurasian avian-like H1N1 swine viruses (for example, A/Swine/Germany/2/1981) but not in those from the supposed avian precursor, A/Duck/Bavaria/1/77. The effect of this mutation on the activity of two avian virus polymerases, the Bav and 50-92 polymerases, was tested in human and pig cells (Fig. 4A and B, respectively). In both cell types, D701N mutation resulted in increased polymerase activity, to a lesser ex- tent than that with the E627K mutation, but the difference was still statistically significant ( P Ͻ 0.01 by unpaired Student’s t test) compared to the wild-type avian polymerase activities. Thus, in pig cells, the luciferase signal from Bav polymerase with the D701N mutation was 2.7-fold higher than that from WT Bav polymerase. The D701N mutation increased 50-92 polymerase activity 4.8-fold in pig cells. In avian cells, neither the E627K nor D701N mutation increased the polymerase activity of either of the avian viruses tested (Fig. 4C). PB2 residues 627K and 701N are absent in swine triple-reassortant viruses and also in the descendant 2009 pH1N1 viruses (Fig. 3; see Table S1 in the supplemental material). However, the G590S/Q591R and T271A mutations present in the TRIG cas- sette-encoded PB2 protein can compensate for this absence (34, 69). These mutations might have been selected and maintained in pigs because they confer a specific replicative advantage in this species. The T271A mutation is also found in seasonal human viruses, in conjunction with the E627K switch, but it did not emerge in the classical swine lineage (Fig. 3). Sequences encoding the G590S/Q591R and T271A mutations were introduced individually into the 50-92 PB2 gene, and minigenome assays were performed to assess the consequences of these changes on polymerase activity in human, swine, and avian cells (Fig. 4A, B, and C). The results indicate that G590S/Q591R and T271A mutations individually increased the activity of avian-origin 50-92 polymerase in both human and swine cells (4.5- and 3-fold, respectively, in human cells and 2.4- and 2.2-fold, respectively, in pig cells). Polymerase activities in DF-1 cells were not significantly affected by these mutations, in agreement with previous publications (34, 69). Thus, the four PB2 mutations tested (E627K, D701N, G590S/ Q591R, and T271A), which adapt avian influenza virus polymer- ase for human cells, also increased activity in pig cells but not in avian cells. PB2 E627K enhances virus replication in pig cells. To test that the results obtained with the in vitro polymerase assay reflected a difference of replication in the context of infectious virus, we generated a set of recombinant viruses by reverse genetics. Viruses that contained wild-type A/Turkey/England/50-92/91 or A/Duck/ Bavaria/1/77 polymerase and NP genes or encoded the mammalian adaptive mutation E627K in PB2 were produced. To elimi- nate host restrictions that mapped to cell entry, interferon response, or vRNP export, we replaced the HA and NA surface protein genes and the NS and M genes of each virus with those from the vaccine strain PR8. PR8-based recombinant viruses have been used to experimentally infect minipigs (116). The multicycle replication of the viruses was compared in swine NPTr cells. Both PB2 E627K-containing viruses grew to higher titers than their wild-type counterparts; the differences in titer were statistically significant except at the last time point, i.e., 48 h postinfection (Fig. 5A). Moreover, Western blot analysis showed that viral gene expression, as exemplified by the M1 protein, was increased in the cells infected with PB2 E627K-expressing viruses over those infected by viruses with wholly avian polymerase constellations, and this was especially evident early during infection (Fig. 5B). These differences confirmed a growth advantage resulting from increased polymerase activity for PB2 E627K-containing viruses in vitro in swine cells, as has been seen in human or simian cells. Pigs are thought to play an important role in influenza virus transmission between birds and humans. They have been proposed to be an intermediate host for the adaptation of avian influenza vi- ruses to humans and for the generation of reassortant viruses with pandemic potential (reviewed in reference 16). Pigs are naturally susceptible to infection with at least some avian influenza viruses, and avian-origin viruses have been isolated from pigs worldwide (reviewed in references 8 and 9). Thus, one might anticipate that pig cells would be more permissive to avian-origin influenza virus polymerase than human cells are. Moreover, it may be that pigs express cofactors for avian influenza virus polymerase that are not present or are different in other mammalian species, such as humans. Using an influenza virus polymerase assay, we investigated the ability of six influenza virus polymerase complexes from different species (turkey, duck, ...

