|Home | About | Journals | Submit | Contact Us | Français|
The 2009 pandemic influenza virus (pH1N1) is a swine-origin reassortant containing human, avian, and swine influenza genes. We have previously shown that the polymerase complex of the pH1N1 strain A/California/04/2009 (Cal) is highly active in mammalian 293T cells, despite the avian origin of both its PA and PB2. In this study, we analyzed the polymerase residues that are responsible for high pH1N1 polymerase activity in the mammalian host. Characterization of polymerase complexes containing various combinations of Cal and avian influenza virus A/chicken/Nanchang/3-120/01 (H3N2) (Nan) by reporter gene assay indicates that Cal PA, but not PB2, is a major contributing factor to high Cal polymerase activity in 293T cells. In particular, Cal PA significantly activates the otherwise inactive Nan polymerase at 37 and 39°C but not at the lower temperature of 34°C. Further analysis using site-directed mutagenesis showed that the Cal PA residues 85I, 186S, and 336M contribute to enhanced activity of the Cal polymerase. Recombinant A/WSN/33 (H1N1) (WSN) viruses containing Nan NP and polymerase (PA, PB1, PB2) genes with individual mutations in PA at residues 85, 186, and 336 produced higher levels of viral protein than the virus containing wild-type (WT) Nan PA. Interestingly, compared to the WT, the virus containing the 85I mutation grew faster in human A549 cells and the 336M mutation most significantly enhanced pathogenicity in a mouse model, among the three PA mutations tested. Our results suggest that multiple mutations in PA, which were rarely present in previous influenza isolates, are involved in mammalian adaptation and pathogenicity of the 2009 pH1N1.
Influenza pandemics emerge when a new virus to which the population has little or no preexisting immunity spreads in humans. Four pandemics have occurred in the last 100 years, caused by the highly lethal 1918 Spanish influenza, the 1957 Asian influenza, the 1968 Hong Kong influenza, and the 2009 swine influenza viruses (14, 26). The last pandemic affected more than 214 countries and caused over 18,000 deaths (32). Overall, the 2009 pandemic influenza virus strain generally caused a mild, self-limiting illness similar to seasonal influenza virus but showed an unusual mortality pattern (9). The virus is now expected to take on the characteristics of seasonal influenza virus and continue to circulate, and a representative strain has been included in the 2010-2011 influenza virus vaccine (33).
The emergence of new human infections occurs through two distinct mechanisms: direct mutation of avian viruses or an intermediate swine reassortment step. The source of the 2009 pandemic was a Mexican swine-origin reassortant virus of the H1N1 subtype (pH1N1). Several reassortment steps led to the emergence of this virus, which possesses HA, NP, and NS genes of the classical swine lineage, NA and M genes from the Eurasian swine lineage, a human PB1 gene that was seeded from an avian influenza virus in approximately 1968, and avian PA and PB2 genes. This combination of gene segments was not previously reported for either human or swine virus (8).
It is becoming increasingly clear that mutations in avian virus genes are required for human adaptation and emergence of pandemic influenza viruses. The viral proteins HA and PB2 have been extensively studied for their contribution to host range. It is well established that preferential binding of HA to α-2,3 or α-2,6 receptors is a major contributing factor to host adaptation, pathogenicity, and transmission (27). In addition to HA, the PB2 component of the viral polymerase complex is known to be important for human host adaptation. A single mutation at PB2 residue 627 significantly affects polymerase activity and replication in mammalian cells (30). In experimental mouse infections with avian H5N1 viruses, the PB2 E627K mutation strongly affects viral pathogenicity (10). It is still unclear how a single mutation at PB2 residue 627 determines viral pathogenicity, although the difference in surface charge of the 627E- or 627K-containing domain of PB2 may affect PB2 interactions with other viral or cellular proteins (16, 31). Another mutation in PB2, D701N, also enhances polymerase activity in mammalian cells and increases virus pathogenicity in mice (6), which could be due to enhanced binding of PB2 to importin α in mammalian cells (7). In addition to PB2 residues 627 and 701, we have recently found that the human residue PB2 271A, a highly conserved host-specific amino acid, contributes to enhanced polymerase activity and viral growth both in vitro and in vivo (2).
