Amino acids in the polymerase complex of shorebird-isolated H1N1 influenza virus impact replication and host-virus interactions in mammalian models

ABSTRACT

A diverse population of avian influenza A viruses (AIVs) are maintained in wild birds and ducks yet the zoonotic potential of AIVs in these environmental reservoirs and the host-virus interactions involved in mammalian infection are not well understood. In studies of a group of subtype H1N1 AIVs isolated from migratory wild birds during surveillance in North America, we previously identified eight amino acids in the polymerase genes PB2 and PB1 that were important for the transmissibility of these AIVs in a ferret model of human influenza virus transmission. In this current study we found that PB2 containing amino acids associated with transmissibility at 67, 152, 199, 508, and 649 and PB1 at 298, 642, and 667 were associated with more rapid viral replication kinetics, greater infectivity, more active polymerase complexes and greater kinetics of viral genome replication and transcription. Pathogenicity in the mouse model was also impacted, evident as greater weight loss and lung pathology associated with greater inflammatory lung cytokine expression. Further, these AIVs all contained the avian-type amino acids of PB2-E627, D701, G590, Q591 and T271. Therefore, our study provides novel insights into the role of the AIV polymerase complex in the zoonotic transmission of AIVs in mammals.

KEYWORDS:

• Influenza virus
• zoonosis
• polymerase
• H1N1
• avian

Introduction

Wild birds and ducks harbour a wide diversity of avian influenza A viruses (AIVs) and constitute the most important environmental reservoir in terms of AIV subtype diversity, with 16 of the 18 hemagglutinin and 9 of 11 neuraminidase subtypes circulating in this reservoir. Human cases of AIV infection generally do not occur directly from contact with these animals, rather through hosts such as poultry and pigs. This is likely simply due to the lack of human contact with wild avian species, whilst contact of poultry and pigs either with these wild animals or through common contact with water or food is more likely compared to humans. However, AIV infection of these animals can also lead to adaptations. These adaptations can potentially increase or decrease the zoonotic risk of these viruses to humans, particularly in pigs and chickens, respectively. Due to this, less is known about the zoonotic risks to humans posed by AIVs in wild bird reservoirs. However, studies have shown that zoonotic potential for mammalian species may be relatively common. A majority of AIVs isolated from wild birds in these studies show the capacity for mammalian replication in the mouse and ferret models of pathogenesis and transmission, respectively [1–4]. However, considering the size of the viral reservoir and the diversity of AIVs within it, studies to date have provided limited insights. Further, there is limited data on virus-host interactions underlying mammalian infection by AIVs from the wild bird reservoir.

Our aim in this study was to gain insights into virus-host interactions involved in AIV infection in mammals and to study those that may facilitate mammalian transmissibility. We used a group of subtype H1N1 AIVs isolated during long-term surveillance of wild birds at Delaware Bay, USA, an important stop-over site for migratory birds [5]. Our previous studies revealed that the viruses could cause disease in mammalian models and transmit by direct contact in a ferret model of human influenza virus transmission without prior adaptation [2]. Further, some AIVs showed some capacity for airborne transmission (AT) [6, 7]. Full genome analysis of this group of viruses with different AT phenotypes revealed amino acid differences between viruses with different transmissibility potential [8]. Eight amino acids were in the influenza virus polymerase genes PB2 and PB1. In AIVs associated with AT, these amino acids were PB2 V67, S152, T199, Q508, and I649 and PB1 I298, S642, and V667 (Table 1) [6].

Table 1. Identities of amino acids in PB2 and PB1 in DE300 and DE274 associated and not-associated with airborne transmission, in orange and green, respectively. Amino acids at these positions in the viruses and plasmids used in this study are shown.

The influenza virus polymerase complex, comprised of PB2, PB1 and PA, is responsible for cap-snatching, transcription initiation and elongation. The importance of the polymerase genes in the viral lifecycle makes them determinants of viral pathogenicity and viral adaptation to other species [9, 10]. Arguably the best studied example of this is the PB2 E627 K mutation, which enhances the activity of the influenza virus polymerase complex in mammalian cells, enhancing viral replication, transmissibility and pathogenicity [11–14]. However, the amino acids in identified in the PB2 and PB1 of these AIVs have largely not been studied in the context of wild bird AIVs. To study these amino acids in PB2 and PB1 we constructed a reverse-genetics A/ruddy turnstone/Delaware/300/2009 (H1N1) (rgDE300), which showed some capacity for AT in a ferret model. We conducted loss-of-function mutations in PB2 and PB1 to determine the impact of these amino acids on the virus lifecycle, mammalian pathogenicity and host responses to gain mechanistic insights into their potential importance in the zoonotic potential of these AIVs in mammals.