Context 2

... for details). The amino acids at these positions in the strains used in the present study are shown in Table 1. The most studied of these PB2 polymorphisms is undoubtedly the PB2 E627K mutation. The only swine lineage of influenza virus that contains PB2 627K is the classical swine lineage that was reintroduced in 1918 into pigs from humans rather than evolving naturally in pigs via an avian source (Fig. 3). In contrast, more than 30 years after introduction of the Eurasian avian-like H1N1 lineage from birds directly to pigs, the PB2 E627K mutation has not been naturally selected. Similarly, 627E is found in PB2 sequences from swine influenza viruses that have the TRIG cassette, in which PB2 and PA originated from a North American avian- origin virus 14 years ago (70, 71). This suggests that there is not a strong selection pressure for 627K to emerge in pigs, in contrast to other mammalian species. To test whether this was because 627K does not affect polymerase function in pig cells, the E627K mutation was engineered by site-directed mutagenesis in PB2 genes from the typical host-restricted avian virus polymerases, the Bav and 50-92 polymerases. Activities of polymerase complexes containing WT PB2 or PB2 with a mutation at position 627 were compared in human, swine, and avian cells (Fig. 4A, B, and C). To demonstrate that the host range mutations did not affect expression of the PB2 mutant proteins, the mutations were additionally engineered into an epitope-tagged 50-92 PB2 gene and expressed individually in each cell type. Western blot analysis using antibody to the C-terminal epitope tag (Flag tag) showed that none of the host range mutations affected the accumulation of PB2 protein in any of the cell types (see Fig. S1 in the supplemental material). This result is in line with several other publications that also show that mutations that affect polymerase activity, such as those at PB2 position 627, do not alter the expression or accumulation of the protein (35, 37, 69, 111). As previously reported (34, 35, 37–40), the introduction of 627K into avian-origin viral polymerases dramatically increased expression of the reporter gene in human cells (Fig. 4A). Interestingly, 627K also significantly increased avian virus polymerase activity in pig cells ( P Ͻ 0.5 by unpaired Student’s t test) (Fig. 4B). Compared to the respective WT polymerases, increases of about 12-fold and 10-fold were observed for Bav E627K and 50-92 E627K polymerases, respectively, meaning that they approached the level of activity of the Eng195 polymerase constellation in pig cells (Fig. 2B). The activity of the H9N2 G1 polymerase was also increased even further in pig cells (see Fig. S2 in the supplemental material) and human cells (data not shown) when the E627K mutation was introduced. Conversely, a reduction of activity in both cell types was observed when the human-adapted Vic PB2 or Ty05 PB2 was mutated (K627E) (see Fig. S2). Similar to the E627K mutation, the D701N adaptive mutation has been reported to emerge when avian influenza viruses are passaged in animals such as mice (36, 41, 59, 112, 113). The D701N mutation in PB2 is associated with enhanced polymerase activity in human cells (35, 36, 113–115). During the natural evolution of influenza virus, this mutation has been selected in the Eurasian swine lineage only (Fig. 3). Indeed, 701N is present in the PB2 proteins of early isolates of the Eurasian avian-like H1N1 swine viruses (for example, A/Swine/Germany/2/1981) but not in those from the supposed avian precursor, A/Duck/Bavaria/1/77. The effect of this mutation on the activity of two avian virus polymerases, the Bav and 50-92 polymerases, was tested in human and pig cells (Fig. 4A and B, respectively). In both cell types, D701N mutation resulted in increased polymerase activity, to a lesser ex- tent than that with the E627K mutation, but the difference was still statistically significant ( P Ͻ 0.01 by unpaired Student’s t test) compared to the wild-type avian polymerase activities. Thus, in pig cells, the luciferase signal from Bav polymerase with the D701N mutation was 2.7-fold higher than that from WT Bav polymerase. The D701N mutation increased 50-92 polymerase activity 4.8-fold in pig cells. In avian cells, neither the E627K nor D701N mutation increased the polymerase activity of either of the avian viruses tested (Fig. 4C). PB2 residues 627K and 701N are absent in swine triple-reassortant viruses and also in the descendant 2009 pH1N1 viruses (Fig. 3; see Table S1 in the supplemental material). However, the G590S/Q591R and T271A mutations present in the TRIG cas- sette-encoded PB2 protein can compensate for this absence (34, 69). These mutations might have been selected and maintained in pigs because they confer a specific replicative advantage in this species. The T271A mutation is also found in seasonal human viruses, in conjunction with the E627K switch, but it did not emerge in the classical swine