The 2009 pH1N1 replicates and transmits well in mammalian species, despite its avian-origin PB2 without 627K or 701N. However, the presence of 591R, which is located near residue 627, alters the surface shape and charge of PB2 and is thought to compensate for the lack of 627K in the 2009 pH1N1 (19, 34). Furthermore, the pH1N1 contains 271A, a human-virus-conserved residue, which contributes to the enhanced polymerase activity of the pH1N1 in human cells (2). In addition to PB2, the 2009 pH1N1 PA is also avian-like, as it contains 7 unique avian virus residues (28P, 55D, 57R, 65S, 100V, 312K, and 552T) and only 3 human virus residues (356R, 382D, and 409N), although the gene was maintained in swine viruses for more than a decade. The contribution of polymerase components besides PB2 to human adaptation of pH1N1 has not been studied in detail. To further analyze the role of the polymerase genes in mammalian adaptation of pH1N1, we compared the activity of polymerase complexes containing various combinations of components from pH1N1 and avian or human viruses. We found that, unexpectedly, the pH1N1 PA gene is key to high pH1N1 polymerase activity in mammalian cells. Further analysis using site-directed mutagenesis identified several pH1N1 residues that enhance avian polymerase activity in mammalian cells. Our results indicate that in addition to PB2, mutations in PA can also significantly affect viral polymerase activity in mammalian hosts.
Influenza viruses A/chicken/Nanchang/3-120/01 (H3N2) (Nan) (18) and A/California/04/2009 (H1N1) (Cal) were provided by R. Webster and R. Webby (St. Jude Children's Research Hospital, Memphis, TN). Influenza virus A/Aichi/2/68 (H3N2) (Aichi) was provided by A. Klimov (Centers for Disease Control and Prevention, Atlanta, GA). Influenza virus A/WSN/33 (H1N1) (WSN) was rescued from cDNAs provided by Y. Kawaoka (University of Wisconsin, Madison, WI). Aichi, Cal, and WSN were propagated in MDCK cells, and Nan was propagated in 10-day-old embryonated chicken eggs (2). MDCK, 293T, and A549 cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 8% fetal calf serum (FCS).
The anti-influenza A virus NP monoclonal antibody (MAb) (clone IC5-1B7, NR-4544) was obtained from BEI Resources. The anti-influenza A matrix protein MAb (GA2B) and the anti-β-actin antibody (AC-15) were purchased from GeneTex (Irvine, CA) and Sigma-Aldrich (St. Louis, MO), respectively.
Synthesis of pCAGGS-NP, -PA, -PB1, and -PB2 for Aichi, Nan, and Cal and pPolI vectors for Nan NP, PA, PB1, and PB2 were previously described (2). pPolI and pCAGGS vectors for rescue of WSN were provided by Y. Kawaoka. Chimera PA genes were created using compatible restriction sites in Nan and Cal PA and cloned into pCAGGS. Mutations in PA genes were created using the QuikChange II site-directed mutagenesis kit (Agilent Technologies, Santa Clara, CA). pPolI-NP-Luc was provided by T. Wolff (Robert-Koch Institute, Berlin, Germany), and pRL-SV40 (Promega, Madison, WI) was provided by L. Martínez-Sobrido (University of Rochester).
Recombinant adenovirus constructs for polymerase genes from influenza virus A/Vietnam/1203/04 (H5N1) (VN1203) have been previously described (1). The human codon-optimized polymerase genes of Cal and influenza virus A/Brevig Mission/1/1918 (H1N1) (1918) were synthesized by GeneArt and Epoch Biolabs, respectively. The recombinant adenovirus constructs expressing PB2, PB1, PA, or NP from Cal and 1918 were created using the AdEasy system (Stratagene, La Jolla, CA) as previously described (1). Briefly, the Cal constructs were created from pShCMV by homologous recombination into AdEasy and propagation in HEK293A cells. For the 1918 constructs, the pAdenoVator-CMV(CuO) plasmid was used, followed by recombination into AdEasy and propagation into QBI-HEK 293CymR cells (Qbiogene, Montreal, Canada). The viruses were purified by CsCl density-gradient centrifugation, and viral titers were determined by quantitative real-time PCR.
Polymerase activity assays in 293T cells were performed as described previously (2). Briefly, 293T cells in a 12-well tissue culture plate were cotransfected with 0.4 μg each of pCAGGS-NP, -PA, -PB1, and -PB2 and 0.1 μg each of pPolI-NP-Luc and pRL-SV40 using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) for 24 h at 34, 37, and 39°C. Luciferase production was measured using the dual-luciferase reporter assay system (Promega, Madison, WI). Polymerase activity was normalized to Renilla expression. All results shown are the averages with standard deviations from three independent experiments. Data were analyzed using GraphPad Prism 5. Unless otherwise indicated, statistical analysis used one-way repeated-measures analysis of variance followed by Tukey's posttest.