Materials and methods

Cell culture

Madin Darby canine kidney (MDCK) cells and 293 T cells were maintained in minimum essential medium (MEM) and Dulbecco's modified Eagle's medium (DMEM), respectively, supplemented with 10% fetal bovine serum (FBS) (Gibco). Cells were incubated at 37°C (5% CO₂).

Generation of plasmids and reverse genetics virus and sequence analysis

The pHW2000 plasmids encoding the eight segments of A/ruddy turnstone/Delaware/300/2009 (H1N1) and A/shorebird/Delaware/274/2009 (H1N1) were constructed as described previously [8]. The pHW2000 plasmids encoding the eight segments of A/Puerto Rico/8/1934 (H1N1) were provided by Prof. Wenjun Song. V67I, S152A, T199A, Q508R, and I649 V were introduced into DE300 PB2 while I298L, S642N, or V667I were introduced into DE300 PB1 using QuikChange site-directed mutagenesis kit (Agilent). Viruses were rescued using the eight plasmid reverse genetics (rg) system as previously described [15]. Viruses were titrated by plaque assay in MDCK cells or by egg infectious dose 50% (EID₅₀) in ten-day-old embryonated chicken eggs. Plaque morphologies were analyzed using Image J (NIH). All plasmids and viruses were sequenced for confirmation. Analysis of the frequencies of amino acids were performed using the Global Initiative on Sharing All Influenza Data (GISAID) (https://gisaid.org).

Polymerase assay and viral replication kinetics

293 T cells were co-transfected with pHW2000 expressing PB2, PB1, PA, and NP genes at 0.1μg per well, a firefly luciferase reporter (pYH_Luc) at 0.1μg per well and a Renilla luciferase reporter (pRL_TK) at 2.5 ng per well. Cells were harvested 24 h post-transfection and luciferase activity was measured using Dual-Luciferase Reporter Assay System (Promega) using Renilla as a control. Negative controls contained pYH_Luc and pRL_TK without influenza virus genes and the PR8 polymerase complex was used as a positive control. To determine viral replication kinetics, MDCK cells were infected at a multiplicity of infection (MOI) = 0.001. At indicated time points, culture supernatant was harvested and stored at −80°C until titrated by tissue culture infectious dose 50% (TCID₅₀) in MDCK cells. Titres were determined by the Reed and Muench method [16].

Expression of mRNA, cRNA, and vRNA

Expression of RNA species were studied in MDCK cells inoculated at MOI = 1 at 2, 4, 6 and 8 h post inoculation (HPI). RNA was extracted using TRIzol (Invitrogen) according to the manufacturer’s instructions. The Uni-12 and Uni-13 primers were used to detect negative-sense viral RNA (vRNA) and positive-sense complimentary RNA (cRNA) whilst oligo dT was used to detected mRNA (supplementary table 1). cDNAs were generated by reverse transcription of 1μg of RNA with a High Capacity cDNA ReverseTranscription Kit (Thermo) using the following programme; 25°C for 10 min followed by 37°C for 120 min and 85°C for 5 min. qPCR was conducted using SYBR green PCR master mix (Vazyme) and specific primers (supplementary table 1) and the following programme; 1 cycle at 95°C for 30s, followed by 40 cycles of 95°C for 10s and 60°C for 30s, and a final cycle of 95°C for 15s, 60°C for 1 min, and 95°C for 15s using a LightCycler 480 II real-time PCR system (Roche). Expression of each transcript relative to β-actin were calculated by 2^−ΔΔCT. Experiments comprised of three technical and experimental replicates.