lineage (Fig. 3). Sequences encoding the G590S/Q591R and T271A mutations were introduced individually into the 50-92 PB2 gene, and minigenome assays were performed to assess the consequences of these changes on polymerase activity in human, swine, and avian cells (Fig. 4A, B, and C). The results indicate that G590S/Q591R and T271A mutations individually increased the activity of avian-origin 50-92 polymerase in both human and swine cells (4.5- and 3-fold, respectively, in human cells and 2.4- and 2.2-fold, respectively, in pig cells). Polymerase activities in DF-1 cells were not significantly affected by these mutations, in agreement with previous publications (34, 69). Thus, the four PB2 mutations tested (E627K, D701N, G590S/ Q591R, and T271A), which adapt avian influenza virus polymer- ase for human cells, also increased activity in pig cells but not in avian cells. PB2 E627K enhances virus replication in pig cells. To test that the results obtained with the in vitro polymerase assay reflected a difference of replication in the context of infectious virus, we generated a set of recombinant viruses by reverse genetics. Viruses that contained wild-type A/Turkey/England/50-92/91 or A/Duck/ Bavaria/1/77 polymerase and NP genes or encoded the mammalian adaptive mutation E627K in PB2 were produced. To elimi- nate host restrictions that mapped to cell entry, interferon response, or vRNP export, we replaced the HA and NA surface protein genes and the NS and M genes of each virus with those from the vaccine strain PR8. PR8-based recombinant viruses have been used to experimentally infect minipigs (116). The multicycle replication of the viruses was compared in swine NPTr cells. Both PB2 E627K-containing viruses grew to higher titers than their wild-type counterparts; the differences in titer were statistically significant except at the last time point, i.e., 48 h postinfection (Fig. 5A). Moreover, Western blot analysis showed that viral gene expression, as exemplified by the M1 protein, was increased in the cells infected with PB2 E627K-expressing viruses over those infected by viruses with wholly avian polymerase constellations, and this was especially evident early during infection (Fig. 5B). These differences confirmed a growth advantage resulting from increased polymerase activity for PB2 E627K-containing viruses in vitro in swine cells, as has been seen in human or simian cells. Pigs are thought to play an important role in influenza virus transmission between birds and humans. They have been proposed to be an intermediate host for the adaptation of avian influenza vi- ruses to humans and for the generation of reassortant viruses with pandemic potential (reviewed in reference 16). Pigs are naturally susceptible to infection with at least some avian influenza viruses, and avian-origin viruses have been isolated from pigs worldwide (reviewed in references 8 and 9). Thus, one might anticipate that pig cells would be more permissive to avian-origin influenza virus polymerase than human cells are. Moreover, it may be that pigs express cofactors for avian influenza virus polymerase that are not present or are different in other mammalian species, such as humans. Using an influenza virus polymerase assay, we investigated the ability of six influenza virus polymerase complexes from different species (turkey, duck, quail, and human) to amplify and express a reporter minigenome in swine cells. Surprisingly, our results indicated that pig and human cells have similar restrictions in supporting the function of influenza virus polymerases from different origins. Low activity was observed in pig cells for polymerases from the two classical avian strains A/Turkey/England/ 50-92/91 (50-92) and A/Duck/Bavaria/1/77 (Bav). Although pigs seem to support the growth of many typical avian influenza viruses, Kida et al. showed that not all strains replicated or induced a serological response in experimentally infected pigs. In addition, the large majority of avian viruses replicated to a lower level in pigs than a swine-adapted virus did (10). In the field, only a minority of avian influenza virus strains seem to naturally spread in the pig population. This suggests that avian influenza viruses with the potential to efficiently replicate and transmit in pigs are rare. Based on our polymerase assay, some candidates might be viruses from the H9N2 G1 lineage or the H5N1 Eurasian lineage Z genotype, such as Ty05 virus, that already bear the E627K mutation. H9N2 A/Quail/Hong Kong/G1/97 (G1) polymerase showed good activity in pig cells and also in human cells (Fig. 2). These results correlate with recent reports showing that H9N2 G1 virus replicates efficiently in human cells (117) and that a virus from the G1 lineage can replicate in 4-week-old pigs and in mice (118). In addition, H9N2 viruses from various genotypes can replicate in ferrets (119), and an H9N2 virus grew to a high titer in a porcine differentiated respiratory epithelial cell ...