A549 cells in a 24-well plate were infected with the respective recombinant adenoviruses at a multiplicity of infection (MOI) of 1,000, expressing rAd5-PA-TAP, -PB1, -PB2, -NP and -pPolI-NP-Luc. For PA from VN1203, an MOI of 100 was used. At 48 h postinfection (hpi), polymerase activity was assayed using the luciferase assay system (Promega). Data were analyzed using GraphPad Prism 5. Equal protein expression levels for all three polymerase (PA, PB1, PB2) genes were confirmed by Western blot analysis.
The recombinant virus WSN-Nan NP/PA/PB1/PB2 has HA, NA, M, and NS genes from WSN and NP, PA, PB1, and PB2 genes from Nan (2). Four viruses with single mutations in Nan PA, T85I, G186S, L336M, and R213K, were rescued as described previously (2). The PA gene of the viruses was sequenced for confirmation. Stocks were prepared in MDCK cells, and infectious titers were determined in MDCK cells using the method of Reed and Muench (25).
To determine the effect of PA mutations on viral protein synthesis, A549 cells were infected at an MOI of 3 and cultured at 37°C. At 3, 6, 9, and 12 h postinfection, total cell lysates were prepared in RIPA buffer. Clarified lysates were separated on a 4 to 12% NuPAGE Bis-Tris acrylamide gel (Invitrogen) in MOPS (morpholinepropanesulfonic acid) running buffer. Proteins were transferred to polyvinylidene difluoride (PVDF) membranes, and NP, M1, and actin proteins were detected by Western blotting using specific antibodies.
To determine the growth of recombinant viruses in vitro, MDCK or A549 cells were infected at an MOI of 0.01 for 1 h and incubated at 34, 37, or 39°C in DMEM containing 0.15% bovine serum albumin (BSA) and 1 μg/ml l-(tosylamido-2-phenyl) ethyl chloromethyl ketone (TPCK)-treated trypsin. At the indicated times postinfection, 10% of the supernatant was removed and stored at −80°C and then replaced with an equal volume of fresh medium. Viral titers were determined in MDCK cells. Data were analyzed using GraphPad Prism 5, and statistical differences were determined using the unpaired t test.
To determine morbidity and mortality, 6-week-old BALB/c female mice (Taconic Farms Inc., Germantown, NY) were anesthetized with Avertin and infected intranasally with 2 × 105 TCID50 (50% tissue culture infectious doses) of recombinant viruses in 30 μl. Body weight was measured daily for 14 days, and mice were observed for signs of illness. To determine virus replication in infected mouse lungs, groups of mice were infected with 2 × 105 TCID50 of the indicated virus, and on days 2, 4, and 6, 3 or 4 mice were humanely sacrificed, and their lungs were removed and homogenized in 1 ml of Dulbecco's phosphate-buffered saline (PBS). The clarified supernatant was stored at −80°C until analysis. Viral titers in lung homogenates were determined in MDCK cells. The experiment was repeated for a minimum of 6 mice per virus per time point. Data were analyzed using GraphPad Prism 5, and statistical differences were determined by unpaired t test. For pathological analysis, lungs were perfused with 10% formalin, fixed, and embedded in paraffin. Four-micrometer tissue slices were prepared and stained with hematoxylin and eosin. Images were taken by using an Olympus inverted microscope. All animal experiments were approved by the University of Rochester Committee on Animal Resources.