Immunofluorescence

MDCK cells were cultured on poly-l-lysine-coated coverslips (Sigma) then inoculated at MOI = 1. At six HPI cells were fixed with ice-cold methanol (Sigma) and stained using an anti-ZO-1 antibody (Invitrogen) then permeabilized using 0.1% (v/v) Triton X-100 in PBS. The secondary antibody Alexa Flour Plus 555 (Invitrogen) conjugate and a mouse anti-influenza A virus nucleoprotein (NP) monoclonal antibody conjugated to FITC (Millipore) were then added. Coverslips were then mounted with ProLong Gold Antifade Mountant with DAPI (Invitrogen) overnight before imaging by confocal microscopy using a Zeiss LSM880. Images were analyzed using Image J.

Western blot

MDCK cells grown in six-well-plates were harvested at two, four, six and eight HPI and lysed in ice-cold RIPA buffer with a protease inhibitor cocktail. Protein was quantified using a Pierce BCA Protein Assay Kit (Thermo Fisher). After blotting on PVDF membranes and blocking, blots were probed with anti-influenza NP antibody (Sino Biological) followed by horseradish peroxidase-conjugated anti-mouse IgG (ABclonal) and SuperEnhanced ECL chemiluminescence substrate was used to develop signal (GBCBIO Technologies). Anti-β-actin antibody (CST) was used as an internal control for protein loading. Band intensities were measured using Image J from three independent experiments.

Mouse experiments

All animal experiments were approved by the first Affiliated Hospital of Guangzhou Medical University Committee on Animal Resources. Six to eight-week-old female BALB/c mice were used. Mice were provided food and water ad libitum. Mice were anaesthetized with isoflurane and inoculated with 3 × 10⁴ EID₅₀ of virus in 30 µL of PBS intranasally and were monitored daily for signs of disease, including weight loss, at least daily for 14 days post inoculation (DPI). Mice were euthanized if humane endpoints were reached (≥25% loss of starting body weight or severe clinical symptoms). To measure virus titres, five mice were euthanized three and six DPI and lungs removed and homogenized in 1 ml of MEM. Clarified supernatants were stored at −80°C until titration by TCID₅₀. Two independent experiments were performed. For histology studies, three mice per group were euthanized three and six DPI. Lungs were perfused and fixed with 10% formalin prior to sectioning and staining.

Five mice per group were euthanized at three DPI to study cytokine mRNA expression. Lungs were homogenized using a TissueLyser II (QIAGEN) and TriZOL (Invitrogen) according to the manufacturer’s instructions. After gDNA removal, 1 µg total RNA was reverse transcribed using random primers (Vazyme). cDNA was diluted tenfold and qRT-PCR was performed on a LightCycler 480 II real-time PCR system (Roche) using 2×PCR master mix (Vazyme) and specific primers (supplementary table 1). Relative mRNA expression was normalized to GAPDH using the 2^−ΔΔCt method. Experiments consisted of two technical replicates.

Homology modelling

The crystal structures of the Influenza A virus polymerase (Protein Data Bank (PDB) code 4WSB) and the electron microscopy structures of the A/little yellow-shouldered bat/Guatemala/060/2010 (H17N10) polymerase complex complexed with RNA (PDB code 6T0 V) were used as model templates [17, 18]. For homology modelling, amino acids were three-dimensionally aligned with the crystal structures via the Phyre2.o server (http://www.sbg.bio.ic.ac.uk/phyre2). Images were generated using PyMOL.

Statistical analyses

Differences between groups were assessed using analysis of variance (ANOVA) or Student’s t-tests using GraphPad Prism v8. P values <0.05 were considered statistically significant.

Results

Amino acids were in regions associated with polymerase complex assembly and RNA binding

PB2 67, 152, 199, 508, and 649 were in the N1, N2, Lid, Cap-627 linker and 627 domains, respectively. PB1 298, 642 and 667 were in the palm, thumb and priming loop domains, respectively (Figure 1A). In terms of known markers of mammalian adaptation, these viruses all contained the avian-type amino acids of PB2 E627, D701, G590, Q591 and T271. Molecular modelling revealed that PB2 T199, associated with AT, were predicted to bind to H634 and P627 of PB1. However, PB2 A199, found in those AIVs not associated with AT, was not predicted to display this interaction (Figure 1, supplementary figure 1). While the interaction between PB2 V67 and surrounding amino acids of PB1 was similar to I67, it is theoretically easier for V67 to combine with PB1. PB2 S152 was located at the RNA binding site, in a flexible loop (Figure 1B). Therefore, S152 may have less RNA recognition and mobility than A152, which could affect RNA binding.