Context 3

... with pig, human, and avian cells, using a panel of influenza virus polymerases of different origins. A/Turkey/England/ 50-92/91 (50-92) (H5N1) and A/Duck/Bavaria/1/77 (Bav) (H1N1) are two typical avian influenza virus strains. We previously described 50-92 as a typical poultry-adapted avian influenza virus whose polymerase displays host range restriction in mammalian cells (35, 39, 91). A/Duck/Bavaria/1/77 is considered an ancestor of the avian-origin H1N1 virus that crossed into pigs in 1979 to generate the Eurasian avian-like swine H1N1 lineage (54, 102). A/Quail/Hong Kong/G1/97 (G1) is a less typical avian influenza virus that is representative of one of the prevalent avian H9N2 virus lineages circulating in southern China (103). Viruses from the G1 lineage have been responsible for two human infections, causing mild illness (104–106), but G1 H9N2 viruses have not been isolated from pigs. A/Turkey/Turkey/1/2005 (Ty05) is a Eurasian lineage HPAI H5N1 virus assigned to clade 2.2. It is the prototype isolate from an outbreak of H5N1 infection in Turkey in 2005 to 2006 in which there were eight confirmed human cases, including four fatalities (107). Finally, polymerases from two human strains were tested: a seasonal H3N2 strain frequently used in laboratory studies, i.e., A/Victoria/3/75 (Vic), and the already- mentioned pH1N1 virus Eng195. The capacities of this panel of influenza virus polymerases to replicate a minigenome in human 293T, swine NPTr, and avian DF-1 cells are presented in Fig. 2A, B, and C, respectively. The data have not been normalized, for example, to expression of a reporter gene such as the Renilla luciferase gene from a cotransfected plasmid, because this can skew results when polymerase genes from different viral sources are compared. In particular, we find that the endonuclease activities encoded by the PA and PA-X genes (108) differ in potency between viral strains (unpublished observations) and that these activities reduce the expression of reporter genes, as previously noted (109). It is also important that the transfection efficiencies of the three cell types used here differ greatly, explaining why the absolute values of firefly luciferase produced in different cells are markedly different. In particular, we and others have previously noted the low transfection efficiency of the DF-1 cell line (37, 39). Nonetheless, it is evident that the patterns of polymerase activity from different viral sources were very similar in human and pig cells but different in the avian cell line. In the two mammalian cell lines, the two typical avian virus strains, 50-92 and Bav, showed low polymerase activities, in contrast to the human-origin H3N2 Vic polymerase, which gave the strongest polymerase activity observed in both human and pig cells. G1 polymerase was more active than either Bav or 50-92 polymerase in human and pig cells, with an activity approaching that of Eng195. Expression of the firefly luciferase reporter gene by the polymerase derived from the HPAI H5N1 Ty05 virus was stronger than expression driven by the G1 or Eng195 polymerase in human or pig cells. In avian cells, the differences between the influenza virus polymerase activities were not dramatic, and the pattern was different. For example, the H9N2 G1 polymerase was the most active, followed by the Eng195 and Vic polymerases. The difference between the 50-92 and Vic polymerases was only 3.3-fold in avian DF-1 cells, whereas 50- and 98-fold differences were observed in swine NPTr cells and human 293T cells, respectively. Taken together, these data imply that a host range barrier at the level of polymerase activity for typical avian influenza viruses exists in pig cells, just as it does in human cells. virus polymerase activity in pig cells. In order to gain some in- sights into the nature of the genetic changes that mediate efficient replication of a typical avian-origin influenza virus polymerase in swine, we first analyzed the polymorphisms at known host range- determining residues of the PB2 component of influenza virus polymerases that are endemic in different hosts, including pigs. The polymorphism frequencies in PB2 at positions 271, 590-591, 627, and 701 were estimated from a sample of complete genome sequences downloaded from the NCBI influenza virus resource (110). A total of 2,259 PB2 sequences were analyzed (see Tables S1 and S2 in the supplemental material for details). The amino acids at these positions in the strains used in the present study are shown in Table 1. The most studied of these PB2 polymorphisms is undoubtedly the PB2 E627K mutation. The only swine lineage of influenza virus that contains PB2 627K is the classical swine lineage that was reintroduced in 1918 into pigs from humans rather than evolving naturally in pigs via an avian source (Fig. 3). In contrast, more than 30 years after introduction of the Eurasian