Polymerase complexes containing an avian PB2 gene function poorly in mammalian cells, but activity can be restored by human adaptive mutations (2, 17, 22). The 2009 pH1N1 contains avian-like PA and PB2 genes but is highly active in mammalian cells (2). To further determine the role of each polymerase segment in the high pH1N1 polymerase activity in mammalian cells, we used a reporter gene assay to analyze the activity of mixed polymerase complexes containing the pH1N1 strain Cal and avian Nan at three different temperatures. The temperatures 34, 37, and 39°C represent the temperatures in the upper respiratory tract in humans, the human deep lung, and the avian intestinal tract, respectively (Fig. 1A). The Cal polymerase complex (Fig. 1A, lane 1) was 30-, 157-, and 132-fold more active at 34, 37, and 39°C, respectively, than the Nan polymerase complex (lane 6) (P < 0.001). Unexpectedly, the Cal polymerase complex containing Nan PB2 (lane 2) maintained high activity at higher temperatures (37 and 39°C) but not at the lower temperature of 34°C. The substitution decreases polymerase activity 25-fold at 34°C but only 2.5- and 2.0-fold at 37 and 39°C compared to that of the Cal complex (P < 0.001 at 34 and 37°C, P < 0.05 at 39°C). Our previous study using the same system showed that Nan PB2, which has 627E, abolished polymerase activity of the human Aichi polymerase at all 3 temperatures, 34, 37, and 39°C, in 293T cells (2). This result indicates that a Cal polymerase component other than PB2 is responsible for the high Cal polymerase activity in 293T cells. Replacement of Cal PB1 or NP with that of Nan did not significantly affect Cal polymerase activity at any temperature (lanes 3 and 5). However, the Cal polymerase containing Nan PA (lane 4) showed significantly decreased activity at all temperatures: 29-, 38-, and 39-fold reduction at 34, 37, and 39°C, respectively (P < 0.001). The strong contribution of Cal PA was confirmed by the characterization of Nan polymerase complexes containing each component of Cal. As described above, the Nan polymerase activity in 293T cells was very low (lane 6). Replacement of PB1 or NP with Cal did not significantly affect Nan polymerase activity (lanes 8 and 10, respectively). The substitution of PB2 (lane 7) slightly increased activity (1.8-, 7.2-, and 6.4-fold at 34, 37, and 39°C, respectively). In contrast, substitution with Cal PA (lane 9) strongly enhanced the Nan polymerase activity at 37 and 39°C, increasing activity 157- and 132-fold, respectively (P < 0.001). Interestingly, this enhancement was not observed at lower temperatures. However, the addition of Cal PB2 restored low-temperature polymerase activity to a level higher than that of the all-Cal complex (lane 12). These results indicate that Cal PA alone can strongly enhance avian polymerase activity at high temperatures but requires Cal PB2 to be highly active at 34°C.
To further characterize the unique contribution of Cal PA to polymerase activity in mammalian cells, we analyzed the activity of complexes containing Cal and a human H3N2 isolate, Aichi, in 293T cells (Fig. 1B). The Cal polymerase complex was as active as Aichi at 37 or 39°C but 6.3-fold less active at 34°C (Fig. 1B, lanes 1 and 6). The addition of Aichi NP (lane 5), PB1 (lane 3), or PB2 (lane 2) to an otherwise Cal complex did not significantly affect polymerase activity, although Aichi PB1 and PB2 slightly enhanced activity at 34°C (about 2.7-fold). Interestingly, the presence of Aichi PA in an otherwise Cal complex (lane 4) strongly decreased activity at all 3 temperatures, suggesting the presence of unique mutations in Cal PA which are required for function of the other Cal polymerase components. Results of gene replacement of Aichi polymerase components with those of Cal indicated that Cal PA functions well with other Aichi components (lane 9). However, Cal PB2, with the remaining components from the Aichi complex, strongly attenuated polymerase activity at all temperatures (lane 7). Cal PB1 and NP only slightly affected the polymerase activity of the Aichi complex. These results suggest that Cal PA contains key residues which are required for the function of the other Cal components and that these residues do not exist in Nan or Aichi PA.
We next determined the contribution of Cal PA to polymerase activity in human lung epithelial A549 cells, which are more closely related to the natural site of infection. Because of low transfection efficiency, we used adenovirus-mediated transduction of A549 cells as we reported previously (1). Similar to our results in 293T cells, the Cal complex was highly active in A549 cells (Fig. 1C, lane 1). In both cell types, the Cal polymerase was most active at 37°C but was about 9- to 13-fold less active at 34°C.