Figure 1. Location of amino acids associated with airborne transmission in the influenza virus polymerase complex. (A) Ribbon diagram showing the interaction of PB2 and PB1 subunits with their respectively amino acid identities. (B) Detail of the impact of amino acids at PB2 67, 152 and 199 in the interaction with PB1. Hydrogen bond between H634 and T199 indicated by a dashed line. Images were based on PDB structure 4WSB.

We next studied the prevalence of the amino acid identities in Table 1 in AIVs overall, in subtype H1N1 AIVs only and in recent H5Nx Clade 2.3.4.4b AIVs in North America (Figure 2). As the viruses in this study were isolated in 2009, our analysis consisted of viruses isolated pre-2009, in 2009 and post-2009. AT-associated amino acids were either not present or rare in H5Nx Clade 2.3.4.4b viruses, with >98% of viruses containing amino acids not associated with AT. In AIVs and specifically subtype H1N1 AIVs, AT-associated amino acids were rare pre- and post-2009, with the highest frequencies in 2009. PB2 Q508 and I649 were the most common AT-associated amino acids, present in 5.3% and 4.9% of AIVs pre-2009 and in 3.7% and 4.6% of AIVs post-2009, respectively. Overall, amino acids not associated with AT predominated in these viruses.

Figure 2. Frequencies of amino acids in (A) PB2 and (B) PB1 associated with airborne transmission in avian influenza viruses in North America. Analysis of the frequencies of amino acids in AIVs and subtype H1N1 AIVs pre-2009, 2009 and post-2009 and in recent subtype H5Nx Clade 2.3.4.4b viruses were performed using the Global Initiative on Sharing All Influenza Data (GISAID) (https://gisaid.org) Data accessed 10 January 2024.

Amino acids in rgDE300 PB2 and PB1 increased viral replication and polymerase activity

We constructed a rg version of DE300 (rgDE300) and rgDE300 viruses in which amino acids in PB2 (rgDE300 PB2 NAT) or PB1 (rgDE300 PB1 NAT) or PB2 and PB1 (rgDE300 PB2 + 1 NAT) were changed (Table 1). Regarding the predicted mammalian transmissibility of these viruses, we have chosen a loss-of-function approach to study these amino acids. We also included the rg viruses A/Puerto Rico/8/1934 (H1N1) (rgPR8) as a positive control and A/shorebird/Delaware/274/2009 (H1N1) (rgDE274) as a comparator to rgDE300. rgDE274 is an AIV with a similar genome to DE300 but encoding NAT amino acids rather than AT amino acid and with HA and NA proteins identical to DE300 (Table 1).

In growth kinetics experiments, rgDE300 reached a significantly higher mean titre at 24 HPI compared to other viruses, the earliest time point at which titres of these viruses were detected (Figure 3A). Viral titres were similar between these viruses at later timepoints. To achieve greater resolution, we examined NP expression in infected cells within eight HPI, corresponding to the first round of replication. NP expression was greater at two and four HPI but not six or eight HPI, in cells inoculated with rgDE300 compared to rgDE300 PB2 NAT, rgDE300 PB1 NAT and rgDE300 PB2 + 1 NAT (Figure 3B and C). rgDE300 also produced plaques of significantly greater mean diameter compared to rgDE300 PB2 NAT, rgDE300 PB1 NAT and rgDE300 PB2 + 1 NAT (Figure 3D and E). The differences in replication led us to use the minigenome assay to study the impact of these amino acids on polymerase complex activity at 37°C. The activity of the rgDE274 polymerase complex was 22% that of rgDE300 (Figure 3F, supplementary table 2). We also found that substituting DE274 PA in DE300 polymerase complex had significantly less effect on polymerase activity compared to substituting DE274 PB2 or PB1 (supplementary figure 2). Mutation of amino acids in the rgDE300 PB2, PB1 or PB2 and PB1 each reduced polymerase activity, with PB2 amino acids having a greater impact in reducing polymerase activity compared to PB1 (yellow boxes, Figure 3F). Overall, these data indicate that PB2 and/or PB1 containing amino acids associated with AT had significant effects on influenza virus replication kinetics.