avian-like H1N1 lineage from birds directly to pigs, the PB2 E627K mutation has not been naturally selected. Similarly, 627E is found in PB2 sequences from swine influenza viruses that have the TRIG cassette, in which PB2 and PA originated from a North American avian- origin virus 14 years ago (70, 71). This suggests that there is not a strong selection pressure for 627K to emerge in pigs, in contrast to other mammalian species. To test whether this was because 627K does not affect polymerase function in pig cells, the E627K mutation was engineered by site-directed mutagenesis in PB2 genes from the typical host-restricted avian virus polymerases, the Bav and 50-92 polymerases. Activities of polymerase complexes containing WT PB2 or PB2 with a mutation at position 627 were compared in human, swine, and avian cells (Fig. 4A, B, and C). To demonstrate that the host range mutations did not affect expression of the PB2 mutant proteins, the mutations were additionally engineered into an epitope-tagged 50-92 PB2 gene and expressed individually in each cell type. Western blot analysis using antibody to the C-terminal epitope tag (Flag tag) showed that none of the host range mutations affected the accumulation of PB2 protein in any of the cell types (see Fig. S1 in the supplemental material). This result is in line with several other publications that also show that mutations that affect polymerase activity, such as those at PB2 position 627, do not alter the expression or accumulation of the protein (35, 37, 69, 111). As previously reported (34, 35, 37–40), the introduction of 627K into avian-origin viral polymerases dramatically increased expression of the reporter gene in human cells (Fig. 4A). Interestingly, 627K also significantly increased avian virus polymerase activity in pig cells ( P Ͻ 0.5 by unpaired Student’s t test) (Fig. 4B). Compared to the respective WT polymerases, increases of about 12-fold and 10-fold were observed for Bav E627K and 50-92 E627K polymerases, respectively, meaning that they approached the level of activity of the Eng195 polymerase constellation in pig cells (Fig. 2B). The activity of the H9N2 G1 polymerase was also increased even further in pig cells (see Fig. S2 in the supplemental material) and human cells (data not shown) when the E627K mutation was introduced. Conversely, a reduction of activity in both cell types was observed when the human-adapted Vic PB2 or Ty05 PB2 was mutated (K627E) (see Fig. S2). Similar to the E627K mutation, the D701N adaptive mutation has been reported to emerge when avian influenza viruses are passaged in animals such as mice (36, 41, 59, 112, 113). The D701N mutation in PB2 is associated with enhanced polymerase activity in human cells (35, 36, 113–115). During the natural evolution of influenza virus, this mutation has been selected in the Eurasian swine lineage only (Fig. 3). Indeed, 701N is present in the PB2 proteins of early isolates of the Eurasian avian-like H1N1 swine viruses (for example, A/Swine/Germany/2/1981) but not in those from the supposed avian precursor, A/Duck/Bavaria/1/77. The effect of this mutation on the activity of two avian virus polymerases, the Bav and 50-92 polymerases, was tested in human and pig cells (Fig. 4A and B, respectively). In both cell types, D701N mutation resulted in increased polymerase activity, to a lesser ex- tent than that with the E627K mutation, but the difference was still statistically significant ( P Ͻ 0.01 by unpaired Student’s t test) compared to the wild-type avian polymerase activities. Thus, in pig cells, the luciferase signal from Bav polymerase with the D701N mutation was 2.7-fold higher than that from WT Bav polymerase. The D701N mutation increased 50-92 polymerase activity 4.8-fold in pig cells. In avian cells, neither the E627K nor D701N mutation increased the polymerase activity of either of the avian viruses tested (Fig. 4C). PB2 residues 627K and 701N are absent in swine triple-reassortant viruses and also in the descendant 2009 pH1N1 viruses (Fig. 3; see Table S1 in the supplemental material). However, the G590S/Q591R and T271A mutations present in the TRIG cas- sette-encoded PB2 protein can compensate for this absence (34, 69). These mutations might have been selected and maintained in pigs because they confer a specific replicative advantage in this species. The T271A mutation is also found in seasonal human viruses, in conjunction with the E627K switch, but it did not emerge in the classical swine lineage (Fig. 3). Sequences encoding the G590S/Q591R and T271A mutations were introduced individually into the 50-92 PB2 gene, and minigenome assays were performed to assess the consequences of these changes on polymerase activity in human, swine, and avian cells (Fig. 4A, B, and C). The results indicate that G590S/Q591R and T271A mutations individually increased the activity of avian-origin 50-92 polymerase ...