As shown, PA from the nonpathogenic avian virus Nan or the conventional human virus Aichi did not function well with the other Cal polymerase components (Fig. 1A and B). We further characterized the compatibility of PA and the other components of the highly pathogenic H5N1 virus VN1203 and the 1918 Spanish influenza virus (Fig. 1C and D) with the Cal polymerase. Both the VN1203 and 1918 polymerases are derived from avian viruses and have PB2 with 627K, but the contribution of other components to polymerase activity in mammalian cells has not been analyzed. Similar to the results with Nan and Aichi shown in Fig. 1A and B, Cal PA was fully functional with the other components of VN1203 or the 1918 virus (Fig. 1C and D, lane 9). Also, the Cal polymerase with VN1203 or the 1918 virus PA functioned very poorly in A549 cells (Fig. 1C and D, lane 4), as observed with Nan or Aichi PA with Cal components in 293T cells (Fig. 1A and B, lane 4). These results indicate that specific Cal PA mutations that are required for Cal polymerase activity do not exist even in the avian-origin highly pathogenic VN1203 or 1918 virus PA proteins.
Replacement of PA in the Nan polymerase complex with that of Cal increased activity over 100-fold at 37 and 39°C (Fig. 1A, lanes 6 and 9), showing a strong impact of Cal PA on an avian polymerase complex, even with a 627E-containing PB2, in mammalian cells. Although the pH1N1 (Cal) PA is avian in origin, sequence analysis identified 20 amino acid differences between the Cal and Nan PA proteins (Table 1). To identify the residues in Cal PA that affect avian polymerase activity, we first created and characterized Nan/Cal chimera PA constructs to determine the region of Cal PA responsible for high polymerase activity in mammalian cells (Fig. 2). Five chimera PAs were constructed (Fig. 2A), and their activities with either Cal (Fig. 2B) or Nan (Fig. 2C) components were determined in 293T cells using a plasmid-based reporter gene assay. Chimeras 1 and 3, which contain the N-terminal 334 residues from Nan, were inactive with both the Cal and Nan polymerase complexes, as observed with wild-type (WT) Nan PA. Chimeras 2, 4, and 5, which contain the N-terminal 334 residues from Cal, were highly active in both the Cal and Nan complexes. However, chimera 5, which contains the N-terminal 463 residues from Cal, was more active than chimera 2 and almost as active as WT Cal, suggesting that multiple residues in Cal PA contribute to enhanced activity.
Because 18 of the 20 amino acid differences between the Nan and Cal PA genes lie in the region 1 to 463 (Table 1), we chose to make amino acid substitutions at all positions in the Cal PA gene and introduced the corresponding Nan residues. None of the single mutations in Cal PA significantly reduced Cal polymerase activity. However, the mutations I85T, S186G, H277S, M336L, R362K, and N409S slightly reduced Cal polymerase activity, between 1.5- and 2.7-fold, in mammalian cells (data not shown). Next, we introduced the corresponding mutations into Nan PA namely, T85I, G186S, S277H, L336M, K362R, and S409N, as well as R213K, which did not affect Cal polymerase activity, for comparison. The effect of the mutations on Nan polymerase activity in 293T cells was measured by reporter gene assay (Fig. 3). Like the inclusion of Cal PA in an otherwise Nan complex, the mutations only slightly affected polymerase activity at 34°C, with up to a 3-fold increase in activity (Fig. 3A). However, at 37°C, the T85I and L336M mutations increased Nan polymerase activity 11.4- and 31.7-fold, respectively (Fig. 3B). At 39°C, the T85I, G186S, and L336M mutations increased polymerase activity 28.5-, 6.4-, and 57-fold, respectively (Fig. 3C). The other mutations (R213K, S277H, K362R, and S409N) did not enhance Nan polymerase activity. These results indicate that multiple residues, especially 85I and 336M, have a strong impact on the contribution of Cal PA to polymerase activity. Interestingly, PA containing multiple mutations of either 85I and 336M or 85I, 186S, and 336M did not further enhance Nan polymerase activity; rather, the activity was lower than that provided by the single mutation at 336M (Fig. 3). These results suggest the presence and involvement of additional residues in Cal PA that are not able to enhance polymerase activity by themselves but contribute to the enhanced activity provided by residues 85I and 336M, possibly by providing structural or functional stability.
To evaluate the role of these PA residues in viral replication, we rescued recombinant WSN viruses containing Nan NP and polymerase genes, with or without PA mutations at positions 85I, 186S, 213K, and 336M. All viruses were successfully rescued. The effect of each mutation on viral protein production was determined by Western blot analysis. Infected cell lysates prepared at 3, 6, 9, or 12 h after infection were analyzed using anti-NP and anti-M1 Ab (Fig. 4). In agreement with the results of the reporter gene assay, viruses containing either 85I, 186S, or 336M in PA produced more viral protein than the WT or 213K virus, as clearly seen in samples prepared at early times after infection (6 hpi). The 85I and 336M mutations most strongly enhanced viral protein production in infected A549 cells (3.6- and 4.5-fold more M1 was produced at 6 hpi than that in the WT, respectively), consistent with the reporter gene assay results in 293T cells.