Figure 3. Amino acids in PB2 and PB1 affected replication kinetics, plaque sizes and polymerase complex activity. (A) Growth kinetics of viruses determined in MDCK cells inoculated at MOI = 0.001. (B) Expression of influenza virus nucleoprotein (NP) following inoculation of MDCK cells with respective viruses at MOI = 1. (C) Relative quantitation of influenza virus NP normalized to β-actin. Band intensities were measured using Image J from three independent experiments. (D) Diameters of plaques formed by respective viruses measured using Image J. Means ± SEM are shown. (E) representative plaque assay. (F) Activities of polymerase complexes determined by plasmid-based polymerase assays in 293 T cells. Genes from DE300 and DE274 are shown in orange and green, respectively, whilst DE300 genes containing NAT mutations are shown in yellow. The PR8 polymerase complex, in purple, was a positive control and reactions containing no influenza virus genes (no DNA) was a negative control. Polymerase activity of DE300 was normalized to 100%. PR8 and no DNA controls were excluded from statistical analysis. Means ± SEM are shown. * P < 0.05; ** P < 0.01; *** P < 0.001; **** P < 0.0001.

Amino acids in rgDE300 PB2 and PB1 increased viral genome replication and transcription

The impact of these amino acids on the polymerase complex as measured by minigenome assay led us to study the activity of these complexes in replicating and transcribing the viral genome by measuring the production of vRNA, cRNA and mRNA in infected cells. The rgDE300 polymerase complex resulted in the greatest amounts of all RNA species of all influenza virus gene segments at two and four HPI (Figure 4, supplementary tables 3–10). Amounts of all RNA species of PB2, PB1, HA, NA and M segments were also greater in cells inoculated with rgDE300 compared to cells inoculated with other viruses at six HPI whilst at eight HPI patterns were less clear (supplementary figure 3 and supplementary tables 3–10). Overall, these data indicate that PB2 and/or PB1 containing amino acids associated with AT had significant effects on the kinetics of influenza virus genome transcription and replication.

Figure 4. Amino acids in PB2 and PB1 affected the kinetics of influenza virus genome replication and transcription. Relative expression of vRNA, mRNA and cRNA of each influenza virus gene segment in MDCK cells inoculated with respective viruses at MOI = 1 at two and four hours post inoculation (HPI) (A to H and I to P, respectively). Expression of gene segments are expressed as fold changes relative to rgDE300 following normalization to GAPDH expression. *** P < 0.001. Means ± SEM are shown.

Amino acids in rgDE300 PB2 and PB1 affected infectivity and intracellular influenza virus staining

rgDE300 showed greater infectivity, with the proportion of cells staining positive for NP post inoculation with rgDE300 being greater compared with cells inoculated with viruses containing NAT amino acids or with rgDE274 at six HPI (Figure 5A). The pattern of intracellular staining was also different in cells inoculated with rgDE300 compared to other viruses. There was a greater presence of punctate staining in the cytosol at six HPI and, in cell inoculated with viruses containing NAT amino acids, staining was more restricted to the nucleus (Figure 5B and C). Therefore, these data indicate that PB2 and/or PB1 containing amino acids not associated with AT impact viral infectivity or kinetics of replication and intracellular trafficking of influenza virus proteins, possibly viral RNP complexes.

Figure 5. Amino acids in PB2 and PB1 affected infectivity and intracellular influenza virus nucleoprotein (NP) staining. (A) Mean percentages of MDCK cells staining positive for NP protein six hours post inoculation (HPI) at MOI = 1. (B) The mean ratio of percent cells staining for NP protein in the cytosol over total percent cells staining for NP protein at six HPI. ** P < 0.01. Means ± SEM are shown. (C) Representative images at six HPI. Green–influenza virus NP, blue–DAPI, red–ZO-1. Scale bars = 20μm.