Context 4

... in swine, we first analyzed the polymorphisms at known host range- determining residues of the PB2 component of influenza virus polymerases that are endemic in different hosts, including pigs. The polymorphism frequencies in PB2 at positions 271, 590-591, 627, and 701 were estimated from a sample of complete genome sequences downloaded from the NCBI influenza virus resource (110). A total of 2,259 PB2 sequences were analyzed (see Tables S1 and S2 in the supplemental material for details). The amino acids at these positions in the strains used in the present study are shown in Table 1. The most studied of these PB2 polymorphisms is undoubtedly the PB2 E627K mutation. The only swine lineage of influenza virus that contains PB2 627K is the classical swine lineage that was reintroduced in 1918 into pigs from humans rather than evolving naturally in pigs via an avian source (Fig. 3). In contrast, more than 30 years after introduction of the Eurasian avian-like H1N1 lineage from birds directly to pigs, the PB2 E627K mutation has not been naturally selected. Similarly, 627E is found in PB2 sequences from swine influenza viruses that have the TRIG cassette, in which PB2 and PA originated from a North American avian- origin virus 14 years ago (70, 71). This suggests that there is not a strong selection pressure for 627K to emerge in pigs, in contrast to other mammalian species. To test whether this was because 627K does not affect polymerase function in pig cells, the E627K mutation was engineered by site-directed mutagenesis in PB2 genes from the typical host-restricted avian virus polymerases, the Bav and 50-92 polymerases. Activities of polymerase complexes containing WT PB2 or PB2 with a mutation at position 627 were compared in human, swine, and avian cells (Fig. 4A, B, and C). To demonstrate that the host range mutations did not affect expression of the PB2 mutant proteins, the mutations were additionally engineered into an epitope-tagged 50-92 PB2 gene and expressed individually in each cell type. Western blot analysis using antibody to the C-terminal epitope tag (Flag tag) showed that none of the host range mutations affected the accumulation of PB2 protein in any of the cell types (see Fig. S1 in the supplemental material). This result is in line with several other publications that also show that mutations that affect polymerase activity, such as those at PB2 position 627, do not alter the expression or accumulation of the protein (35, 37, 69, 111). As previously reported (34, 35, 37–40), the introduction of 627K into avian-origin viral polymerases dramatically increased expression of the reporter gene in human cells (Fig. 4A). Interestingly, 627K also significantly increased avian virus polymerase activity in pig cells ( P Ͻ 0.5 by unpaired Student’s t test) (Fig. 4B). Compared to the respective WT polymerases, increases of about 12-fold and 10-fold were observed for Bav E627K and 50-92 E627K polymerases, respectively, meaning that they approached the level of activity of the Eng195 polymerase constellation in pig cells (Fig. 2B). The activity of the H9N2 G1 polymerase was also increased even further in pig cells (see Fig. S2 in the supplemental material) and human cells (data not shown) when the E627K mutation was introduced. Conversely, a reduction of activity in both cell types was observed when the human-adapted Vic PB2 or Ty05 PB2 was mutated (K627E) (see Fig. S2). Similar to the E627K mutation, the D701N adaptive mutation has been reported to emerge when avian influenza viruses are passaged in animals such as mice (36, 41, 59, 112, 113). The D701N mutation in PB2 is associated with enhanced polymerase activity in human cells (35, 36, 113–115). During the natural evolution of influenza virus, this mutation has been selected in the Eurasian swine lineage only (Fig. 3). Indeed, 701N is present in the PB2 proteins of early isolates of the Eurasian avian-like H1N1 swine viruses (for example, A/Swine/Germany/2/1981) but not in those from the supposed avian precursor, A/Duck/Bavaria/1/77. The effect of this mutation on the activity of two avian virus polymerases, the Bav and 50-92 polymerases, was tested in human and pig cells (Fig. 4A and B, respectively). In both cell types, D701N mutation resulted in increased polymerase activity, to a lesser ex- tent than that with the E627K mutation, but the difference was still statistically significant ( P Ͻ 0.01 by unpaired Student’s t test) compared to the wild-type avian polymerase activities. Thus, in pig cells, the luciferase signal from Bav polymerase with the D701N mutation was 2.7-fold higher than that from WT Bav polymerase. The D701N mutation increased 50-92 polymerase activity 4.8-fold in pig cells. In avian cells, neither the E627K nor D701N mutation increased the polymerase activity of either of the avian viruses tested (Fig. 4C). PB2 residues 627K and 701N are absent in swine triple-reassortant viruses and also in the descendant 2009 