To determine the effect of the PA mutations on viral growth, multicycle growth kinetics of the recombinant viruses were determined in A549 cells at 34, 37, and 39°C (Fig. 5). Cells were infected at an MOI of 0.01 with either the WT, 85I, 186S, 336M, or 213K virus. The viral titers were determined at various time points after infection. All viruses grew similarly at 34°C (Fig. 5A), as expected from the results of the reporter gene assay, which showed no enhancement at 34°C. At 37°C (Fig. 5B) and 39°C (Fig. 5C), the 85I mutant grew faster and to higher titers than the WT or the other mutant viruses. The viral titers of the 85I and WT viruses were statistically different by unpaired t test at 24 and 36 hpi at 37°C (P = 0.012 and 0.041, respectively) and 24 and 36 hpi at 39°C (P = 0.018 and 0.049, respectively). The 336M virus, however, grew in a manner similar to that of the WT virus.
Next, we determined if mutations at 85I, 186S, and 336M increase viral pathogenicity in a murine model. Six week-old BALB/c mice were infected with 2 × 105 TCID50 of viruses, and body weight was measured daily for 14 days (Fig. 6A). Mice infected with the WT, 85I, and 186S viruses showed similar patterns of minor weight change. In contrast, mice infected with the 336M mutant virus lost up to 17% of their body weight following infection. Viral growth in lungs was determined at days 2, 4, and 6 postinfection in two independent experiments, for a minimum of 6 mice per virus per time point (Fig. 6B). Viral lung titers of 85I and 186S virus-infected mice were not higher than those of WT-infected mice at any time. The titer of the 336M virus was similar to that of the WT on days 2 and 4, but the 336M mutant retained 3.4-fold more virus on day 6 than the WT.
Because the 336M mutant virus maintained a higher viral load for an extended period, resulting in more weight loss than that caused by WT infection, we performed histological analysis of lungs from mice infected with either the WT or 336M virus on day 6 postinfection. Hematoxylin- and eosin-stained lung sections of WT-infected mice showed mild immune infiltrates in the alveolar air spaces and slight thickening of the alveolar walls and bronchioles (×10 and ×20 magnifications are shown in Fig. 7A and B, respectively). In contrast, the 336M virus-infected mice showed increased immune infiltrates in the alveolar air spaces and thickened bronchiolar tissue compared to those of the WT-infected mice (Fig. 7C and D), which is consistent with the observed weight loss and increased viral load in the lungs.
Finally, we determined the presence of the PA residues 85I, 186S, and 336M among various influenza A virus isolates. Over 11,000 full-length PA sequences were available in the Influenza Research Database (IRD) (29) as of 28 February 2011. First, we analyzed the sequence variation of the 2009 pH1N1 PA proteins and found that there is little variability in the PA gene of 2009 pH1N1 isolates. PA 85I, 186S, and 336M were almost completely conserved among the 2009 pH1N1 isolates (Table 2). Interestingly, these residues were rarely found in other influenza A isolates. Among over 8,500 non-pH1N1 viruses, only 15, 65, and 29 isolates have residues 85I, 186S, or 336M, respectively. Importantly, of the available swine influenza virus PA sequences, almost none isolated before the pandemic have these residues. These sequence data may suggest that the key PA residues we identified in this study were introduced recently, just before pH1N1 introduction to the human host.