Amino acids in PB2 and PB1 impacted pathogenicity in mice

All viruses, with the exception of rgPR8, did not cause sufficient morbidity to cause any animals to reach humane endpoints. rgDE300 caused a mean peak weight loss of 8.2 ± 1.4% at seven DPI whilst viruses with NAT amino acids did not cause significant weight loss compared to mock infected animals (Figure 6A and B). Titres of rgDE300 in the lungs were significantly greater compared to rgDE300 PB2 NAT and rgDE300 PB2 + 1 NAT but not rgDE300 PB1 NAT at three DPI whilst at six DPI viral titres were similar amongst all viruses, with the exception of rgDE274 (Figure 6C). In the lungs, rgDE300 was associated with more severe interstitial pneumonia compared to rgDE300 PB2 NAT, rgDE300 PB1 NAT and rgDE300 PB2 + 1 NAT, characterized by alveolar interstitial consolidation and extensive inflammatory cell infiltration (Figure 6D-H). Measurement of cytokine expression in the lungs at three DPI revealed that the expression of C-X-C motif chemokine ligand (CXCL)−10, CXCL-2, interleukin (IL)−6, IL-10 and interferon (IFN)-β were all significantly upregulated in mice inoculated with rgDE300 compared to rgDE300 PB2 NAT, rgDE300 PB1 NAT and rgDE300 PB2 + 1 NAT (Figure 7A-E). The expression of the host antiviral gene MX-1 was also increased in mice inoculated with rgDE300 compared to rgDE300 PB2 NAT, rgDE300 PB1 NAT and rgDE300 PB2 + 1 NAT (Figure 7F). Overall, these data further demonstrate that mutation of these amino acids in PB2 and PB1 in DE300, particularly those in PB2, impacted replication in vivo, leading to an attenuated inflammatory response and reduced pathogenicity in the mouse model.

Figure 6. Amino acids in PB2 and PB1 were associated with reduced disease severity and lung viral titres in mice. (A) Mean weight loss and (B) survival in mice inoculated with respective viruses. (C). Mean lung viral titres at three and six days post inoculation. Titres of rgDE274 were not detected. Means ± SEM are shown. rgPR8 was included as a positive control. ** P < 0.01; *** P < 0.001; **** P < 0.0001. Representative sections of lungs of PBS mock-inoculated (D) and mice inoculated with (E) rgDE300, (F) rgDE300 PB2 NAT, (G) rgDE300 PB1 NAT and (H) rgDE300 PB2 + 1 NAT. Sections were stained with hematoxylin and eosin.

Figure 7. Amino acids in PB2 and PB1 altered the host lung cytokine response following avian influenza virus inoculation. Expression of (A) CXCL-10, (B) CXCL-2, (C) IL-6, (D) IL-10, (E) IFN-β and (F) MX-1 mRNA in mouse lungs at three days post inoculation. Means ± SEM are shown. * P < 0.05; ** P < 0.01; *** P < 0.001; **** P < 0.0001.

Discussion

The influenza virus polymerase complex is a critical determinant of interspecies adaptation and transmission. Our previous studies identified a group of AIVs isolated from wild birds in Delaware Bay, USA that demonstrated different transmissibility in a ferret model of human influenza virus transmission [2, 6]. As the genomes of these viruses were similar, we could identify amino acids that may contribute to mammalian transmissibility. As the amino acids in the PB2 and PB1 genes had not been previously studied in this context, it was the aim here to determine their effect on the virus lifecycle and mammalian pathogenicity [6].

AIV polymerases do not function efficiently in mammalian cells but adaptive mutations can overcome species barriers. PB2 627 is arguably the most recognized mammalian adaptation, with a change from the AIV residue glutamine (E) to lysine (K) resulting in enhanced activity in mammalian cells and at the lower temperature of the mammalian upper respiratory tract (33°C) compared to the avian gastrointestinal tract (41°C) [11–14, 19]. The ANP32 proteins, including ANP32A, are important host factors here, mediated largely by the identity of the residue at PB2 627 [20, 21]. PB2 T271A and PB2 D701N can also enhance polymerase activity in mammalian cells. In viruses lacking PB2 K627, PB2 D701N disrupts a salt bridge, increasing the accessibility of a nuclear localization signal (NLS) to importin alpha and enhancing nuclear import [22, 23]. PB2 D701N can also change importin alpha preference from type 3, required by AIVs for efficient replication, to type 7, for which mammalian influenza viruses show specificity [23–25]. PB2 K627, N701 and/or A271 also confer greater transmissibility in the guinea pig model [14, 26, 27]. Interestingly, the AIVs studied here all contained the avian-type amino acids PB2 E627, D701, G590, Q591 and T271, suggesting that the amino acids studied here could potentially represent new mechanisms of AIV mammalian adaptation.