pH1N1 viruses (Fig. 3; see Table S1 in the supplemental material). However, the G590S/Q591R and T271A mutations present in the TRIG cas- sette-encoded PB2 protein can compensate for this absence (34, 69). These mutations might have been selected and maintained in pigs because they confer a specific replicative advantage in this species. The T271A mutation is also found in seasonal human viruses, in conjunction with the E627K switch, but it did not emerge in the classical swine lineage (Fig. 3). Sequences encoding the G590S/Q591R and T271A mutations were introduced individually into the 50-92 PB2 gene, and minigenome assays were performed to assess the consequences of these changes on polymerase activity in human, swine, and avian cells (Fig. 4A, B, and C). The results indicate that G590S/Q591R and T271A mutations individually increased the activity of avian-origin 50-92 polymerase in both human and swine cells (4.5- and 3-fold, respectively, in human cells and 2.4- and 2.2-fold, respectively, in pig cells). Polymerase activities in DF-1 cells were not significantly affected by these mutations, in agreement with previous publications (34, 69). Thus, the four PB2 mutations tested (E627K, D701N, G590S/ Q591R, and T271A), which adapt avian influenza virus polymer- ase for human cells, also increased activity in pig cells but not in avian cells. PB2 E627K enhances virus replication in pig cells. To test that the results obtained with the in vitro polymerase assay reflected a difference of replication in the context of infectious virus, we generated a set of recombinant viruses by reverse genetics. Viruses that contained wild-type A/Turkey/England/50-92/91 or A/Duck/ Bavaria/1/77 polymerase and NP genes or encoded the mammalian adaptive mutation E627K in PB2 were produced. To elimi- nate host restrictions that mapped to cell entry, interferon response, or vRNP export, we replaced the HA and NA surface protein genes and the NS and M genes of each virus with those from the vaccine strain PR8. PR8-based recombinant viruses have been used to experimentally infect minipigs (116). The multicycle replication of the viruses was compared in swine NPTr cells. Both PB2 E627K-containing viruses grew to higher titers than their wild-type counterparts; the differences in titer were statistically significant except at the last time point, i.e., 48 h postinfection (Fig. 5A). Moreover, Western blot analysis showed that viral gene expression, as exemplified by the M1 protein, was increased in the cells infected with PB2 E627K-expressing viruses over those infected by viruses with wholly avian polymerase constellations, and this was especially evident early during infection (Fig. 5B). These differences confirmed a growth advantage resulting from increased polymerase activity for PB2 E627K-containing viruses in vitro in swine cells, as has been seen in human or simian cells. Pigs are thought to play an important role in influenza virus transmission between birds and humans. They have been proposed to be an intermediate host for the adaptation of avian influenza vi- ruses to humans and for the generation of reassortant viruses with pandemic potential (reviewed in reference 16). Pigs are naturally susceptible to infection with at least some avian influenza viruses, and avian-origin viruses have been isolated from pigs worldwide (reviewed in references 8 and 9). Thus, one might anticipate that pig cells would be more permissive to avian-origin influenza virus polymerase than human cells are. Moreover, it may be that pigs express cofactors for avian influenza virus polymerase that are not present or are different in other mammalian species, such as humans. Using an influenza virus polymerase assay, we investigated the ability of six influenza virus polymerase complexes from different species (turkey, duck, quail, and human) to amplify and express a reporter minigenome in swine cells. Surprisingly, our results indicated that pig and human cells have similar restrictions in supporting the function of influenza virus polymerases from different origins. Low activity was observed in pig cells for polymerases from the two classical avian strains A/Turkey/England/ 50-92/91 (50-92) and A/Duck/Bavaria/1/77 (Bav). Although pigs seem to support the growth of many typical avian influenza viruses, Kida et al. showed that not all strains replicated or induced a serological response in experimentally infected pigs. In addition, the large majority of avian viruses replicated to a lower level in pigs than a swine-adapted virus did (10). In the field, only a minority of avian influenza virus strains seem to naturally spread in the pig population. This suggests that avian influenza viruses with the potential to efficiently replicate and transmit in pigs are rare. Based on our polymerase assay, some candidates might be viruses from the H9N2 G1 lineage or the H5N1 Eurasian lineage Z genotype, such as Ty05 virus, that already bear the E627K mutation. H9N2 ...