The 2009 influenza pandemic was due to a swine-origin influenza strain that was a reassortant of human, avian, and swine genes. The viral ribonucleoprotein is composed of a North American avian-derived PA and PB2, a human H3N2-derived PB1, and a classical swine-derived NP (3, 8). In general, the activity of avian polymerase complexes in mammalian cells is very poor, but some mutations are known to enhance activity, including PB2 158G, 627K, and 701N (5–7, 10, 12, 30, 36). The 2009 pH1N1 viruses replicate and transmit efficiently in humans but have the avian-type low-pathogenicity residues 158E, 627E, and 701D in PB2, suggesting the involvement of other residues for efficient replication of pH1N1 in the mammalian host. Two residues in PB2 have been identified which contribute to pH1N1 polymerase activity: residue 591R, which locates close to residue 627 in the three-dimensional structure (19, 34), and residue 271A, which is highly conserved among human viruses and enhances avian polymerase activity in mammalian cells (2). However, in this study using a reporter gene assay, we found that the pH1N1 PA has a stronger impact on polymerase activity than the pH1N1 PB2 in mammalian cells. Although both the PA and PB2 of the pH1N1 virus are of avian virus origin, the avian Nan virus polymerase containing Cal PA was over 20 times more active than the Nan complex containing Cal PB2 at higher temperatures (Fig. 1A, lanes 7 and 9). Interestingly, the human Aichi and the highly pathogenic VN1203 and 1918 virus polymerase components function as efficiently with Cal PA as with their homologous complexes, unlike with Cal PB2, which resulted in significantly reduced activity (Fig. 1). Replacement of Cal PA with that of the Nan, Aichi, VN1203, or 1918 virus resulted in reduced activity of the Cal polymerase complex, indicating that Cal PA contains residues that, together with Cal PB2, complement polymerase activity in the mammalian host.
Enhanced activity of the Nan polymerase complex provided by Cal PA was observed at 37°C and 39°C but not at 34°C (Fig. 1A, lane 9). However, the additional inclusion of Cal PB2 enhanced activity at 34°C (lane 12), suggesting that Cal PB2 contains residues required for the complex to be active at lower temperatures. Because the Aichi, VN1203, and 1918 virus complexes containing Cal PA were active at lower temperatures, it is likely that one or more PB2 residues present in the 2009 pH1N1, seasonal human virus, and highly pathogenic avian viruses is necessary for activity at lower temperatures. In fact, we previously showed that the PB2 E627K mutation strongly activates the Nan polymerase complex even at 34°C (2). Cal PB2 591R, which is considered to compensate for the lack of 627K, may allow for low-temperature activity of the pH1N1. Replacement of Cal PB2 with that of Nan reduced activity slightly at 37 or 39°C (about 2- to 2.5-fold) but strongly at 34°C (25-fold; Fig. 1A, lanes 1 and 2), which clearly indicates that PB2 from a conventional avian virus does not contain the residue(s) required for low-temperature activity. The underlying mechanism for the temperature dependence of the influenza polymerase is unknown, although interactions with host inhibitory or enhancing factors could be an interesting subject for future research (20, 21).
The role of PA in host adaptation is not as well characterized as the role of PB2. However, a recent study indicates the PA mutation T97I increases polymerase activity of an avirulent avian virus in a reporter gene assay, as well as virulence in a mouse model (28). In our study, we have shown that multiple residues, especially 85I, 186S, and 336M in Cal PA, contribute to the enhancement of avian polymerase activity in mammalian cells, which is essential for mammalian host adaptation. The addition of each mutation to Nan PA significantly increased Nan polymerase activity (Fig. 3). However, individual mutations in Cal PA at positions 85, 186, or 336 to Nan residues only slightly decreased Cal polymerase activity, supporting the idea that multiple residues in Cal PA contribute to enhanced Cal polymerase activity in mammalian cells. PA possesses the endonuclease activity required for the “cap-snatching” mechanism (4, 35), and the crystal structures of two domains of PA have been determined (4, 11, 24, 35). The amino acids that increased polymerase activity, 85I, 186S, and 336M, are all surface exposed on their domains. 85I and 186S are located in the domain containing the endonuclease active site. While 186S is located on the opposite side of the structure from the endonuclease active site, it is part of a proposed bipartite nuclear localization signal, residues 124 to 139 and 186 to 247 (23, 35). The residue 85I is located on the surface adjacent to the endonuclease site. The branched, hydrophobic isoleucine at this position in the pH1N1 PA is very different from the polar neutral threonine in most other strains and could potentially affect the endonuclease activity of PA. The residue 336M is located on the opposite end of PA from the PB1-binding site, and no clear function is linked to this residue. However, because of its location at the surface of the molecule, it could be involved in interactions with cellular proteins that inhibit or enhance polymerase activity. It is likely that multiple mutations contribute to the highly active Cal polymerase through multiple complementary unknown mechanisms.