Our analysis of AIVs isolated from wild birds in North America, including the more recent H5Nx Clade 2.3.4.4b, indicated that the PB2 and PB1 amino acids studied here are not common, although several appear to be maintained. Considering that these amino acids facilitate replication in a mammalian background, it is possible that they remain uncommon as they do not confer a selective advantage in avian hosts. Further studies in an avian background may provide insight into this question.

PB2 67, 152, and 199 were located on the surface of the polymerase complex, indicating that interaction with host factors were likely involved. PB2 199 has been previously identified to differentiate avian and human influenza viruses, with A and S being strongly associated with AIVs and human influenza viruses, respectively [28, 29]. T199 in DE300 is biochemically more similar to S compared to A and thus may facilitate mammalian replication. PB2 contains NLS domains at 449–495 and 736–739 [30], which are the sites of importin-α interaction. As residue 508 was close to the first PB2 NLS, Q508R may alter importin-α interaction and affect PB2 accumulation in the nucleus, as observed with PB2 701 in the highly pathogenic subtype H7N7 AIV [23]. PB2 649 was located in the 627 domain, the site of the E627 K mutation necessary for transcription and replication, and could thus impact these functions [31].

Reassortments involving the introduction of AIV PB1 and contemporary human influenza viruses have resulted in the emergence of the causative viruses of the 1957 H2N2 and 1968 H3N2 influenza pandemics [32]. Mutations in PB1 also facilitate mammalian adaptation and replication, with the acquisition of PB1 216 by the 2009 pandemic H1N1 influenza virus leading to a reduction in polymerase fidelity that could increase the frequency of adaptive mutations [33]. PB1 L298I has also been identified, along with R386 K and I/A517 V, as a set of mutations in the pandemic 2009 H1N1 influenza virus that increased viral fitness [34].

Cytokine expression has been correlated to influenza virus polymerase activity, which was also evident here, with the rgDE300 polymerase complex demonstrating the highest activity and eliciting greater cytokine responses in the mouse lung (Figure 3F and Figure 7) [35–37]. This also correlated with increased weight loss, lung pathology and cellular infiltrates in the lungs of mice inoculated with rgDE300 compared to those inoculated with rgDE300 PB2 NAT, rgDE300 PB1 NAT and rgDE300 PB2 + 1 NAT (Figure 6).

These PB2 and PB1 amino acids also had a significant impact on early viral replication kinetics. These results were somewhat correlated by mouse lung viral titres, which were significantly greater in mice inoculated with rgDE300 compared to other AIVs (Figure 3A and Figure 6C). Further, rgDE300 was more pathogenic compared to rgDE300 PB2 NAT, rgDE300 PB1 NAT and rgDE300 PB2 + 1 NAT, which was similar to observations in the ferret upper respiratory tract [8]. The pattern of intracellular staining was affected such that NP staining was more restricted to the nucleus in cell cultures inoculated with rgDE300 PB2 NAT, rgDE300 PB1 NAT and rgDE300 PB2 + 1 NAT compared to rgDE300 (Figure 5). This could indicate that trafficking of viral ribonucleoprotein complexes, consisting of vRNA, NP, PB2, PB1 and PA, may have been affected, which may have contributed to the slower viral replication kinetics observed (Figure 3A).

This study had several limitations. We did not study the impact of individual amino acids. As we focused on mammalian replication and pathogenicity, we did not study these viruses in avian models to determine if any fitness costs were evident.

Overall, these data demonstrated that PB2 and/or PB1 containing amino acids associated with AT conferred greater replication kinetics on rgDE300, which was associated with increased pathogenicity and altered host responses. This leads us to hypothesize that they contribute to the zoonotic potential of these AIVs.

Acknowledgements

We thank Zhenting Huang and Zhanpeng Jiang for their assistance and contributions to this study. We thank the laboratory of Professor Yuxia Zhang at the State Key Laboratory of Respiratory Diseases/Guangzhou Medical University for their support and assistance with this project. We acknowledge the Theme-based Research Scheme (Project No. T11-712/19-N) of the Research Grants Council of the Hong Kong SAR Government, grants from InnoHK, an initiative of the Innovation and Technology Commission, the Government of the Hong Kong Special Administrative Region, financial support provided by the Pasteur Network programme from the Pasteur Network Foundation and Guangzhou Medical University.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

Pasteur Network programme from the Pasteur Network Foundation; Research Grants Council, University Grants Committee: [Grant Number T11-712/19-N].