The PA 85I, 186S, and 336M mutations identified by reporter gene assay actually enhanced viral polymerase activity when introduced into viruses, confirming that these mutations in an avian PA can enhance avian virus polymerase activity in mammalian host cells. The 85I and 336M mutations strongly enhanced viral protein synthesis at an early stage of infection (Fig. 4). The virus containing the 85I mutation grew faster and to a higher titer than WT virus at 37 and 39°C, which is consistent with the reporter gene assay results. However, viruses containing the 186S or 336M mutations did not show enhanced multistep growth in A549 cells. While the 336M mutation enhanced polymerase activity more than the 85I mutation as determined by the reporter gene assay or viral protein synthesis, the enhanced polymerase activity by the 336M mutation did not directly reflect the multistep growth of the mutant virus. This result may suggest that the mutation at residue 336 caused a deficiency in a process other than transcription, possibly through unbalanced genome replication, impaired transport of nucleocapsids, or defective virus assembly.
When we examined pathogenicity in vivo, we found that none of the mutations resulted in lethal infection, in sharp contrast to the PB2 627K mutation (2), which indicates that these PA residues are host adaptation but not pathogenicity factors in the mammalian host, as expected from the fact that wild-type pH1N1 viruses are not pathogenic to mice. However, the 336M mutation increased morbidity, as evidenced by weight loss and lung pathology, as well as the lung viral titer at late times after infection. Unexpectedly, the 85I mutation, which increased multistep growth in cultured cells, did not enhance in vivo virus growth or morbidity in mice. These data may support the idea that pathogenicity is not simply reflected by better virus growth as determined in cultured cells but also by the host response, including elevated cytokine release and inflammation, which can then affect lung pathology and virus growth.
Sequence data analysis indicates that few influenza strains contain one or some of the PA residues 85I, 186S, and 336M (Table 2). Although almost all of the pH1N1 viruses contain these three key residues, very few other influenza isolates have these PA residues, regardless of the host. Not one non-pH1N1 strain has all three residues in PA. The majority of the strains containing PA 85I isolated prior to emergence of the 2009 pandemic are avian viruses (13 out of 15 non-pH1N1 viruses). PA 336M can be found in both avian and human isolates. Seventeen human isolates and 12 avian isolates contained PA 336M. Interestingly, multiple highly pathogenic H5N1 human isolates contain PA 336M (A/Hong Kong/483/97, A/Hong Kong/485/97, A/Hong Kong/532/97, and A/Hong Kong/542/97). Although A/Hong Kong/483/97 and A/Hong Kong/485/97 contain PB2 627K, two strains, A/Hong Kong/532/97 and A/Hong Kong/542/97, were pathogenic to mice (15), lack the previously identified virulence markers in polymerase PB2 158G, 271A, 627K, and 701N and PA 97I genes, and have the PA 336M mutation revealed in this study. Some H5N1 viruses isolated from Hong Kong chickens around the same time also contain PA 336M. Further study is needed to determine the contribution of PA mutations to the highly pathogenic phenotype of H5N1 viruses containing PB2 627E. It is possible that the enhancement of polymerase activity by PA 336M contributes to host adaptation and pathogenicity of these H5N1 viruses, although it is unlikely that this single residue explains the highly pathogenic phenotype of the viruses.
It is not clear when and how the PA 85I, 186S, and 336M mutations were introduced into the swine virus PA gene, as surveillance of swine influenza virus prior to the 2009 pandemic was extremely limited. Also, despite efficient viral replication in miniature pigs, infection with the pH1N1 was asymptomatic, which may explain the lack of detection of the virus in pigs before emergence of the pandemic (13). This pandemic highlights the need for continued and expanded influenza virus surveillance, especially in swine, which continue to play a major role in the emergence of pandemic influenza A viruses. It should be noted that identification of the polymerase mutations that affect host adaptation provides critical information for identifying potential future pandemic viruses.
This work was supported by the New York Influenza Center of Excellence (NYICE), a member of the NIAID CEIRS network, under NIH contract HHSN266200700008C. J.L.M. was supported by NIH T32DE007202; E.A.D. was supported by NIH T32AI007362.
We thank A. Klimov, Y. Kawaoka, R. Webster, R. Webby, L. Martinez-Sobrido, and T. Wolff for reagents. We also thank L. Johnstone and the Pathology/Morphology Imaging Core (Department of Laboratory Animal Medicine, University of Rochester School of Medicine and Dentistry) for histology preparation.
Published ahead of print on 11 May 2011.